Prerender · Monthly All-Hands

July All-Hands

Four nominations, seven investigations, and a customer who audited us. Let’s get into it.

July 2026
Presenter: Joe

Icebreaker

Something interactive to open with. Everyone plays from their own screen.

Presenter: Joe

Agenda

01

Values & Recognition

Living the values — four nominations this month

02

Team Updates

GTM, Product, Engineering, and Customer Success

03

People Update

Open roles and referrals

04

MCP Team Presentations

Seven teams, five minutes each

05

Celebrating Mistakes

Learning out loud as a team

06

Miscellaneous Topics

Sweepstake results, tool access, and the Q3 survey

Presenter: Joe

Our Values

The principles we hold each other to — a quick reminder before we recognise the people who lived them.

Ownership
Taking responsibility for outcomes, not just tasks
Autonomy
Empowered to act and decide within your domain
Shared Outcomes
Designing for the whole team to own results together

Living the Values is our program to recognise colleagues who’ve shown a specific example of living the Prerender values. Each all-hands we highlight nominations to help us remember what these values look like in practice.

Anyone can nominate a teammate — send a short summary to Lizzie. Four nominations this month.

Presenter: Charles

All Three Values in Action

🏆
Belmin
All Three Values

Ownership. Belmin took responsibility for web application security testing — not as a task handed to him, but as an outcome he owned end to end.

Autonomy. When testing surfaced issues, he leveraged Claude to fix them himself rather than just bouncing them back to engineering.

Shared Outcomes. He managed an overloaded deployment pipeline, prioritising the critical work when it mattered so the whole team kept moving.

Presenter: Anna

Ownership in Action

🏆
Aliya
Ownership

Ownership isn’t just about getting your own work done — it’s about staying with a problem when it gets messy, and telling people the truth about it before they have to ask. Aliya has done that every single month, on every project she leads.

March — pricing and free-trial migration. When she hit a billing mismatch for some Enterprise Plus customers, she didn’t just flag “something’s broken.” She traced it back to missing pricing flags and inconsistent manual migration steps from before her time, and came to us with a fix, not just a problem. She also quantified a revenue timing gap from mid-cycle upgrades and brought it to leadership before anyone asked her to. She went looking for the bad news so we could act on it early.

July — Moonshot POC data quality. Aliya found two issues that were quietly making the reports wrong: URL normalisation collapsing distinct pages, and crawl data including non-Googlebot traffic. Instead of letting the reports go out, she held them, had a constructive discussion with the team, and made sure they were fixed before release to customers.

Presenter: Janine

Autonomy & Shared Outcomes in Action

🏆
Kristina
Autonomy · Shared Outcomes

I’d like to nominate the whole CS team — you are all brilliant — but today I want to particularly thank Kristina for living Autonomy and Shared Outcomes.

She returned from maternity leave in May and hit the ground running: no ramp-up, no hand-holding. Since May our median first-reply time dropped from 2.5 hours to under an hour, and P90 came down from 1.8 days to 6.6 hours and has held there. These are team results — sustaining 24-hour coverage takes all hands. Kristina holds the night shift with her little one beside her, 18:00–02:00 CEST (01:00–09:00 PHT).

She knows exactly when to escalate, consistently filing well-structured Linear tickets — correctly identified, documented and routed to Engineering: a caching rules bug, a Google Shopping query param causing render spikes, a missing CSV export column. Across complex cases — missing cache inventory, rendering issues on high-traffic sites, overage alerts, billing edge cases — she handles technically demanding questions with patience and precision, and flags risk before it escalates.

She’s recognised beyond CS, too. When Sales needed a deep render optimisation analysis on a complex enterprise account, Kristina delivered autonomously: Googlebot patterns, URL waste, bot behaviour, concrete savings estimates. James’s response: “amazing, thank you so much for all this.”

Thank you, Kristina.

Presenter: Tiff

Ownership in Action

🏆
Carli
Ownership

Two things this quarter.

The Prism audit tool. Carli took it on and is getting it customer-ready essentially single-handedly. She’s project managing it herself — pulling the right people into the right conversations for clarification — and she designed the customer-facing report elements from scratch rather than waiting for someone to spec them for her. She’s also been deliberate about how she runs it: getting early feedback on a draft rather than building it by committee, which is why it’s actually moving.

The AI Insights launch. When we lost Product Marketing cover mid-launch, she stepped straight in. She found the launch wasn’t ready, and got every piece of comms written and shipped in under a week. She didn’t wait to be asked, and she didn’t let it slip.

It’s the clearest example of treating a deliverable as yours forever that I’ve seen this quarter.

Presenter: Tiff

FIFA — The Customer Who Audited Us

Three years as one of our biggest customers — operating the whole time without a contract.

Contract Value
$180k+
Over two years
Security Controls Evidenced
13
Plus a penetration test against our dashboard, run by FIFA’s own team
Departments Signed Through
4
Mid–World Cup
What It Took
Thirteen security controls, each with evidence. A penetration test against our dashboard, run by their team. Every finding fixed and confirmed by FIFA’s own tester.
What We Got
A Tier 1 enterprise security review we can now point to. Every enterprise conversation after this one is shorter.
Presenter: Tiff

AI Builders — Demand We Didn’t Have to Create

Matchr built their entire site with Lovable. It launched invisible in search. They had never heard of us. Claude recommended us, Claude did the integration, and they were live in under two hours — and we have the case study to prove it.

Two Active Tech Partner Conversations
Bubble and Base44 — both AI builders.
The Same Pattern, at Platform Scale
AI builders let anyone ship a site in a day, and every one of those sites is fully client-side rendered. Fast to build, invisible to crawlers.
Why It Matters
The AI build boom is producing our problem faster than anyone can build around it.
Presenter: Venessa

AI Insights — Launch Update

A new tool in the Prerender dashboard that catches AI crawler access issues at the source, before they affect your bottom line.

What Users Get
Visibility
See the exact pages AI crawlers can reach, what they can’t, and where access breaks down.
Diagnostics
Run diagnostics to identify where they’re blocked, and get step-by-step fixes.
Prioritisation
A prioritised page-level breakdown, so you know what to fix first.
Daily Health Status
So you’re never left guessing.
Exportable Reports
Diagnostics reports you can hand straight to your development team.
Biweekly Status Reports
So you always know where you stand.
Presenter: Venessa

AI Insights — Early Numbers

Where users went after the in-app announcement — and how deep they got.

Saw the announcement banner Clicked through the banner Visited the AI Insights tab Explored the Diagnostics tab Ran diagnostics on their site Exported a diagnostics report 234 31 80 41 28 3 13% of the 234 who saw it 34% of the 234 who saw it 51% of tab visitors 68% of Diagnostics visitors 11% of those who ran diagnostics 0 117 234 users

Steps are shown in journey order, but the percentages have different bases — note that tab visits (80) exceed banner clickthroughs (31), because users also reached the tab without going through the banner. So these steps are not a strictly nested funnel.

Presenter: Venessa

Nexus 2.0 — Fully Revamped 🚀

Cost
80% cheaper
New model: GPT-5.6 Luna
Speed
3–4× faster
Same task, less waiting
Answer Quality
78% → 94%
Measured on a rebuilt evaluation suite
Key Changes
Two Modes: Integrate & Verify
Helps new users set up step by step, or troubleshoot when something isn’t working.
Automated Stack Detection
No more long list of questions. Some domains still require probing questions.
Live Integration Check
Runs a check directly to confirm Prerender is working and spot common setup mistakes.
Rebuilt Evaluation Suite
Built from real customer conversations.

Why it matters. The old Nexus was rigid, and lacked the autonomy to guide users in the way it judged best. The new one is faster, cheaper, more accurate, and helpful for both technical and non-technical users — moving us toward our 40% integration-rate goal.

Presenter: Aliya

Google Search Performance — Limited Preview

A new report for 60 self-serve customers, generated once per customer, showing how pages served through Prerender appear in Google Search — visibility gaps, impressions, clicks, and indexing-related signals.

Customers emailed Engaged with the preview Started synchronisation Reports delivered Customer interviews completed 60 19 11 9 + 2 2 32% of those emailed 58% of those who engaged 9 delivered, 2 in progress — all 11 accounted for Feeding report improvements 0 30 60 customers

The POC is already doing what we needed it to do: validating customer interest, improving the report based on real feedback, and uncovering technical gaps we would not have found from internal testing alone.

Presenter: Charles

Engineering Update

The velocity improvements we reported last month continued unabated through July.

Velocity — Merged Pull Requests
2× Q1
A new benchmark of performance for the team — more than doubled from Q1
Production Uptime
10 weeks
Consecutive weeks at 100%
5xx Render Error Rate
−75%
From improvements to Prerender
Vulnerabilities Remediated
15,000+
Cleared from production ahead of the pen test
Security & Compliance
The SOC 2 audit penetration test is scheduled for early August. In preparation, we have remediated more than 15,000 vulnerabilities from production.
Challenge
Volatility in the price, performance and availability of frontier models is the new normal, and is expected to continue.
Presenter: Charles

Deploy Frequency

PRs merged per week, Feb 9 – Jul 20 2026. Q1 ran in the twenties and low thirties; since May we’ve held in the fifties to seventies — more than double the Q1 rate.

0 20 40 60 80 20 20 26 32 30 25 19 25 31 32 9 36 71 66 75 49 77 57 67 29 55 48 67 75 Feb 9 Feb 23 Mar 9 Mar 23 Apr 6 Apr 20 May 4 May 18 Jun 1 Jun 15 Jun 29 Jul 13

Total merges to master per week · git log origin/master --first-parent bucketed by ISO week (Mon-start) · Omega and human merges combined · Excludes the partial current week (W31, Jul 27+).

