Sprint - Feb 10 - Feb 21

[timgl]

Sprint Planning

List of incidents + any action items that haven't been scheduled yet

• JS Static assets not loading
  • https://app.incident.io/posthog/incidents/314
  • https://app.incident.io/posthog/incidents/315
  • Actions:
    • integrating canary alerts for prod environments to PagerDuty as a last to-do.
    • No incident analysis as far as I can tell?
• CDP services unhappy
  • Simple roll back while rolling out, no further action required I think
• Livestream outage
  • Root cause changing broker addresses without changing prots
  • We have alerting on livestream
• Billing service slamming US Cloud
  • Zach "put in a few things to make it better next time" (just checking what)?
  • We changed the rate limiting and haven't seen it since

Team sprint planning

For your team sprint planning copy this template into a comment below for each team.

# Team ___

**Support hero:** ___

## OKR

1. OKR, status (red/yellow/green) and action points if yellow/red

## Retro

<!-- Grab the high and low priority items from last time and add whether that item was completed or not -->
- 

### High priority

-

### Low priority / side quests

-

[k11kirky]

Team LLM Observability

Support hero: Peter

OKR

To be defined

Retro

Red: Support wasn't set up as a separate category so I missed a range of issues

• Did not test with the various configurations of ai app setups
  Green: We rolled out support for new SDK's pretty quick for customer requests

Next Sprint

High priority

• Any Support issues that prevent people from using LLMO
• Users tab on LLM Observability (persons based view to see top users + cost)
• Break out user input from system prompt
• Support all major providers on JS / PY
  • OpenAI
    • beta endpoints
  • Anthropic
  • Bedrock
  • Langchain
  • Vercel
    • generateObject
• Cost overrides for models we don’t support
• Documentation, tutorials, examples and cookbooks

Low priority / side quests

• Test Coverage

[ioannisj]

Team Surveys

• Support hero: @ioannisj
• No planned PTOs

🟢=finished 🟡=in progress 🔴=won't finish ⚪=not started

Retro

Retro - Team Notes

• Got lonely and a bit lazy and missed a couple of standup updates during Luca's onboarding week. Sidetracked from high priority tasks because of other important SDK issues that surfaced @ioannisj
• in person onboarding is awesome but tiring as well (jetlag was horrible / low productivity ) @marandaneto (+1 @lucasheriques)
• feels like we have to ramp up the skills on debuggability across the tech stack from SDK/FE to API and DB @marandaneto (+1 @lucasheriques)
• support hero has been rough - context switching between tickets and other work is hard. lot of time lost on debugging, trying to consolidate what I learn on the contributing.md file @lucasheriques

Retro - Tasks

High priority

• 🟢 In-person Lucas onboarding @marandaneto @lucasheriques
• 🟢 Survey targetting platform, eg Mobile only, Web only, etc @marandaneto @lucasheriques
• 🟢 🐛 Unable to Save/Launch Survey @lucasheriques
• 🟡 Reach out to AEs for customers actively using surveys, start building a closer relationship with them @lucasheriques
• 🟡 Lay the foundation for Mobile surveys on iOS and React Native @marandaneto @ioannisj
  • 🟡 The goal for this sprint is that surveys are loaded and cached with all conditions in React Native
  • 🟡 iOS survey display logic (will likely have something by Friday on this)
  • ⚪ iOS survey question branching logic (will likely roll-over to this sprint)
• ⚪ 🐛 Surveys not capturing responses @lucasheriques
• ⚪ 🐛 When filtering survey responses, the charts are not updated @lucasheriques
• 🔴 🐛 Wrong graph series after changing question order @lucasheriques
  • Figure out if the originalQuestionIndex coming from the BE is persisted since the survey first iteration to fix the responses
• 🔴 🐛 Changes are lost on save when editing a survey that contains errors @lucasheriques
• 🔴 🐛 Required answer on a survey @lucasheriques

Low priority / Side quest

Surveys:

• 🟢 🐛 Survey was shown to only one user, but it says 3 unique users have seen it @lucasheriques
• 🟢 🐛 Requested URL is incorrect when PostHog host includes a path @ioannisj
• 🟢 🐛 fix: failing XCFranework builds @ioannisj
• 🟢 🐛 fix: properly identify Mac Catalyst apps @ioannisj
• 🟢 Set locale and timezone info if not using Expo @marandaneto
• 🟡 🐛 Support multiple instances for iOS SDK - Customer requested this and it also popped up quite a few times these last two weeks + parity with Android SDK @ioannisj
• 🟡 🐛 Inconsistent Color Scheme for NPS Feedback Categories @lucasheriques
  • 🟢 Follow up to see if it's still needed as implementation changed
• ⚪ 🐛 Surveys - Increase the number of allowed choices for single and multiple choices questions @lucasheriques

Session replay:

• 🟢 🐛 fix: session replay and auto capture works with 'with' method @marandaneto
• 🟢 🐛 fix: do not allow default integrations multiple times @marandaneto
• 🟢 🐛 fix: touches fall back to single touches if out of bounds @marandaneto

OKR

• 🟡 Improve quality
• 🟡 Mobile Surveys
• ⚪ Talk to customers
• ⚪ Finish up WIP features

High priority

• iOS survey display logic @ioannisj
• iOS Basic type popup display @ioannisj
  • major focus will be on how to properly present this in the host view hierarchy without breaking navigation
• RN - survey display logic and start with survey question branching logic @marandaneto
• Allow a survey to pop up repeatedly @lucasheriques
• feat: new filters for surveys answers @lucasheriques (replaced the previous one)
• Send intro to up to 5 customers to get feedback based on the questions we prepared @lucasheriques
• When filtering survey responses, the charts are not updated @lucasheriques
• 🐛 Wrong graph series after changing question order @lucasheriques
  • Figure out if the originalQuestionIndex coming from the BE is persisted since the survey first iteration to fix the responses
• 🐛 Changes are lost on save when editing a survey that contains errors @lucasheriques
• 🐛 Required answer on a survey @lucasheriques
• 🐛 Surveys not capturing responses @lucasheriques
• Surveys - Increase the number of allowed choices for single and multiple choices questions @lucasheriques

Low priority / Side quest

• 🐛 Build fails for React Native when compiled as a static library @ioannisj
• Adds referrerURL automatically @ioannisj
  • This popped up this last two weeks. Will investigate
• Surveys: customize text color of feedback widget @lucasheriques
• 🐛 Session replay won't start/stop if reloadFeatureFlags is called and session recording flag flipped @marandaneto

[benjackwhite]

Team Can Delete Plugins (CDP)

• Support: @benjackwhite

Retro

• ✅ Inline plugin destinations @MarconLP and @benjackwhite
  • All there - will be testing if they scale by end of week
• ✅ Rollout new Ingestion services and remove old code @benjackwhite
  • We did it! Didn't remove the old code as we are waiting till all plugins are gone for this
• ✅ Get Team 2 using Hog transformations and get all existing transformations translated to hog functions @meikelmosby
  • All the missing features for parity are there with "hog-plugins" setup and tested
• 🟡 Google Ads and LinkedIn ads destinations finally out! @MarconLP
  • Available in early access, but not fully released
• ✅ Get proper production metrics for Blobby V2 on production @pl
  • Specifically things like write costs and batch efficiency so we know how complex a solution we actually need
  • Play with the various settings to get a feeling for what the right balance is
  • Ideally get to the point where we are writing actual S3 files (albeit to a test bucket)
• 🟡 Made the changes to be able to work off of WS but not actually rolled out

Discussion

• CI - flakey plugin server and E2E cypress tests are flakey and slow

OKRs

• 😍 People choose us for their CDP first 🟡
• 🖕 Plugins be gone 🟢
• 🚀 Warpstream for events 🟢
• 🎈 Write-optimized session replay ingestion 🟢
• 💪 We make changes and improvements to the Node services with confidence 🟢

High priority

• Onboarding Eli! (everyone)
• Delete plugins code @benjackwhite
• Delete onevent and webhooks services @benjackwhite
  • Migrate Zapier stuff to Hog Functions (this allows us to finally kill webhooks services)
• Read the new S3 data for comparison and write the relevant data to Clickhouse @pl
• UX improvements to release customisable hog transformations @meikelmosby
  • Top 5 common requests that could be solved with new transform templates
  • Making editable transformations a reliably experience
• Move existing destination templates into nodejs @MarconLP
• Put up TikTok and Bing pixel site destinations @MarconLP

Low priority

• Come up with a plan to test capture Warpstream for real (even if it is just for a few hours)

[pauldambra]

Team Retain-you

Support hero: @pauldambra

Retro

• can we do a better job of tracking cool ideas? @annikaschmid
• OMG support load @pauldambra
• CI has been horrible @pauldambra
• good to see validation that "i have too many to watch" is a real customer problem to solve @veryayskiy