Presenter: Janine

Customer Success & Support

Support has always been in execution mode — Customer Success is now too. Here’s what we shipped in July.

Engagement Models — Live
Defining our book of business.
AI Response Agent
For Tech-Touch customers — currently in test review.
Revised Render Alert Emails
T1 Approaching T2 At Risk T3 Exceeded
PLG Lifecycle Stages
Defined and measurable. HubSpot setup in progress.
NPS Program — Launching 28 Aug
Product & GTM: join the webinar on 6 Aug 😊
Presenter: Joe

People Update

Joining Soon
Trung Tran — Brand and Content Strategist
Starts 1 September 2026.
Jens d’Hondt — Applied Research Scientist
Starts 1 September 2026.
Open Positions
Product Marketing Manager
Now open.
Referrals
Referrals Matter — a Lot
The best hires we’ve made came through people already here. If you know someone good, put their name forward.
Seven teams · Five minutes each

MCP Team Presentations

Seven teams, seven investigations — findings from the work each team did this month.

1. Upsell Radar
Who’s outgrown their plan and we serve well?
Tiff, James, Carli, Vitor
2. Support → Product Signal
What do tickets say to fix — does the roadmap match?
Anna, Fernanda, Fiona, Marcin
3. Renewal 360
A full account 360 into one renewal brief, in minutes.
Amanda, Giovanni, Hannah, Janine
4. Error-Page Tax
Which accounts burn renders on non-200 pages?
Aliya, Lizzie, Marcelo, Kristina
5. AI-Crawler Readiness
Who gets AI-bot traffic, and how well do we serve each bot?
Lucas, Marcos, Peter, Flor
6. Churn-Risk Watchlist
Which high-MRR accounts are silently degrading?
Taki, Rafael, Moss
7. Platform Pain — The Timeout Club
Worst-timeout accounts — our infra or their site?
Belmin, Charles, Jácint, Venessa
MCP Team 1 · Tiff, James, Carli & Vitor

Upsell Radar

Who’s outgrown their plan and we serve well — so we don’t upsell accounts we’re failing?

The Goals
Find the Over-Limit Accounts
Accounts at or over their plan render limit.
Gate by Service Health
Only upsell accounts we serve well.
Rank the Real Opportunity
Attach MRR to rank, and exclude anyone already in conversation.
MCP Team 1 · Tiff, James, Carli & Vitor

What We Said in Bilbao

The slide this team published at the offsite showcase — “The Overage Achievers.”

Lead With 3 High-Confidence Targets — James
Start with the Pro accounts that clear the health gate comfortably: Wonderschool (323% usage), HiDubai (320%), and menivim.net (262%). Hold the remaining 83 medium-confidence accounts until they’re 15+ days into their billing cycle.
Confidence — Carli
High for the 3 lead targets — backed by 19–24 days of billing data and clean health metrics. The other 83 are medium-confidence at 7–12 days elapsed; directionally right but projections may shift before month-end.
Wildcard & Top Account — Vitor
FAIRsharing (Starter, 1,924% usage) has never once been within plan in 24 months — the easiest upgrade on the board. The largest qualified account is Homedepot MX at $23,760 MRR, 217% usage, 93.6% cache hit.
Do Not Upsell — Tiff
Skip accounts we’re failing: Magic Betting (14.7s render time), OVHcloud (4% cache hit rate), Hello Storage (20% cache hit rate). Watch yearly Enterprise re-signs — verify term start date, since prior-term renders can inflate the overage signal.
MCP Team 1 · Tiff, James, Carli & Vitor

Three People, Three Different Lists

Same data. Same MCP servers. Same question.

Widest Net
197
accounts — loose health gate
Tightest
86
accounts
Third Run
In between
A third list, somewhere between the two

The health gate flips between 10/10 and 2/10 accounts passing depending on which metric is queried — the most likely root cause of the three-way divergence at the offsite.

MCP Team 1 · Tiff, James, Carli & Vitor

The Definition Was Wrong

Two criteria in the tightened definition were unworkable — and both failures were invisible until the numbers were re-run.

The accounts weren’t the problem. The definition was.

“15+ consecutive days over limit” can’t be measured on monthly plans
The usage counter resets every billing cycle. Nine of ten accounts reset between 4 and 18 July, so nobody had accumulated 15 days. Moores sits at 980% of its limit and shows only ~6 days over. The one account that passed did so because it’s annual — which the definition explicitly excludes. Replaced with persistence across two billing cycles.
“Render time < 3s” was measuring the wrong thing
Most requests are served from cache at ~0.05s. Averaged across everything, the cache hits drown out the slow renders and every account looks fast. Measured on cache misses only, 8 of 10 sit between 3.2s and 9.3s.
MCP Team 1 · Tiff, James, Carli & Vitor

The Health Gate Caught One of Our Own Leads

Wonderschool was one of the three “high-confidence” targets on the Bilbao list.

Wonderschool has a 36% cache hit rate.

Had we acted on the offsite list, we would have opened an upsell conversation with a customer we are failing.

MCP Team 1 · Tiff, James, Carli & Vitor

The Definition, Rewritten

Qualifying definition v3, revised 27 July — run across the full paid base rather than a sample.

Qualifying Criteria
  • Usage ≥ 100% of render limit
  • Over 100% of limit in each of the last 2 billing cycles
  • Cache hit ratio ≥ 70%
  • Cache-miss render time < 3 seconds
  • Timeout rate < 5%
  • Active subscription — not $0 MRR, not recently resigned, not annual
  • Not already at Preliminary stage or beyond in the HubSpot existing-business pipeline
The Funnel
Paid accounts ≥100% of limit, reliable data
102
Pass the service health gate
~30
Ready to act on, ranked by MRR then usage
10

Previous-cycle overage verified in ClickHouse. All 10 cross-checked against HubSpot for open deals and live conversations. Only 3 of the original Bilbao top 10 survived — HiDubai, blacksheep and MQ Sweden, the last of which the offsite list had missed entirely.

MCP Team 1 · Tiff, James, Carli & Vitor

10 Accounts Ready to Act On

# Account Plan MRR Usage Cache hit Cache-miss render Prev cycle over?
1HiDubai1Pro$349275%95.1%1.75sYes
2pickafund.comPro$349182%97.4%2.50sYes
3MQ Sweden AB (mq.se)2Pro$349170%98.0%0.53sYes
4Alternative Airlines3Pro$349170%97.7%2.64sLikely
5Wempe (wempe.com)4Pro$349146%86.6%2.43sYes
6polette / blacksheep.io5Pro$349140%87.1%2.62sYes
7overgear.comPro$349127%94.9%2.72sLikely
8prezman.frPro$349117%94.6%2.76sLikely
9silpo.ua (Fozzy)6Pro$349104%70.3%1.41sYes
10alternativeflooring.comGrowth$149182%88.6%1.25sLikely
Likely: ClickHouse raw counts hit 94–99% of limit in the previous cycle. Billing counts run ~2× ClickHouse rows for adaptive (mobile + desktop) accounts, so these are almost certainly over — billing-grade confirmation still open. 1Closed-won Jul 2025 — due a reconnect. Dormant 2nd Pro subscription. 2The Bilbao list missed this one entirely. 3James owns the relationship — route through him. 4No contacts in CRM. 5James already in live conversation (15–17 Jul) — confirm before outreach. 6Cache hit sits exactly on the 70% floor — no margin.
MCP Team 1 · Tiff, James, Carli & Vitor

The Health Gate Is Worth More Than the List

Accounts We’re Serving Badly — Nothing To Do With Upsell
  • thunderpick.io averages 174s per render
  • Intercom times out on 39.6% of requests
  • Duda, TorBox, Viva Aerobus and Lufa Farms are all heavy users on 6–9s renders

Given NRR is the H2 priority, that list may matter more than this one. It belongs with support, not sales.

HubSpot Cannot Do This Today

Cross-checking the ten exposed the CRM as much as the accounts:

  • Six have no owner assigned
  • Four have no proper company record — they’re named after an email address
  • Two don’t exist in HubSpot at all
  • One has a record but zero contacts

Render performance, cache hit ratio and latency don’t live in HubSpot in any form. This is the honest answer to “couldn’t we just pull a report?” — and a concrete specification for the cleanup already underway.

MCP Team 1 · Tiff, James, Carli & Vitor

Recommendation

The method holds, with one condition — left vague, it produces confident answers that are wrong.

It only works when the definition is precise about which metric it means.

Next Steps
  • Hand the 10 to Sales for outreach, with the two James accounts routed through him first
  • Route the service findings to support as a separate workstream — a retention signal, not an expansion one
  • Confirm previous-cycle overage for the four Likely accounts
  • Feed the CRM gaps into the HubSpot cleanup as a worked example
Confidence — Medium to High
Usage and service health data are strong and were verified in ClickHouse. Four of the ten need one more confirmation step. CRM context is the weakest layer — thin enough that a live conversation on one account only surfaced because we checked contact-level activity rather than deal stage.
MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Support to Product Signal

What do tickets say to fix — highest volume / frustration / MRR pain — does our roadmap match?

The Goals
Rank the Themes
Rank ticket themes by volume and by frustration (sentiment).
Weight by Revenue
Weight the themes by MRR at risk.
Test the Roadmap
Compare the top themes against the current roadmap.
MCP Team 2 · Anna, Fernanda, Fiona & Marcin

What We Said in Bilbao

The slide this team published at the offsite showcase — “Billing/Usage Is Our #1 Customer Pain by Every Measure — but the Roadmap Is Aimed Elsewhere”

The Mismatch
Billing/usage drives 1,486 tickets, 31.8% frustration, and $164.5k MRR at risk. Meanwhile the roadmap’s AI-crawler/machine-visibility focus appears in just 22 tickets.
Recommendation
Add a customer-facing billing-transparency feature (overage clarity, usage visibility) as a roadmap priority.
Supporting Signals
Integrations and diagnostics are the most covered.
Confidence — Medium
A lot of items didn’t match our taxonomy.
MCP Team 2 · Anna, Fernanda, Fiona & Marcin

From Roadmap Signal to Revenue Risk

Three analyses in one view — feedback to MRR to churn.

01 · Roadmap Signal

What 14 months of support feedback says the roadmap should address.

Source: 1,196 HubSpot tickets (June 2025 – July 2026) classified on the 3-level taxonomy and attributed to the roadmap. Roadmap-relevant = feature/friction/performance signal only. Only non-Done backlog is counted, so “unaddressed” means no roadmap item and no currently-open backlog item — some were shipped earlier (Done), and areas like data-viz, recache and cache saw substantial recent delivery.

02 · MRR Impact

The MRR sitting behind open tickets right now.

Source: Prerender IRIS hubspot.tickets joined to the account CSV, 365-day window. Resolved = “Closed”.

03 · Churn Analysis

The behavioral signal that predicts churn before a ticket is ever filed.

Source: ClickHouse audit_log plus analytics.customer_identity_map and Chargebee MRR, Jan 1 – Jul 22 2026 (Oct–Dec 2025 excluded for pricing/trial-migration noise). Statistics reported as observed associations, not causal claims. Mixpanel page/funnel and render-volume analyses specified but pending.

MCP Team 2 · Anna, Fernanda, Fiona & Marcin

01 · Roadmap Signal — Feedback Attribution

Every ticket classified on a 3-level taxonomy, then attributed to the product roadmap — 1,196 classified tickets, roadmap-relevant only.

Roadmap-Relevant, Unmet by the Roadmap
Unmet by the roadmap
106
of 261 roadmap-relevant asks
Tickets classified
1,196
On the roadmap
155
Unaddressed1
90
Attribution Funnel
Stage Tickets
Roadmap-relevant261
→ On the roadmap155
→ Unmet → in open backlog16
→ Unaddressed190
1No roadmap item and no currently-open backlog item. Only non-Done backlog is counted.

The roadmap is a small, deliberate slice. Only 22% of inbound (261) is roadmap-relevant; the other 78% (935) is operational/support, not roadmap material.

The real gap is ~90 asks. Of the 106 roadmap-relevant asks with no roadmap item, only 16 map to a currently-open backlog item — leaving ~90 with no roadmap item and no open backlog item.

MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Is It Roadmap Material — and Is It on the Roadmap?

The same 1,196 tickets, split twice.

Is the Feedback Roadmap Material? — Relevant vs Operational
261
935

Roadmap-relevant 261 (22%) · Operational / support 935 (78%)

Of Roadmap-Relevant, Is It on the Roadmap? — On Roadmap vs Unmet
155
106

On the roadmap 155 · Unmet 106

MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Where Demand Lands, and Where It Doesn’t

The 155 attributed tickets by roadmap item, against the 90 with no roadmap item and no open backlog item.

Roadmap Items Receiving Demand — 155 Attributed Tickets
Roadmap itemTickets
Improving Integration131
Onboarding revamp8
Granular Bot Classification6
Multi-WS / Access rights3
GSC + Prerender integration2
Activation experiment2
AI Crawler Telemetry1
Referrals1
Change Propagation1
Unaddressed Demand by Area — 90 Tickets
AreaTickets
rendering32
billing_and_usage15
url_filtering13
recache_rules9
api7
cache_management4
sitemap_integration3
account_management2
data_visualization2
mobile_rendering1
ui_ux1

Captured roadmap demand is almost all one item — Improving Integration (131 of 155). The machine-visibility bets draw ~0 current inbound: forward bets, not responses to today’s demand.

MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Key Learnings

Five conclusions from attributing 14 months of support feedback to the roadmap.

1
The roadmap is a small, deliberate slice of inbound. Only 22% (261) of tickets are roadmap-relevant; 78% (935) is operational/support — bugs, billing, account ops and noise.
2
Of roadmap-relevant feedback, 41% (106) is unmet by the roadmap. Only 16 map to a currently-open backlog item, leaving ~90 genuinely unaddressed.
3
Unaddressed demand clusters in rendering, billing/usage control, URL filtering, recache rules and API — reliability & control refinements the machine-visibility-heavy roadmap does not cover.
4
Captured roadmap demand is almost all one item — Improving Integration (131 of 155). The machine-visibility bets draw ~0 current inbound: forward bets, not responses to today’s demand.
5
Read “unaddressed” with the Done caveat: only non-Done backlog is counted, so areas substantially delivered in the last ~12 months (data visualization, recache rules, cache management) still surface here. The durable forward gap is rendering, billing/usage control and URL filtering.
MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Feedback Structure: Product Area → Topic

Level 1 to Level 2, the nine largest areas — volume, the feedback-kind mix, and the specific thing customers wrote in about.

Product area Tkts Feedback kind Level 2 topics
billing_and_usage353billing_dispute 260, account_request 37, bug_report 36, feature_request 13, onboarding_question 5, integration_help 1, noise 1plan_upgrade_confusion 110 · overage_alert 73 · incorrect_charge_dispute 43 · payment_failure 28 · cost_cap 25 · bot_traffic_cost_surprise 23 · refund_request 18 · invoice_management 8 · usage_breakdown_by_domain 8 · payment_method_change 6 · account_reactivation 5 · billing_days_remaining 3 · discount_request 2 · unattributed 1
rendering243bug_report 206, performance_complaint 29, integration_help 5, billing_dispute 2, positive_signal 1render_reliability 152 · render_timeout 36 · blank_page 31 · soft_404_prevention 11 · html_stripping 8 · custom_request_headers 2 · render_request_behavior 2 · blocked_scripts_config 1
account_management210account_request 187, bug_report 18, onboarding_question 2, feature_request 1, noise 2plan_change 98 · cancellation 66 · notifications 15 · user_settings 11 · account_reactivation 9 · domain_addition 6 · duplicate_account 2 · referral_program 1 · legal_compliance 1 · unattributed 1
integration_setup150integration_help 120, bug_report 28, feature_request 1, positive_signal 1cloudflare_config 63 · cdn_config 32 · framework_specific 25 · nginx_apache_config 11 · integration_not_detected 10 · dns_proxy_mode 5 · token_auth_error 4
cache_management40bug_report 30, integration_help 4, performance_complaint 4, feature_request 2recache_queue_management 15 · bulk_delete 13 · cache_manager_reliability 4 · bulk_recache 3 · exact_url_search 2 · export_cache_data 2 · cache_list_pagination 1
recache_rules35bug_report 23, feature_request 5, performance_complaint 4, onboarding_question 1, integration_help 1, account_request 1per_url_ttl 10 · recache_ignore_rules 8 · content_change_detection 8 · per_directory_ttl 4 · sitemap_based_schedule 2 · recache_jitter 2 · render_speed_control 1
url_filtering29integration_help 13, bug_report 13, feature_request 3query_param_exclusion 8 · partial_url_match 8 · regex_pattern_support 7 · allowlist_mode 3 · ignored_url_status_code 3
access_and_permissions29account_request 15, bug_report 11, integration_help 2, onboarding_question 1team_user_management 16 · role_based_access 6 · login_access 3 · sso 2 · token_security 1 · two_factor_auth 1
onboarding_and_education25onboarding_question 13, bug_report 8, feature_request 2, integration_help 2integration_setup_confusion 6 · integration_checker_issues 6 · product_concept_confusion 6 · docs_and_guides 3 · waiting_state_ux 2 · metric_meaning 1 · unattributed 1
MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Feedback Structure — the Long Tail

The remaining nine areas, same breakdown. Together with the previous slide this is all 1,196 classified tickets.

Product area Tkts Feedback kind Level 2 topics
api16bug_report 7, performance_complaint 5, feature_request 3, integration_help 1recache_api_filters 8 · insights_export_api 2 · cache_status_api 2 · api_token_scoping 2 · search_api_total_count 1 · cache_clear_api 1
seo_analytics15bug_report 13, integration_help 2indexing_impact_reporting 12 · seo_scoring 2 · google_search_console_integration 1
data_visualization11bug_report 9, feature_request 2charts 5 · sort_and_filter 4 · insights_export 1 · domain_list_view 1
sitemap_integration10bug_report 6, integration_help 2, onboarding_question 1, performance_complaint 1sitemap_management_ui 6 · automatic_sitemap_polling 4
unattributed10noise 7, bug_report 1, integration_help 1, unattributed 1
machine_visibility8integration_help 4, bug_report 3, feature_request 1bot_classification 6 · telemetry_crawler_activity 1 · eligibility_status 1
crawler_visibility7bug_report 6, feature_request 1crawler_activity_report 4 · url_discovery_source 2 · per_bot_breakdown 1
mobile_rendering3integration_help 2, onboarding_question 1enable_mobile_rendering 2 · mobile_desktop_usage_visibility 1
ui_ux2performance_complaint 1, bug_report 1inconsistency 1 · unattributed 1
MCP Team 2 · Anna, Fernanda, Fiona & Marcin

02 · MRR Impact — Open Support Tickets

Of 2,578 tickets across active accounts, 2,520 are resolved — only 58 remain open, and 40 of those belong to 25 active accounts carrying $48,565 MRR.