## what we shipped

Replay

High priority

• 💀 Goal 4: Squash distractions
  • 🟡 ship our own rrweb https://github.com/PostHog/posthog-rrweb @pauldambra
    • replay side out
    • so many bugs every time we make changes here
  • ✅ fix the safari bug @pauldambra
• 👩‍🔬 Goal 2: Churn and retention exploration
  • ✅ more cross selling into replay (probably from experiments) @veryayskiy
    • we did it from web analytics
    • for experiments exists already - going to add the new cross sell metric
  • ✅ measure that cross selling @veryayskiy
  • 🟡 propose first version of replay churn metrics @veryayskiy

Side quests and lower priority

• 🎤 Goal 3: Watch less and get more value
  • 🪓 internal and test account filtering in a personless world @pauldambra
    • just deprioritised based on other load
  • 🟡 first version of "chat with replay"

OKRs 🟡

1. OKR, status (red/yellow/green) and action points if yellow/red

• ✅ 🚀 Goal 1: Get Alex rocketing
• 🟡 👩‍🔬 Goal 2: Churn and retention exploration
  • talking to users and picking metrics
• 🟡 🎤 Goal 3: Watch less and get more value
  • in the short term, looking at playlists and filters to see what we can get
• 🟡 💀 Goal 4: Squash distractions
  • need support to be less tricky things heavy

High priority

• 👩‍🔬 Goal 2: Churn and retention exploration
  • formalise first version of replay churn metrics @veryayskiy
• 🟡 🎤 Goal 3: Watch less and get more value shipping chat
  • get some test users on "chat with replay" @veryayskiy
  • how do we do "AI" opt-in settings @veryayskiy
    • is it one setting for all of PostHog.
    • Do Team AI product manager already want to do it?
• 🟡 🎤 Goal 3: Watch less and get more value making filters better
  • split saved filters and playlists @veryayskiy (with @pauldambra supporting)
    • also what the hell do we call them 🤣
      • collections folders pinned saved playlist
    • unwind some of the abstractions that makes this hard to work with
• 💀 Goal 4: Squash distractions @pauldambra
  • move E2E tests from Cypress to Playwright - #DevEx
  • re-bundle rrweb without post-css so we can test our own built in recording

Side quests and lower priority

• 🟡 🎤 Goal 3: Watch less and get more value
  • link from analytics (particularly funnels) to replay @pauldambra

[jurajmajerik]

Team Experiments

Retro

Anders:

• 🟡 Correct experiment results
  • Difficult to fix with existing query runners so -> RFC: New experiment query runner
  • WIP proof-of-concept of a new query runner
  • Some architectural changes required, important to align on work so we build towards the same future state -> Goal description and work break down
• 🟢 Growthbook deep dive
• 🟢 Misc
  • Violin plots
  • Stop throwing errors when there is no experiment data
    • Postponed as it makes sense to bake into the new query runner and UI changes

Daniel:

• 🟡 Graduate data warehouse integration out of beta
  • Added support for all account filters
  • Added support for a Stripe subscriptions table joined with customers table. This is set up with a shared metric in US production, but the query OOMs.
• 🟢 [customer] satisfaction (small things out of product council meeting):
  • Added tag and quick select support for shared metrics
  • Removed blocking results loading animation
• 🟢 Misc:
  • Added support for creating an experiment with an existing feature flag
  • Switched win probability to calculate against the control, not all variants

Juraj:

• 🟡 No-code experiments UI revamp
  • Fixed a bunch of bugs (e.g. 1, 2), refactored and got familiar with the codebase. Check the issue for the remaining work.
• 🟡 Start on credible interval timeseries chart - built a UI for this
• 🟢 Misc:
  • Added a p-value workaround for a customer until we implement Frequentist
  • Various UI fixes + set up a Storybook for the new UI

Sprint goals

Anders:

• Anders will own the new query runner and split the work into manageable tasks. We aim to pair every day to make quick progress and clarify the unknowns.
  • Support the trends metrics we do today (count unique, count total and continuous)
  • Data warehouse support
  • Two step funnels and explored how we want to implement multi-step funnels

Daniel:

• Support hero
  • Support Anders with new query runner implementation
  • Graduate data warehouse integration out of beta - blocked by new query runner

Juraj:

• Support Anders with new query runner implementation
• Fix no-code experiments:
  • Continue working through the UI and refactoring list
  • Make sure all transforms get correctly applied in posthog-js and write tests for this.

[MichaelKutsch-ph]

Team maINFRAim

Support: @danielxnj @MichaelKutsch-ph

Retro

• Move Django apps to new chart version: split out local evaluation @MichaelKutsch-ph
• continue argoCD alerts with prometheus alerts instead @bciaraldi
• Infra Costs - waiting on both GCP and Google for an offer/quote @MichaelKutsch-ph @frankh
• Get a argoCD workflow going for great DX for teams - helmfile + argoCD is in place. ArgoCD-only in the making
• Prep Warpstream for go-live. Delayed by CDP chart migration - will pick up next sprint @frankh
• 🟡 Kafka failover to s3 - tests are ongoing, look promising, waiting on review @frankh

OKR

1. 🏃 Next level ArgoCD - Continuous Delivery
2. ☁️ 🕸️ Scalable networking
3. 💰 Mr. Bezos, bring down those costs!
4. 🎮 A stable pre-prod environment

High priority

• Move Django apps to new chart version: surveys @MichaelKutsch-ph
• ArgoCD only workflow @MichaelKutsch-ph
• Move Django apps to new chart version: decide @danielxnj
• continue argoCD alerts with prometheus alerts instead @bciaraldi
• Infra Costs @MichaelKutsch-ph
• Prep Warpstream for go-live @frankh
• Kafka failover to s3 @frankh

Low priority / side quests

• move old capture to new capture @MichaelKutsch-ph
• Evaluation of Robusta.dev (better alerts, AI and fancy things...)

✅=finished 🟡=in progress 🔴=won't finish ⚪=not started

[robbie-c]

Team Web Analytics

Support hero: @robbie-c + @rafaeelaudibert

Retro

Main quest

Robbie

• 🟢 Add revenue tracking from event properties to web analytics

Rafa

• 🟢 Actual Core Web Vitals release
• 🟢 SDK performance
• 🟢 Basic Web Vitals on Toolbar

Side quests & support:

Robbie

• CDP Reddit ad conversion destination for 11labs
• Track cross-sell product intents
• Fixing flaky web analytics tests

Rafa

• Loading state skeleton overhaul
• Toolbar setup UX improvements
• Path cleaning UI/UX improvements

OKR

• Finish cookieless mode
• Build marketing analytics
  • Add revenue data to web analytics
  • Get ad spend data into data warehouse and use it for per-channel analytics
• Build per-page reports
  • Core web vitals

Sprint goals

High priority

Rafael

• Web analytics cross selling
• RFC on per-page reports

Robbie

• Continue helping out with 11labs CDP stuff
• Cookieless next steps - decide endpoint

[mariusandra]

Team Dev-T-Rex

OKR

• 🟡 Build a framework where we can have 50+ products? @mariusandra
  • 🟡 Defining and splitting the common layer from the rest of the apps
  • ⚪ Defining and implementing boundaries between apps
  • ⚪ UX changes to support the new structure
  • ⚪ Every product is able to export operations (backend) and widgets (components) => AI, notebooks etc.
  • ⚪ Testing independently - if you change things on one product, we only run tests for that product
• 🟡 Levelling up our product quality. @adamleithp
  • Design system that is self documenting
  • Mobile support
  • Finally figure out (explain) modals vs drawers
  • More standardised components
• 🟢 Anti-goal for Q1: @corywatilo
  • Fixing CI when it randomly breaks (each team owns their tests)
  • Rewrite everything
  • Implement everything ourselves

Retro

• Adam:
  • 🟢 Keep working on the design system.
    • Color system PR ready(ish). Merging this week or EOD.
  • 🌲 Project "look mom, no mouse"
    • Tree view good, keep on iterating on the rest.
  • ⚫ Standardize UX for highly used interactive elements
    • ongoing
• Marius: Gotta build fast
  • 🟡 Integrate nx for build dependency management and caching. Seriously consider nx cloud (watch this 10min video).
    • Got nx working, but decided to try turborepo instead ($5-10k/mo vs $0/mo).
    • Part 1 (pnpm workspaces) merged, now extracting and merging all other parts. Hoping to get turborepo merged still this week.
  • 🔴 Split lemon-ui from the frontend and bundle it as a single workspace package
    • After turborepo works, I'll split out the lemon UI.
  • 🔴 Map the boundaries between projects - for all languages
    • After turborepo works for node projects, I'll integrate the other languages
• Cory:
  • ⚫ Keep iterating on the designs.
    • Kept iterating