Open tickets
40
Active accounts
25
MRR with open tickets
$48,565
Waiting on us
23
Open Tickets by Status
23
8
4
4
1
Status Detail — MRR at Stake
StatusTicketsMRR at stake
Waiting on us23$40,616
New8$6,242
Open4$34,113
Waiting on customer4$2,678
Waiting for something1$349

The ball is in your court. 23 of 40 open tickets are “Waiting on us” — carrying ~$40.6k MRR. These are the fastest wins: a response closes them.

MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Open Tickets by Area

Tickets and the MRR of the active accounts behind them — which open tickets, in which areas, hold how much revenue.

AreaTicketsMRR of active accounts
billing_and_usage10$7.4k
unclassified7$7.2k
account_management6$33.7k
rendering5$1.0k
cache_management3$29.9k
integration_setup2$398
access_and_permissions1$1.8k
seo_analytics1$412
url_filtering1$1.8k
recache_rules1$149
crawler_visibility1$149
Do not sum this column. An account’s full MRR is attributed to every area it has an open ticket in, so the per-area figures overlap by design. The unduplicated total is the $48,565 carried by 25 accounts.
MCP Team 2 · Anna, Fernanda, Fiona & Marcin

03 · Churn Analysis — Behavioral Signals

Do customers behave a certain way before they ticket — and do silent customers who never ticket behave the same way before they churn? They do.

Churn · silent, no-ticket
59%
the load-bearing signal
Churn · silent + ticketed
86%
Churn · active users
22%
Paying-base churn
43.8%
What the Analysis Covers
StageCount
Negative tickets in window558
With company_id438
Resolved to product identity215
— already churned78
No-ticket comparison sample437

One behavior dominates. In-product login collapse — not billing-page thrashing — is the earliest, most consistent churn precursor, and it is statistically significant (RR 1.89, OR 3.15, χ²=33.3, p<0.001) even among customers who never contact support.

Silent risk is invisible risk. Customers who submit negative tickets churn at 36% — below the 43.8% base. The highest-risk group is the silent majority: 43% of paying customers logged in zero times, and 59% of them churned — with no support signal at all.

MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Canonical Behavioral Pattern Library — P1 to P3

Generalized from audit_log event sequences: logins, account and billing-state changes, manual cache clears, cancellations and reactivations.

The Silent Fade — P1 · Strongest

All taxonomies · engagement collapse

steady logins → login rate down → only system/billing updates → non-renew / cancel

Human logins decay toward zero while the account still shows automated billing-cycle updates. Churned accounts averaged a fraction of retained logins in every cohort (rendering 6.5 vs 242; billing 32 vs 90; account-mgmt 13 vs 85).

Prevalence universal · Lead time weeks–months · Discriminative very high

Billing Shock → Cancel — P2

billing_and_usage

charge / overage event → billing-state updates → “stop billing” ticket → cancel

A Chargebee charge/renewal change precedes an urgent billing ticket then cancellation. Aligns with Prerender’s #1 billing-surprise driver: overage charges. 73% of churned billing-cohort accounts fired a cancel event.

Cohort churn 36% · Lead time days · Discriminative medium

Troubleshoot & Abandon — P3

rendering · cache_management

manual cache-clear burst → recache attempts → rendering ticket → usage down → cancel / resolved · retained

Repeated manual cache-clears signal a customer fighting a rendering problem. Caveat: cache-clearing alone does not predict churn (~50% of both churned and retained clear cache) — it becomes predictive only when followed by login decline (leads to P1).

Cohort churn 29% · Lead time days–weeks · Discriminative low alone

MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Canonical Behavioral Pattern Library — P4, P5 and the Gap

The onboarding failure mode, the save window, and the one journey the audit log cannot see.

Onboarding Stall — P4

integration_setup

signup → integration ticket → low sustained logins → never activates → cancel

Early-tenure accounts that ticket about setup and never reach steady rendering. Highest cohort churn (52%), lower MRR ($49–$149). A distinct onboarding motion, not renewal-stage churn.

Cohort churn 52% · Lead time weeks · Discriminative medium

Cancel → Reactivate — P5 · Save Window

account_management · billing

cancel event → grace / outreach → reactivate

18% of churn-flagged accounts later reactivated. Retained accounts also show cancel events (account-mgmt 45%), i.e. successful saves. The interval between SITE_USER_CANCELLED and period-end is a concrete, actionable save window.

Reactivation 18% · Window cancel → period-end · Action win-back

Not Yet Observable — Gap

Requires Mixpanel / render volume

The billing-page → pricing-page → checkout-abandon journeys are page-view sequences that audit_log does not record. Confirming event ordering around tickets and quantifying render-volume decline needs Mixpanel funnels plus the ClickHouse analytics table. Login-frequency is used here as the validated engagement proxy.

MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Ticket vs No-Ticket — Is Silent Churn Real?

Bucketing both cohorts by in-window login count shows the same behavioral signature predicts churn in both.

Silence is the tell.

Zero-login customers churn at 59% vs 22% for active ones — a 2.7× spread, with no ticket ever filed. The behavioral signature works with or without a ticket, which is precisely what makes it usable for proactive, silent-churn detection.

MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Churn by Engagement — Both Cohorts

Cohort B is a random sample of 437 paying customers (live Jan 1 2026, plan > $0) who submitted no negative ticket.

Cohort B · No-Ticket Paying (n=437)
EngagementnChurn
Silent · 0 logins18759%
Low · 1–9 logins11941%
Active · 10+ logins13122%
Base churn for this sample: 43%.
Cohort A · Negative-Ticket Customers (n=215)
EngagementnChurn
Silent · 0 logins2186%
Low · 1–9 logins3468%
Active · 10+ logins16023%
Base churn for this cohort: 36%.

Ticket + silence is close to terminal. A negative ticket from a customer who has stopped logging in churns at 86%. The same silence signal, amplified.

Cohort A vs B headline. Ticketed customers churn at 36%; the no-ticket paying base at 43.8%. Support contact is mildly protective. The behavioral signature (login collapse) is the common thread — it works with or without a ticket, which is precisely what makes it usable for proactive, silent-churn detection.

MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Churn Validation & Signal Strength

Appropriate statistics, not invented metrics.

Relative risk
1.89×
silent vs engaged (no-ticket)
Odds ratio
3.15
of churn if silent
Chi-squared
χ²=33.3
p < 0.001 · highly significant

Treating “zero logins in the window” as a churn classifier on the no-ticket cohort: precision 59%, recall 59%, F1 0.59. As a first-pass, single-feature CS trigger that requires no support ticket, that is strong. Combining it with tenure, plan tier, and render-volume decline should push precision higher.

Behavioral signature Population Churn rate vs base Leading-indicator strength
Silent + negative ticket2186%+2.4×strongest — near-deterministic
Silent · no ticket18759%+1.4×strong, invisible to support
Low logins · ticketed3468%+1.9×strong
Low logins · no ticket11941%~baseweak alone
Active · ticketed16023%0.6×protective
Active · no ticket13122%0.5×protective

MRR at stake in the blind spot. Within this 437-customer sample alone, $9,214/mo of still-active MRR sits in silent (zero-login) no-ticket accounts — exactly the accounts today’s ticket-driven CS motion never sees. Extrapolated across the ~3,600 paying base, the silent-but-active pool is materially larger.

MCP Team 2 · Anna, Fernanda, Fiona & Marcin

Recommendation

Three analyses, one answer — and it is not the one the roadmap is currently built around.

Add a customer-facing billing-transparency feature — overage clarity, usage visibility — as a roadmap priority.

What the Volume Says
billing_and_usage is the largest area at 353 tickets, of which 260 are billing disputes — led by plan_upgrade_confusion (110) and overage_alert (73). It is also the second-largest block of unaddressed roadmap-relevant demand (15 of 90), and P2 names overage charges as Prerender’s #1 billing-surprise driver.
What to Do Next
  • Work the 23 “Waiting on us” tickets — ~$40.6k MRR, closable with a response
  • Ship a zero-login watchlist as a first-pass CS trigger; layer tenure, plan tier and render-volume decline to push precision higher
  • Use the P5 save window — the interval between cancel and period-end
  • Close the observability gap with Mixpanel funnels plus the ClickHouse analytics table — specified but pending
Confidence — Medium
Bilbao’s caveat stands: a lot of items didn’t match our taxonomy. “Unaddressed” counts only non-Done backlog, so recently delivered areas still surface. Statistics are observed associations, not causal claims. The billing-page to checkout-abandon journey is not observable in audit_log at all.
MCP Team 3 · Amanda, Giovanni, Hannah & Janine

Renewal 360

Assemble a full account 360 into one renewal brief in minutes — and templatize it to scale beyond enterprise.

The Goals
Pick a Real Account
An imminent renewal with real ticket history — not a clean test case.
One Brief, Four Layers
Money, usage, health and support assembled into a single document.
Land a Narrative
A clear renewal position — right-size or expansion, stated outright.

In Bilbao we started generic — everyone hands-on across the whole brief. It didn’t work. We split it instead: Giovanni on usage, Janine on support interaction, Amanda on pricing and picking the right customer for the setup.

MCP Team 3 · Amanda, Giovanni, Hannah & Janine

What We Said in Bilbao

The slide this team published at the offsite showcase — “Faire’s Usage Ran 47% Over Allowance — the July 3 Invoice Hit ~$54,900, Nearly 3× Their $19,200 Contract”

Recommendation: Right-Size the Renewal
Don’t renew at $19,200; propose an upgrade covering ~110M+ renders, framed around the customer’s growth. Target $66,000–$67,000/yr, raising price per 1,000 renders toward $0.60+.
Confidence: Medium
Usage was reconciled via a canonical billable-renders query, but the renewal price target is unsettled and the spike’s recurrence is unconfirmed.
Account Health Is Uncertain
Inconsistent data, 47% over allowance, price per 1,000 renders is below our target.
Open Questions Before Finalizing
Was the Jul–Oct 2025 spike (85.7M renders) one-time or recurring? Also investigate the Dec–Mar usage drop (~60K/mo) — avoidance, seasonality, or removed domains — and reconcile early data conflicts (Redash vs IRIS).
MCP Team 3 · Amanda, Giovanni, Hannah & Janine

Since Bilbao: A Second Account, End to End

Babylist’s renewal is due Sep 25 and the account is already in T3 overage.