High priority

• Marius
  • Initial PR (or RFC) for "tree view" backend
  • Keep on building
    • Extract, bundle and test products separately from the rest of the frontend (and app in general).
    • Run as many tests tests in CI with turbo affected as we can
    • Package lemon-ui as a separate package
    • Integrate products in languages (python, rust) into turborepo
• Cory
  • Work on designs, build higher fidelity mockups (search, sidebar, etc)
• Adam
  • Work on implementing designs

Low priority / side quests

• Marius
  • Continuous Hog/QL support

[daibhin]

Team Error Tracking

@oliverb123 on batch imports for one week

Retro

🟢 505 users (up from 205) in beta across 430 teams
🟢 Node manual and autocapture ready PostHog/posthog-js-lite#366
🟢 Add PostHog manual exception capture to Django #27836
🟢 Much better docs (thanks to @ivanagas) PostHog/posthog.com#10527
🔴 Dropped individual exception loading & linking. Focused on redesigned issue page instead #28083

OKR

🟢 Migrate PostHog off Sentry
🟢 Ship the damn thing

High priority

Migrate PostHog off Sentry

• Alerting config for issue created & reoccurred @daibhin
• Rollout Node capture on plugin server @daibhin

Ship the damn thing

• Beta feedback & review @daibhin / @oliverb123

  • Create a dashboard in PostHog from usage reports
  • Watch session replays
  • Reach out to most active users
  • Setup 2-3 user interviews

• Small touchups @oliverb123

  • Fix Kafka offset bug (switch to using transactions)
  • Add first_seen timestamp to Django fingerprint model
  • Improve handling of deeply nested / malformed exception properties

Low priority / Side quests

• Python chained exceptions issue ordering
• [Carried over] Issue exception loading ("latest", "earliest" or uuid)

[timgl]

Team Batch Exports

Support hero: ___

OKR

• 🟢 Reliability/stability
  • Canary deployments
  • Have alerting + runbooks set up for most failure modes

• alerting for specific exports

• 🍊 UI/UX (haven't started yet)
  • Backfill UI
  • Split up logs into “user reporting + user actionable” and “internal logs”, only show the first group in the UI.
  • Make it obvious you need teams plan for 5 min exports
• Notifications (slack/email)
  • Tim to figure out what CDP is doing, come up with a plan
  • 🟢 Exports using HogQL
  • 🟢 Allow any model to be exported

Retro

• 🟢 Property filters for batch exports Property filters for Batch Exports #26415 (use hogql in the backend to do this) @tomasfarias - done
• 🟢 Export sessions table Feature request: allow sessions table to be exported via batch exports #25387 @tomasfarias - almost there, should be done before end of week
• 🟢 Use distributed events table for short term backfills @rossgray - should be done this week
• 🟢 Backfill UI Batch export: Backfill UI #27788 @rossgray - Done, want to make a few more improvements but it's mostly there

High priority

• Batch export: check connection before saving destination #28149
• S3 batch export: Parallelize upload for increased performance #28129
• Batch export: check connection before saving destination
• Help out data warehouse team

Low priority / side quests

[aspicer]

Team Product Analytics

🎉 @Sriram-bk joined the team! 🎉
🎉 @HamedMP starting next week! Onboarding in Munich week of 2/17 🎉

Q1 2025 objectives

1. Main Goal: Make it a no-brainer for customers to use our insights

   • Research and improve insights
     • [Sandy] Funnels
       • Filters in exclusion steps
       • Optional Steps
       • Time to convert trends
     • [Thomas] User Paths
     • [Anirudh] Retention
       • Revenue Retention - connect to data warehouse
     • Data management / Non-existing tools
   • [Thomas] Better path cleaning support
   • Taxonomic Filter Improvements
     • Better organized, easier to use
     • Information about properties (schema and statistics)
   • [Anirudh] Make other non-event sources a first-class citizen of the insight ecosystem
   • Metric to track - Reduce usage of HogQL as a workaround/crutch for missing UI features

2. [Anirudh] Small Goal: Alerts v2

   • Alerts on funnels
   • Notifications through Slack

3. [Sandy] Small Goal: Shared dashboards with passwords

4. [Sandy & Anna] Research Goal: Understand activation and contraction

Retro

Support hero: @thmsobrmlr @anirudhpillai

• [Anirudh] Retention
  • 🟢 Insight UX
  • 🟢 Retention on event properties
  • 🟢 (Bonus) Expanding date range
• [Thomas] 🟢 Start on paths insights changes
  • UI / UX Wins
• [Sandy] 🟡 Start on funnels changes
  • Time to convert trends
• [Sandy] 🟢 Fixing issues with hogql cohort calculation

Sprint Plan

Support hero: @aspicer @thmsobrmlr

• [Thomas] Improve user paths insight:
  1. Improve visual styling
  2. Implement "Other (i.e. all remaining values)"
  3. Improve path cleaning filters
• [Thomas] Onboarding Hamed
• [Sandy] Funnel Improvements
  • Filters in exclusion steps
  • Optional Steps
  • Time to convert trends
• [Sandy] Analyze support load
• [Sri] Ramping up and helping with support
  • Actors query fixes for first time user matching
• [Anirudh] Breakdowns on retention

[dmarticus]

Team Finally Launching A Groundswell SDKs (Flags)

Retro

Onboarding Lucas, finishing hackathon projects, starting work on the Java SDK, sparking customer joy, and finishing new flags

What went well

• @haacked – On vacation, the snow looks sick
  • 231 downloads so far of the PostHog Nuget package!
• @havenbarnes
  • Launched remote config flags, monitoring usage
  • Finished encrypted payloads for remote config flags, SDK changes still needed
  • Started new Java SDK derived from posthog-android
• @dmarticus
  • sparked customer joy
  • new flags are ready for real traffic! I wrote a synthetic test bed to verify how they work in both posthog-js and posthog-python and I've confirmed that person targeting works for every scenario. Am going to start rolling out real traffic this week – starting with team 2, and then ideally some percentage of real traffic later in the week

What could've gone better

• still tuning our decide deployments in EU, needed to make sure infra changes to handle the load
• support SLA rate went way up; our support engineers were underwater and I didn't do a good enough job helping out on the non-escalated SLA violators. Shoutout to Haven for jumping on that Monday and Tuesday.

This sprint

Themes: release new flags to production traffic, meaningful progress on Java SDK, continue to spark joy with small features, start working on behavioral cohort targeting

Time off: @haacked 5 days, @dmarticus 2 days

• @dmarticus

  • Roll out new flags to a meaningful percentage of production traffic
    • start with all traffic for team 2, then do % rollouts
  • Implement quota limiting for new flags
  • Finish non-production-rollout-blocking tasks for new flags
    • All "necessary for release" tasks from here: Feature Flags Rewrite: List of things to get done #22131
  • Deliver RFC for behavioral cohort targeting

• @havenbarnes

  • Launching encrypted-payload remote config flags PR
  • Launching new posthog-java SDK
  • More revs on observability work

• @haacked

  • Fix some issues found in other client libraries while developing posthog-dotnet
  • .NET SDK Beta release
  • .NET SDK documentation is published
  • Unblock Unity projects from using the .NET SDK
  • Support the ASP.NET Feature Management API

Support

• 2/10-2/17 @dmarticus
• 2/17-/24 @havenbarnes

Q1 2025 Objectives

(I'm going to turn these into sub-issues so I don't have to type so much)
✅=finished 🟡=in progress 🔴=won't finish ⚪=not started

🟡 Performance & Reliability

• 🟡 Complete migration of feature flag evaluation traffic from the legacy /decide endpoint to the new /flags endpoint (@dmarticus)
• ⚪ Establish comprehensive monitoring and observability for the feature flag platform (@dmarticus, @havenbarnes)
• ⚪ Enhance Grafana metrics to differentiate between infrastructure failures and logical errors (@dmarticus)
• ⚪ Collaborate with infrastructure team to create infrastructure cost analysis and dependency mapping with Michi (@dmarticus)
• 🟡 Improve observability of the feature flag evaluation process by tying flag evaluation results to the original request context (@havenbarnes)
• ✅ Experiment with sampling request parameters for high-value customers to have better observability for support requests (@havenbarnes)
• ⚪ Implement actionable Sentry alerts in #alerts-feature-flags channel (@havenbarnes)
• ⚪ Integrate PostHog error tracking with new flags endpoint (@dmarticus)
• 🟡 Implement read-only flags (specifics to be defined, see the following RFC) (@havenbarnes)