58.9M of 60M renders used — ~98% — projecting ~72.6M by renewal.

At a daily burn of ~199,000 renders. Cache hit rate is 46.8%, well below the 70% healthy threshold, and 2,105,650 4xx errors in the last 30 days confirm removed campaign or product pages still being crawled — burning render quota with no SEO benefit.

MCP Team 3 · Amanda, Giovanni, Hannah & Janine

The Usage Picture

Eight numbers, each with the read attached — the layer Giovanni owned.

Metric Value Read
Quota60M renders/yr
Used (period-to-date)58.9M (98.2%)Overage locked in
Projected at renewal~71.7M (119%)T3
Daily burn~196K renders/day
Cache hit rate47%Below 70% healthy
Live 4xx rate~90%12 months running — ~2M wasted renders/mo
Render mix~82% recache / ~18% on-demandRecache is the main consumer
12-mo trendDoubled YoYPeaked ~9.7M/mo (Apr), recache cut sharply since June — growth plus config-driven

Overage is driven by chronic 4xx errors and aggressive recache TTLs, not clean traffic growth. Fixing both likely brings usage back inside 60M.

MCP Team 3 · Amanda, Giovanni, Hannah & Janine

The Money: One Rate, Four Reference Points

Per 1,000 renders — the current rate sits above the minimum but below the commercial target.

List price
$1.00
per 1,000 renders
Current rate
$0.495
legacy, below target
Commercial target
$0.60
target floor for this account
Minimum acceptable
$0.38
below this the account underperforms commercially
Discount Should Follow Deal Size
The discount from list price should be determined by deal size — larger deals justify a deeper discount, but $0.60 is the target floor here.
Overage Rate Is Out of Date Too
This account’s overage rate is $0.50/1,000; the current list overage rate is $1.25/1,000. If the cache and 4xx error issues are fixed, Babylist may not need more renders than their current 60M quota — making a rate increase the right lever at renewal, not a larger plan.
MCP Team 3 · Amanda, Giovanni, Hannah & Janine

Health Is Mixed and Billing Trust Is Fragile

The two layers you only get by assembling the whole 360 — and the two that change how we open the conversation.

Account Health — Mixed

Genuinely over quota, but inflated by a ~90% 4xx rate and weak 47% cache efficiency. Effective rate ($0.495/1k) sits below the $0.60 target.

Real optimization headroom before any plan change — the account is a rate-increase candidate, not a size-increase one.

Support — Quiet and Clean on the Surface
  • 6 tickets in 12 months, all closed, 83% neutral
  • But two billing disputes plus a previously lapsed/cancelled renewal
  • One frustrated ticket (Nov 2025, 401s plus slow renders), resolved

Billing trust is fragile — an unannounced overage would land badly. Worth confirming performance has held.

MCP Team 3 · Amanda, Giovanni, Hannah & Janine

Recommendation: Right-Size the Renewal

Renew at 60M renders. Do not increase the plan size.

Move the rate, not the quota.

The Ask at Renewal
Move the rate from the legacy $0.495/1,000 toward the $0.60 commercial target, bringing annual contract value from $29,700 to $36,000/yr — an increase of $6,300 (+21%). The new list overage rate of $1.25/1,000 should be included in the renewal proposal.
Why Not a Bigger Plan
The current render overuse is driven by 4xx errors — a ~90% live 4xx rate, running 12 months — not genuine traffic growth. Fixing the error source before renewal could bring usage back comfortably within quota — and strengthen the case for the rate increase.
MCP Team 3 · Amanda, Giovanni, Hannah & Janine

What We Still Need to Answer

Three open questions before the proposal goes out — and how much of this we’d stand behind today.

Open Questions Before Finalizing
  • Is the June/July recache drop permanent? It swings annual demand between ~50M and ~95M and determines the right quota.
  • What’s the source of the 4xx errors — which crawler, which dead URLs — and who owns the fix before renewal?
  • Why does the current rate ($0.495) sit below the $0.99/1k quoted at the last renewal — legacy pricing we’re honoring?
Confidence — High
Usage, billing and support figures are pulled directly from source systems — ClickHouse is authoritative on render counts — so the account picture is solid. The one caveat is forward demand: the recent recache reduction means the renewal-period projection is a linear estimate with no prior-year annual baseline. That doesn’t affect the core narrative.
MCP Team 4 · Aliya, Lizzie, Marcelo & Kristina

Error-Page Tax

Which accounts and error types burn renders on non-200 pages — what is the waste and who do we contact?

The Goals
Rank the Accounts
Top accounts by billable non-200 renders, with the 3xx / 4xx / 5xx split.
Quantify the Waste
Dollars, plus the cache angle.
Split by Error Type
So the outreach is right — a 404 and a 503 are not the same conversation.
MCP Team 4 · Aliya, Lizzie, Marcelo & Kristina

What We Said in Bilbao

The slide this team published at the offsite showcase — “~101.6M Renders Went to Non-200 Pages, Burning ~$3,095 for Us in 1 Month”

Where the Waste Is
~101.6M renders went to non-200 pages. 4xx client errors lead at 38.3M ($1,166) and are the most actionable to clean up.
Non-200 Cost Breakdown for Customer
4xx $1,166, 3xx $978, 6xx $829, 5xx $120. Customer-side render value is estimated at $120k–$145k (range $105k–$170k).
Recommendation
Work with CS to set thresholds and severity, and quantify the impact of the error-page tax on accounts.
Confidence
Medium to High. Internal render counts and costs are solid; customer-side value is an estimate pending billing data.
MCP Team 4 · Aliya, Lizzie, Marcelo & Kristina

We Were Counting the Wrong Thing

Last 30 days. Internal render-node cost is better approximated by render duration, not raw render count.

7.86% of renders. 43.13% of render time.

Roughly $20k/month of internal render-node cost. A count-based model put non-200 cost at about $3.6k/month — the duration-based estimate is ~$19.8k/month, about 5.5× higher.

MCP Team 4 · Aliya, Lizzie, Marcelo & Kristina

The Method

How the internal non-200 cost figure is built, and the inputs behind it.

Internal non-200 cost = non-200 render time ÷ total render time × projected monthly render-node cost — that is 43.13% × $46,000 ≈ $19,840/month.

Inputs, last 30 days
Metric Value
Total renders2,340,244,062
Non-200 renders184,037,533
Total render time740,213 hrs
Non-200 render time319,278 hrs
Projected render-node cost$46,000/mo
MCP Team 4 · Aliya, Lizzie, Marcelo & Kristina

Breakdown by Status

Last 30 days. Share of count, share of render time, and the cost that follows the time.

Status Share of all renders Share of total render time Est. monthly render-node cost
3xx3.76%7.87%~$3.6k
4xx3.87%24.38%~$11.2k
5xx0.23%10.88%~$5.0k
Non-200 total7.86%43.13%~$19.8k
6xx is not included in this breakdown until Engineering confirms how it maps in the current data source.
MCP Team 4 · Aliya, Lizzie, Marcelo & Kristina

Where the Cost Actually Sits

Every bar is a percentage of the same monthly total — count on top, render time below.

Share of all renders Share of total render time 0% 5% 10% 15% 20% 25% 30% 3xx 4xx 5xx 3.76% 7.87% 3.87% 24.38% 0.23% 10.88%

The biggest driver is 4xx: only 3.87% of render count, but 24.38% of total render time, or about $11.2k/month. 5xx is much smaller by count at 0.23%, but expensive by duration because average render time is about 53.7 seconds, representing about $5.0k/month.

MCP Team 4 · Aliya, Lizzie, Marcelo & Kristina

Customer Impact

We know this affects customers. We cannot yet say which ones, cleanly.

What We Know
We already see technical support escalations, and some churn cases are related to customers not understanding or trusting the value they get when renders are spent on errors.
What We Do Not Have Yet

A clean account-level view showing:

  • which customers are most affected
  • which status types are driving the issue per account
  • whether the issue is recurring or temporary
  • whether the affected renders are inside plan allowance or creating Extra Render exposure
  • whether there are related CS tickets, escalations, or churn signals

So the next step is not to estimate customer-side dollars broadly. The next step is to connect this render data to accounts and customer context.

MCP Team 4 · Aliya, Lizzie, Marcelo & Kristina

Non-200 & Timeout Render Exposure

Accounts whose crawler traffic is generating error-page or timeout renders — every one of which is billable. Ranked by render volume weighted for billing exposure, reliability escalations and churn risk.