🟡 Customer-Driven Innovation

• ⚪ Behavioral cohort targeting for flags (RFC here) (@dmarticus)
• 🟡 Feature management (@havenbarnes)
• ⚪ Automatically manage flag staleness (@haacked)
• ⚪ Customer research-driven features (TBD based on the findings of the market research and customer interviews)

⚪ Strategic Research

• Identify and analyze key user segments:

1. Current active users
2. Competitor users
3. Custom solution implementers
4. Potential users unaware of feature flagging

• Conduct customer interviews to identify gaps and growth opportunities
• Document findings and recommendations for product roadmap.
• More context here

🟡 SDK Spike

• 🟡 Evaluate feasibility and tradeoffs around adding feature flags to the Java SDK
• ⚪ Investigate improving the Rust SDK (add feature flag evaluation, implement the suggestions from the open issues etc.)
• 🟡 Research .NET SDK demand + feasibility @haacked
• 🟡 Fix open issues in the Python and Node SDKs: (Python, Node)

🟡 User Experience Improvements

Low-hanging, medium-impact, and largely context-free fruit that can be picked up by any team member during available capacity:

• ⚪ Consolidate boolean and multivariate flag interfaces
• ⚪ Show flag evaluation results more clearly in the flag editor UI
• ⚪ Improve the experience of syncing flags between environments
• 🟡 Anything else from the current feature flag backlog

[zlwaterfield]

Team GrowthBAC

Retro

Retro items

• @zlwaterfield
  • RBAC
    • ✅ last client side PR
    • 🏗️ migrations - write + run for team 2
    • ❌ get team 2 live
  • 🏗️ customer.io emails set up for posthog (if time, get 2fa emails flowing)
  • 🏗️ metabase -> posthog move complete, this includes the DW view for invoice with annual
  • 🏗️ explore dagster more for usage reports
• @patricio-posthog
  • 🏗️ Mobile session replay pricing
  • ✅ Change period_start for annual invoices, to count them on the month the contract was signed
• @joshsny
  • ✅ Run experiment to push session replay to more users in onboarding (with the aim to make it default on)
  • ✅ Go through competitors onboarding (Mixpanel & Amplitude, June.so & Statsig)
  • ✅ Add cross selling instrumentation for use across the repo and add stats to growth dashboard
  • ✅ Finish off experiment to product analytics cross sell
  • ✅ Resurrect the quick start section: open it by default
  • ✅ Split quick start into Product Analytics, Feature Flags & Session Replay, so we are ready for adding other products (added surveys, data warehouse and experiments also)
  • RFC for paid plan by default + credits for quick start (nice to have) decided against doing this
  • Pet project: Eager query loading for top N queries that are grouped by interval day/week/month (unlikely to have time for). Turns out this is being actively worked on by @anirudhpillai so I am chatting with him

Extra

• ✅ improve how tax is consider in mrr (don't double count)
• ✅ moving around usage report times and rate limit to reduce failures (working?)
• ✅ debugging quota limiting issues (shipped lots of logging)

Q1 2025 Objectives

✅=finished 🟡=in progress 🔴=won't finish ⚪=not started

• 🟡 Usage dashboard (Pato)
• Pricing changes(Pato)
  • 🟡 Finish Mobile Replay pricing
  • ⚪ Query API pricing
  • ⚪ Clean up billing for identified / unidentified events
• 🟡 RBAC (Zach)
• ⚪ Enterprise add-on (Zach)
• ⚪ Improve SAML (Zach)
• 🟡 Use intent data to onboard users into multiple products (Josh)
• ⚪ Surface intent data later in the users journey (Josh)
• ⚪ Move core RevOps flows off of Zapier into the billing system (tbd)

Extras (nice to haves if we have time or feel inspired)

1. 🏗️ Postgres (pgbouncer + read replica) - infrastructure handling this
2. Self serve account deletion
3. Case insensitive emails
4. 🏗️ Move all transactional emails to Customer.io
5. 3D Secure improvements - email flows
6. 🏗️ Move revenue dashboards from Metabase into PostHog
7. Get credit cart input experiment over the line
8. Billing endpoint performance improvements (split up)
9. Proper callbacks for time-sensitive checks