Accounts on Watchlist
12
of 74 with render-failure tickets in 30d
Non-200 Renders (30d)
9.27M
Error pages + redirects, all billable
In Extra-Render Exposure
3
2 over limit · 1 at 97%
Urgent / Churn Escalations
4
Including 1 explicit churn threat
# Account Plan / MRR Non-200 renders (30d) Avg render Allowance / exposure CS signal
1Popken Fashion GroupEnterprise+ · $13,3336,977,4846.6sInside · 3.9%Urgent
2RELAYTOPro · $349976,2861.3sOver limit · 198%Escalation
3BingoPlusPro · $34974,3522.9sInside · 17%Churn threat
4Bill McIntoshStarter · $4947,5642.4sOver limit · 277%Overage T3
5FanaticalPro · $349341,1313.4sInside · 69%Escalation
6ExponentGrowth · $14996,2281.2sAt limit · 97%Escalation
7Fortnum & MasonPro · $349264,1181.7sInside · 53%Escalation
8Sling / DishPro · $349306,7302.4sInside · 62%Ticket
9GreentubePro · $34935,1553.6sInside · 10%Escalation
10WAM / SoshPro · $34971,8292.9sInside · 16%Escalation
11Sportbet / fullslotGrowth · $14919,7235.5sInside · 28%Ticket
12Riders SharePro · $34936,2633.1sInside · 16%Auto-alert
Window: last 30 days for render volume, 90-day trend. Snapshot 25 July 2026. Source: Iris — ClickHouse analytics, HubSpot tickets, Chargebee. 34 issue tickets across these accounts, all-time. How exposure is read: the pre-computed overage table had not aggregated the current billing period for 11 of the 12 accounts, so exposure compares observed 30-day billable renders against the plan’s monthly render limit. Bill McIntosh was confirmed independently as Tier 3. “Inside” = under limit; “At limit” = 90–100%; “Over” = above limit.
MCP Team 4 · Aliya, Lizzie, Marcelo & Kristina

Who We Contact, and What We Say

Four priority actions off the watchlist — the accounts where doing nothing has a cost.

P1 · BingoPlus — CS Save

Explicit churn language plus chronic ~20K/wk failures and latency complaints since March. Owner outreach and a reliability plan this week.

“…we will stop using” — 10 tickets, 8 in the last 180 days.

P2 · Popken — Reassure & Fix Source

~7M error renders looks alarming but sits inside a 240M custom limit (3.9%). Confirm no overage, then fix the 404 / redirect source.

“Pause rendering to avoid overages”

P3 · RELAYTO + Bill McIntosh — Overage

Genuine Extra-Render charges driven by errors — 2× and 2.8× plan. URL-filter / soft-404 config plus a credit conversation.

“Excessive x10 billing” · “URGENT: spending out of control”

P4 · Exponent — Pre-empt

At 97% of the 100K Growth plan, with errors making up 99% of billable usage. One bad week tips into overage — tune filtering now.

“prerender / exponent costs”

MCP Team 4 · Aliya, Lizzie, Marcelo & Kristina

What This Number Is, and What It Is Not

An 80/20 internal cost estimate, not exact accounting.

Caveats
  • Based on the last 30 days, not a long-term average yet
  • Uses projected $46k/month net render-node cost, pending Finance/Engineering confirmation
  • Uses render duration as a proxy for capacity consumed
  • 6xx excluded until Engineering confirms how it maps in the current data source
What It Sizes
Not every non-200 is automatically waste or customer-fixable. This analysis sizes the internal render-node capacity consumed by non-200 responses — nothing more, and that is the point of it.
MCP Team 4 · Aliya, Lizzie, Marcelo & Kristina

Recommendation

Use this as the internal cost baseline, then add the account-level breakdown as the next step.

We already have the signals. We just don’t use them proactively yet.

Next Steps
  • Confirm the $46k/month render-node cost baseline with Finance/Engineering
  • Confirm how 6xx maps in the current data source
  • Build the account-level breakdown by customer, status type, render count and render time
  • Connect the highest-impact accounts to CS context: open tickets, escalations, renewal risk, churn signals
  • Use the diagnostics work to turn this from reactive investigation into proactive detection
Confidence — Medium to High
Render volume and render-time data are strong enough for an 80/20 internal cost estimate. The $46k/month render-node cost baseline still needs Finance/Engineering confirmation. Customer-side and account-level impact is directionally real, but not quantified cleanly yet.
MCP Team 5 · Lucas, Marcos, Peter & Flor

AI-Crawler Readiness

Who gets AI-bot traffic and how well do we serve each bot? The Machine Visibility story.

The Goals
Profile Per-Bot Traffic
For a top account — renders, cache, render-time, error%.
Compare AI Bots vs Google
Find who we serve worst.
Build the Narrative
Assemble the Machine Visibility story with MRR and tier context.
MCP Team 5 · Lucas, Marcos, Peter & Flor

What We Said in Bilbao

The slide this team published at the offsite showcase — “AI bots already drive ~30% of renders and reach 1 in 3 customers — but we serve indexing bots far worse than readers (64% vs 97% cache hit)”

AI Bots Are Already a Meaningful Share of Traffic
  • ~1 in 3 active customers had at least one AI-bot render in the last 30 days
  • ~30% of renders for top accounts came from AI bots
  • Most Growth and Starter accounts show near-zero AI-bot activity today — the exposure is concentrated in larger, content-heavy customers
  • The count likely understates true reach: stale integrations, gated content, and smaller sites all suppress the number
Two Fundamentally Different Types of Bots, Two Different Problems
  • Reader bots: on-demand, user-triggered — cache hit rate drives response speed
  • Indexing bots: continuous, autonomous — render quality determines whether content gets indexed at all
  • Volume is dominated by indexing bots; reader bots are smaller in volume but better served
We Serve Reader Bots Well. Indexing Bots Are a Different Story.
  • Reader bots: 91–97% cache hit rate, sub-second render times, error rates under 3%
  • Indexing bots: 64–75% cache hit rate, 3–4 second render times on miss, double-digit error and timeout rates
  • The highest-volume indexing bot has the worst cache hit rate of any major bot we measured
Cache Hit Rate Can Hide a Silent Failure for New Content
  • Accounts with 85–95% cache hit rates can still have 50–70% render-fail on uncached URLs
  • When an indexing bot hits a page that isn’t cached yet, it fails silently and moves on — that content never gets indexed
  • Cache hit rate masks the problem entirely; render-fail-on-miss is the metric that surfaces it
MCP Team 5 · Lucas, Marcos, Peter & Flor

How Far AI Bots Already Reach

Finding 1, re-run in July — a sampled estimate, not a full census.

Active customers with an AI-bot render
~830
of 2,522 — about 1 in 3, last 30 days
Share of renders at top accounts
~30%
of all renders came from AI bots
Reach among customers with any traffic
~43%
~24% of “active” customers had zero traffic in the window
Where the Exposure Sits
Most Growth and Starter accounts show near-zero AI-bot activity today — exposure is concentrated in larger, content-heavy customers.
Method — and Its Limit
Based on a random sample of 258 active customers (33% had ≥1 AI-bot render, 95% CI 695–970), because scanning all 2,522 accounts hits ClickHouse’s per-query scan cap. Existence-check per customer for AI-crawler-token user-agents in ClickHouse prerender.analytics, trailing 30 days. The true rate is sensitive to how “active” is defined.
MCP Team 5 · Lucas, Marcos, Peter & Flor

Why That Number Is Probably Low

Two integration issues mean some AI-bot traffic never reaches Prerender at all.

Cloudflare Worker Integration Timing
The documented Cloudflare Worker integration template didn’t include AI-bot user-agents (GPTBot, ChatGPT, ClaudeBot, Perplexity, Amazonbot) until they were added on 10 Feb 2025 — Gemini and google-extended followed later, on 21 Aug 2025. Accounts that set up this integration before that date likely never allowlisted AI bots, so that traffic is never forwarded to Prerender. Those accounts show 0 AI-bot renders even if bots are actively crawling them, unless the users proactively modified their integrations.
CloudFront Header Forwarding
CloudFront customers whose cache or origin-request configuration doesn’t forward the User-Agent header can’t have bot detection trigger at all for that traffic — regardless of when they integrated. That traffic is invisible to us entirely. User-Agent is one of five required forwarded headers in our own CloudFront docs.

Sources: revision history of the GitHub Gist linked from the Cloudflare integration v2 docs; the CloudFront integration docs, confirmed in the workshop continuation thread in July 2026.

MCP Team 5 · Lucas, Marcos, Peter & Flor

Two Bot Types, Two Different Problems

Before serve quality, the split that makes the rest of the story readable.

Reader Bots — ChatGPT-User, Perplexity-User, Claude-User
Browse pages in real time when a user asks a question. Cache hit rate drives response speed.
Indexing Bots — GPTBot, ClaudeBot, Applebot, Meta-ExternalAgent, Bytespider, PerplexityBot
Crawl autonomously to build AI search indexes. Render quality determines whether content gets discovered at all.
Volume Is Dominated by Indexing Bots
Indexing bots
~77%
~10.2M AI-bot requests
Reader bots
~23%
~3.0M AI-bot requests

Sample of 220 accounts confirmed to receive AI-bot traffic, over a trailing 30-day window in ClickHouse prerender.analytics. Sample totals were extrapolated ~14× elsewhere in the analysis to approximate fleet-wide volume. Directionally representative, not an exact count.

MCP Team 5 · Lucas, Marcos, Peter & Flor

Serve Quality, Bot by Bot

The extremes are stark. The middle is messier than the Bilbao slide claimed.

Bot Type Cache hit Render on miss Errors / timeouts
ChatGPT-UserReader97.1%~406ms0.4% errors
PerplexityBot1Indexing91.1%
Perplexity-User2Reader83.9%not measurednot measured
Claude-User2Reader77.3%not measurednot measured
GPTBot, ClaudeBot, Meta-ExternalAgent, BytespiderIndexing70–75%2.8–3.6sdouble-digit
Applebot3Indexing64.3%~3.6s11.3% timeouts
Anthropic-AI4Indexing36.4%
1Classified indexing, but sits inside the top reader-bot tier. Either PerplexityBot is called out as an exception, or the reader/indexing split needs a clearer dividing line before this goes in front of the all-hands. 2The two lowest-volume reader bots — Claude-User 9,646 requests, Perplexity-User 6,008. Render time and error rate are missing for both, so “sub-second renders, low error rates” for reader bots is only actually demonstrated for ChatGPT-User. 3The single highest-volume bot measured, and the worst served among bots with meaningful volume. 4Lower still, but at just 778 requests — excluded as noise rather than a counterexample. ClickHouse prerender.analytics, 220-account sample confirmed to have AI-bot traffic, trailing 30 days. Directional, not exact.
MCP Team 5 · Lucas, Marcos, Peter & Flor

Cache Hit Rate Hides a Silent Failure on New Content

The most counterintuitive finding — a within-AI-bot-traffic effect, isolated to AI-bot requests only.

Fresh content is the content most likely to fail.

Northern Tool
94.9% AI-bot cache hit rate — and a 69.2% render-fail rate on cache misses.
EndPrize Solutions
85.7% cache hit — and 54.1% render-fail on misses.
Why Nobody Notices
Misses are a small share of total AI-bot requests, so these failures barely move the blended cache-hit percentage. Nothing in the top-line number prompts anyone to look closer — and on a dashboard blending all crawler types, AI-bot miss failures dilute further still. Render-fail-on-miss is the metric that surfaces it.

Caveats: drawn from 7 of the top-10-MRR accounts (3 omitted for near-zero AI-bot traffic) — not fleet-wide, may not generalise to smaller accounts. A deeper look at one other account in this dataset found its render-fail-on-miss was mostly expected 3xx redirects (locale/canonical routing), not real breakage; Northern Tool and EndPrize haven’t been checked the same way, so 69% and 54% could be similarly overstated. Render-on-miss is slow regardless of outcome — 6–13 seconds across these accounts.

MCP Team 5 · Lucas, Marcos, Peter & Flor

We Can’t Really Be Sure We’re Serving the Real AI Crawlers

Finding 5 — network-wide, Jul 21–24 2026, sampled as four 12-hour slices.

Probes for secrets in four days
337,049
~84K/day for .env files, cloud keys and config — 37.6% drew an HTTP 200
Provably spoofed, of AI-crawler traffic
~0.40%
116,007 of 28,657,173 requests
Provably spoofed, of all crawler traffic
~0.08%
116,007 of 141,505,954 requests
What the Spoof Test Can Prove
Only requests to bare-IP hosts, or with duplicated / proxy-chained user-agents. Within the AI cohort the spoof is almost entirely proxy-chained or duplicated UAs; bare-IP hits are ~0 there. The dramatic bare-IP scanning wore a Googlebot UA, so it counts under search engines, not AI. For context: all hard-spoof across every crawler is ~0.29% of crawler traffic (416,063), and AI crawlers are ~20% of all crawler traffic.
Three Things to Keep This Honest
  • This is a floor, not the real spoof rate. The common case — an attacker sending a clean GPTBot or ClaudeBot UA against a normal domain — is invisible here; confirming it needs the client IP checked against each vendor’s published ranges, which this table doesn’t store. True spoofing is higher, possibly much higher.
  • The earlier “almost certainly impersonation” call was about the sensitive-file-probing subset, not all AI traffic. Most AI-crawler volume is ordinary content crawling.
  • Don’t read 0.40% as “spoofing is negligible” — read it as 0.40% is provably forged; the rest is unverified, not verified-clean.

Hypothesis: it’s a very small portion that might get unproportional huge attention in AI Insights. Source: IRIS ClickHouse prerender.analytics, Jul 21–24 2026, network-wide.

MCP Team 5 · Lucas, Marcos, Peter & Flor

Each Bot Behaves Differently

Finding 6, and the questions we’d resolve before this goes any further.

We cannot say that bots behave in a similar manner when distributing their crawl budget.

The Product Question
Crawl-budget distribution differs bot by bot — we compared them on the Disney account. We wonder whether surfacing this to the user would bring them value.
Open Before We Go Further
  • Settle the reader vs indexing dividing line, or name PerplexityBot as an explicit exception
  • Get render time and error rate for Claude-User and Perplexity-User
  • Re-check Northern Tool and EndPrize render-fail-on-miss for expected 3xx redirects
  • Optional next cut: sensitive-file-probing traffic as a share of AI and of all crawler traffic — a rough earlier estimate puts it around half a percent of AI traffic, and it can be computed precisely
MCP Team 6 · Taki, Rafael & Moss

Churn-Risk Watchlist

Which high-MRR accounts are silently degrading — the watermelon: green outside, red inside — and what is the save-list plus outreach?

The Goals
Shortlist the High-MRR Accounts
Start from where the revenue actually sits.
Find the Watermelon
Green on paper, degrading render health underneath.
Produce a Save-List
With the outreach angle attached to each account.
MCP Team 6 · Taki, Rafael & Moss

What We Said in Bilbao

The slide this team published at the offsite showcase — “7 High-MRR Accounts Are Silently Degrading Watermelons — 2 Renew This Week and Need Outreach Now”

Act This Week

Trigger immediate outreach on the two time-critical renewals: Betika (Jun 27) and Radio Rottu (Jun 30).

Both are paying overages with no upgrade conversation underway.

Save-List Plan

Hand the four chronic $349-Pro overage-payers to CS & Account Management as Stage-2 Enterprise-upgrade cases.

Queue the remaining accounts by technical urgency.

Separate Tracks
  • Roobet — high-urgency technical case needing support outreach (not upsell) after an unrecovered May 20 cache clear.
  • Chips.gg (12% cache hit rate, URL explosion) — needs crawl/hygiene fixes before any upsell conversation.
Confidence — Medium
Cache-hit findings are clean and trusted, but the list only screened ~11–22 accounts — not the full high-MRR population. Several accounts have paid $1,700–$3,200/mo in overages for 12–24 months.
MCP Team 6 · Taki, Rafael & Moss

What Happened to the 7 Flagged Accounts

Followed up on 27 July, one month after the offsite watchlist was published.

All 7 renewed. Zero cancellations.

But two lost ~90% of their revenue without cancelling, and one $18,000 invoice is unpaid.

MCP Team 6 · Taki, Rafael & Moss

The Watchlist, One Month On

Every account on the Bilbao list, with its renewal outcome and how its overage moved.

Account Flagged as Renewal Outcome Overage then Overage now
BetikaDegrading, acuteJun 27Renewed, paid$2,875$4,489 (+56%)
Radio RottuChronic, recache loopJun 30Renewed, paid$1,950$5,731 (+194%)
elenastrinasparcoDegrading, perfJul 17Renewed, paid$1,801$2,253
Sanistål A/SChronic, 2+ yrsJul 13Renewed, paid$2,440$2,498 (flat)
Swank Films FRDisengagedJul 2Active — $349, zero overage$4,064$0
insight.coSoft-watch onlyJul 18Active — $729$4,105$380 (−91%)
digitelematica / eurospinWatch, unpaid invoiceJul 9$18,000 still unpaidn/an/a

The urgency was right but the window closed — Radio Rottu’s bill has nearly tripled. And one month is too short to score the screen: these are month-to-month plans, and the one confirmed churn signal has ~88 days of lead. Worth re-checking in September.

MCP Team 6 · Taki, Rafael & Moss

We Can’t Predict Who Will Cancel — Two Signals Survived

Nine possible causes were tested against a standard set in advance. One passed; a second holds with live examples.

A Customer Moving Themselves to a Cheaper Plan
  • A customer who downgrades cancels at 1.7× the normal rate, about 88 days later — roughly three months to react
  • Only shows up in mid-sized customers (~$1,000–$10,000 a year). No sign of it under $1,000 a year
  • The ceiling is the data, not the technique: 140 prediction models were tried, and the best was barely better than guessing
A Sudden Enormous Bill
  • Being permanently over the plan limit does not cause people to leave — those accounts cancel at the normal rate
  • A one-off spike is different. Worst 5% of months: ~1.5×. A bill roughly 7–8× the plan price (worst 1%): ~2.2×
  • Survived controls for the pricing migration and for declining usage — one of only two signals that did
  • Likely undercounted: this product is set-and-forget, so customers often don’t notice the bill until well after it lands

Live now: Betika was billed $4,489 in extras on a $349 plan (12.9×). Radio Rottu, $5,731 on a $349 plan (16.4×). Both past the worst-1% threshold, both still climbing, no intervention on record. Caveat: very few accounts sit in that band — suggestive, not proven.

MCP Team 6 · Taki, Rafael & Moss

We Already Ask Customers If They’re Happy

The rating tells us nothing about who will leave. The comment box does.

UMUX responses in the dashboard
111
June 2025 – July 2026
Cancellation rate, unhappiest to happiest quarter
No pattern
36%, 40%, 41%, 38%. Excluding the pricing change, the unhappy ones left slightly less often (relative risk 0.85×)
Rows in the feedback table
0
Responses live only inside the survey tool — not in Postgres, not in HubSpot, not in IRIS. No read integration in the codebase
The Written Comments Are Different
  • 6 people said outright they were looking to leave — 4 did
  • 7 complained the free plan was scrapped — all 7 left
Two Gaps in Who Gets Asked
  • The survey requires the customer to have logged in within 90 days — so the people drifting away never see it
  • No survey fires on a bill jump, a downgrade or a usage drop. Every trigger is time-since-signup
MCP Team 6 · Taki, Rafael & Moss

About a Third of Accounts Are Billed for Rendering Broken Pages

They don’t leave over it — and it only costs them money if they’re near their limit.

Spend ≥10% of monthly allowance on error pages
29.5%
of accounts; 14.8% spend a quarter or more
Typical error share — cancelled vs retained
2.6% / 2.1%
Indistinguishable, and even the worst 10% of each group match
Target list — big and wasteful
64 of 531
accounts (12%)
Only the Biggest Quarter Actually Pays for It

Comparing accounts of similar size, extra charges only rise in the biggest quarter (over ~283,000 renders/month): 69% of the wasteful ones get billed extra vs 49% of the clean ones, and 39% pay more in extras than their plan costs vs 21%.

In the three smaller size groups, wasteful accounts paid the same or less. In the smallest, 3% vs 12%. Under the limit a wasted render is free; over it, every one is billed.

Customers Do Notice
One wrote that we “render anything that soft 404s and charge for it.” Of 8 who complained about this, only 2 left.
MCP Team 6 · Taki, Rafael & Moss

The Watermelon We Actually Found

Both of these accounts are green on any subscription dashboard.

A customer can stop paying us almost everything without ever cancelling.

Two Live Examples
  • Swank Films — extras went from $4,064/month to zero. Subscription still active.
  • insight.co — $4,105 to $380. Still active.

No churn tool catches this, because nobody cancelled.

Two Different Risks
On accounts where extras run 8–17× the plan fee, revenue risk and churn risk are different things — and only the churn one is currently being watched.
MCP Team 6 · Taki, Rafael & Moss

What We’d Do Next

Ideas coming out of the follow-up, plus what the analysis does and does not cover.

Suggested Actions / Ideas
  • Bill spike detector — not only upwards, but maybe also downwards. Might be a feature idea for internal use
  • Diagnostics project — already under way; POC/MVP soon to be scoped
  • Email warnings — there is a HubSpot workflow that warns customers about render limits and overages; possible expansion
  • Opportunistic surveys and AI analysis of replies — not sure if this exists, might be interesting. Could trigger at crucial points, like when downgrading plan or reaching render limit
Sources and Caveats
  • The downgrade signal comes entirely from Alexander Malyshev’s self-serve churn analysis (14 Jul). The bill-spike finding is his too, with live examples added. The survey, waste and silent-revenue-loss items were verified directly against live data for this follow-up
  • The waste analysis used a 30-day exposure window (March 2026) with outcomes measured after 1 May 2026, deliberately avoiding the Oct–Nov 2025 pricing migration — which produced ~89,700 starter-plan cancellations and would otherwise swamp any signal. So it describes accounts that survived the repricing, not the whole customer base
  • Implementation note for anyone wiring this up: the survey tool receives login_users.id, not users.id. Billing is keyed on users.id. Joining on the wrong one gives a silent partial join, not an error

These are month-to-month plans and the one confirmed churn signal has ~88 days of lead — one month is too short to score the screen. Worth re-checking in September.

MCP Team 7 · Belmin, Charles, Jácint & Venessa

Platform Pain — The Timeout Club

Who are the worst-timeout accounts, and is it our infra or their site?

The Goals
Rank the Worst
Rank the worst-timeout accounts.
Us vs Them
Diagnose us (our ceiling) vs them (their origin).
Triage the Fix
Own-it vs route-back — leading with the concentration story.
MCP Team 7 · Belmin, Charles, Jácint & Venessa

What We Said in Bilbao

The slide this team published at the offsite showcase — “~10 Accounts Drive ~53% of Platform Timeouts — and It’s Customer-Side”

Cause: Mostly Them, Not Us
Origins return 200/404 responses, renders run to full budget, and timeout rates are uniform across our nodes and regions. Concentration is extreme — DSW alone is ~18% — so fixing a handful of accounts moves the global number the most.
Breakdown of the Problem
  • ~29% are soft-404s burning render budget on dead URLs (DSW, Platekompaniet, MOSH, Goody)
  • ~16% are default-ceiling configuration issues (Figma, Moonrail, FleetPride, Brora)
  • ~8% are genuinely slow origins (888/Evoke, Enlabs)
Recommendation: Split “Us” vs “Them”
Ship the fixable ~45% now: short-circuit soft-404 renders, and raise default 20s timeout ceilings to 30/60s for affected accounts. Launch customer outreach via HubSpot with a render-readiness playbook.
Confidence — High
Customer-side cause confirmed by three independent checks plus uniform-across-region evidence, with concentration figures matched by two analysts.
MCP Team 7 · Belmin, Charles, Jácint & Venessa

The Verdict: It’s Their Sites, Not Our Infra

12 accounts · ~1.9M failed renders in 7 days · one root cause.

We add ~0ms of delay. The wait happens inside their page.

Renders Don’t Run Slow — They Hang
When crawled, each page either renders fine in seconds or hangs until our clock runs out and we cut it off. Not “a bit slow” — stuck. True across all nodes and both regions.
Why It Matters
  • An SEO and AI visibility leak on our top enterprises
  • They’re billed for the failures
MCP Team 7 · Belmin, Charles, Jácint & Venessa

The Long Tail

Every account is the same shape — and it isn’t the shape a timeout ceiling can fix.

One Shape, Every Account
Most renders finish fast, a large share hang to the deadline, and almost nothing sits in between.
The Time Limit Is Not the Problem
The cutoff is already 5–90× a normal render, and accounts on a 60s budget still hang. More time won’t fix a page that is stuck.

This is where July departs from Bilbao. The offsite recommendation was to raise the 20s ceiling to 30/60s; on fresh data, the accounts already sitting on a 60s ceiling — currently Mosh, Enlabs and iFAST (888 moved off) — hang anyway.

MCP Team 7 · Belmin, Charles, Jácint & Venessa

Who’s in the Club

Two ways to read “worst,” side by side.

By Volume — June

FleetPride + DSW ≈ 770k timeouts.

Both since fixed — DSW went from ~18% of fleet timeouts to ~zero, FleetPride collapsed to 0.37%. The roster genuinely changes as accounts get fixed; that change is real signal, not methodology noise.

By Severity
  • UpToDate — 98%
  • Universal Orlando — 93%
  • iFAST — 84%

Barely rendering at all.

The prioritization rule: MRR × severity, not raw volume.

MCP Team 7 · Belmin, Charles, Jácint & Venessa

Root Cause — and Who Owns the Fix

Two customer-side causes, inferred from render-time shape. Direct per-render confirmation needs engine logs.

Never signals ready
~63%
window.prerenderReady never set
Request never finishes
~25%
A third-party or network request hangs
Combined
~88%
The #1 cause every day, in both regions

The fix lives on the customer’s side. Our job: detect it, hand them the playbook, and make render-readiness part of enterprise onboarding.

MCP Team 7 · Belmin, Charles, Jácint & Venessa

Timeout Club — Standard Definition v1

One canonical way to produce the list, so every analysis returns the same clients for the same window. No hand-rolled queries.

The Definition
  • Source: public.domain_metrics only — the hourly per-domain rollup
  • Window: trailing 7 days ending at report date, re-run monthly, with a trailing-14-day persistence gate to filter one-off crawler bursts
  • Population: on-demand renders for the outreach ranking — the crawler and SEO-facing traffic
  • Filters: exclude staging hosts; require ≥ 5,000 timeouts
  • Ranking: priority = (MRR / 100) × timeout rate, tiebreak on absolute timeouts; top 5 for outreach; deprioritize $0 / no-plan accounts
What It Guarantees — and What It Doesn’t

Same window, same clients. That kills the “everyone’s analysis disagrees” problem.

It does not produce the same companies every month. The roster genuinely changes as accounts get fixed or newly break — that change is real signal, not methodology noise. Never quote a prior month’s list for current outreach.

Enforced by one materialized view (PRE-3213), one saved query, one owner.

Re-verified on fresh 7-day data on 27 July. Every output is labelled with its exact dates.

MCP Team 7 · Belmin, Charles, Jácint & Venessa

The Repeatable CS Workflow

A monthly loop, and the deliverable of this workshop. Future state: this becomes a dashboard — manual now, automated later.

Diagnosis done and verified.

Next: ship the readiness playbook, then run the CS loop on the top 5.

Rank
The top 5 by revenue × severity.
Diagnose
Each account on its own.
Build
The customer-ready case.
Reach Out
But only once we have the fix to hand them.
Presenter: Joe

Celebrating Mistakes

Mistakes are how we build a stronger, more resilient team. Let’s celebrate the courage it takes to try, fail, and improve.

Step 1
Submit your mistake and learnings using the form
Step 2
The winner gets a gift card
Step 3
Submissions can be anonymized if preferred

No entries this month — but that doesn’t mean we didn’t make mistakes. Let’s keep this going, so everyone feels comfortable sharing what went wrong and what they learned from it.

Presenter: Joe

Miscellaneous Topics

World Cup Sweepstake Results 🏆

Congratulations to Giovanni (Spain), who wins a $100 gift card.

And a round of applause for Peter (Turkey), the first to bow out of the tournament. Peter — we’ll donate $100 to a charity of your choice for taking part.

New Tool Access Request Process

Submit an issue in Linear via the ACCESS team board.

Include the tool name, why access is required, and the urgency of the request. Follow the same process for requesting an entirely new tool.

Presenter: Joe

Q3 People Survey — Results

19 participants. Overall sentiment is strong and positive.

“I can see myself at Prerender next year” “Prerender is a great place to work” “My workload feels manageable and sustainable” “I feel informed about what’s happening across Prerender” 8.6 8.3 7.6 7.5 0 2 4 6 8 10
Feeling Informed — 7.5
One of our lower-scoring areas. We’re bridging it with the #fromjoesdesk weekly overviews.
Manageable Workload — 7.6
Before we jump to fixes, we want to understand what’s really driving this: is it specific to certain teams, tied to resourcing gaps, or something else? We’ll dig in and come back with next steps.
Prerender · Monthly All-Hands

Thank You

See you next month. Keep building.

July 2026