This sprint

Time off: N/A

• @zlwaterfield
  • Support week 1
  • RBAC
    • fully roll out for team 2
    • misc improvements / fixes from team
    • roll out to % of new users
    • think more about enterprise add-on
  • customer.io emails set up for posthog (if time, get 2fa emails flowing)
  • metabase -> posthog move complete, this includes the DW view for invoice with annual
  • explore dagster more for usage reports
• @patricio-posthog
  • Support week 2
  • Finish Mobile Replay Pricing (backend almost there, needs mostly migrations for current customer, frontend needs some tweaking)
  • Check again Forecasting, it's jumping too much.
• @joshsny
  • Redo first page of onboarding (include all sources, highlight invites, search and other improvements)
  • Improve setup for NextJS users since it's the most popular SDK but it's awkward to setup
  • Redo plans page in onboarding
  • Brainstorm improvements for surfacing demo data in the app
  • Improvements to weekly report emails (if we get feedback within the sprint)

[Twixes]

Team AI PM

Support hero: both

OKR

Replace 10 full-time data analysts around the world with AI (we need a better metric actually).

Retro

High priority

• ✅ Core memory shown in the UI @skoob13
• ➡️ Archival storage @skoob13 - deprioritized, as we're evaluating the impact of memory
• ➡️ Stickiness/lifecycle insights @Twixes - deprioritized based on lack of user feedback
• 🔁 AI in "Add insight" on dashboards @Twixes - rollover
• Address user feedback:
  • 🟠 Keep showing RAG actions that were taken after insight creation is done @Twixes
  • 🔁 "Stop response" button @skoob13 / @Twixes - requires Django upgrade
  • ✅ Ability to type while assistant is generating @skoob13 / @Twixes
  • ✅✅✅ …and more (BIG ONE: shipped conversational UX)
• ✅ Dogfood our LLM observability product (shipped UX improvements, SDK reviews)
• ✅ New: Efficient query results format for agent @skoob13

Low priority / side quests

• ✅ Fix empty message just before schema generation kicks off @skoob13

High priority

• Side panel Max across the app @Twixes
• Natural language "Add insight" on dashboards @Twixes
• Semantic search on actions @skoob13
• Max able to use actions @skoob13
• History of conversations @skoob13
• Investigate Django 5.0 upgrade for straightforward "Stop response" button @skoob13
• More user feedback

[fuziontech]

Team ClickHog

Objective

Goals for Q1:

• Query Scheduling Service
  • Paid API use case is covered
  • We should have all queries running through this
  • Basic load shedding
  • Differentiate API queries based on $
• Iceberg continuation
  • Keep on top of Altinity builds for CH
  • Query logs into Iceberg
  • Ingestion pipeline work (monitoring & automation)
• Cluster topology consolidation and cost reductions
  • We are on one cluster
  • Query performance is equivalent
  • Costs are down
  • Schema management
• Ongoing HogQL/Performance improvements
  • Query Scheduling
  • Productizing Performance Tooling
  • Better query attribution (in query comments)
    • By end of query we do not run queries with no annotations
    • Some sort of Type for queries that will run

Board

https://github.com/orgs/PostHog/projects/85/views/2

Retro

🔥 Incidents 🔥

Happy cluster. Happy Life.

Last Sprint's Goals

• Deletes merged and running in Dagster @fuziontech ✅
• Houseplan forked and running some migration? @Daesgar ✅
  • Not on houseplant but the goal was complete (we can do schema management on coord)
• EU Squash job running from Dagster @tkaemming ✅
• Duplication investigation @tkaemming ❌
• Plugin server HTTP header bonanza @orian ♻
  • Changed approach and will do this more iteratively
  • Query limit is shipped
  • Tagging queries is shipped
• Pass query service headers from Django for rate limiting @orian ⏸
• Clickhouse in K8s @Daesgar ⏸
• BONUS ZK -> 1 AZ saving costs $$$ @Daesgar 🚀
• Cross join rework shipped @tkaemming ✅

This Sprint's Goals

TL;DR

• Duplication investigation @tkaemming
• Expose estimate for queries on API @orian
• Expose whether or not a new query is utilizing indexes @orian
• Start getting users on Query API and iterate & learn @orian
• Clickhouse in K8s @Daesgar
• Backups to Dagster @fuziontech
• Metabase dashboard for delete compliance @fuziontech
• Clean up weekend EU disk alerts @fuziontech @Daesgar
• Investigate Person Properties into a map (similar to event props) @tkaemming
• Look into slack sensor on Dagster for alerting on compliance job @fuziontech

[EDsCODE]

Team Data Storehouse

OKR Q1 2025

Objective

Objective: World class query experience

• Key Results:
  • SQL copilot
  • Draft mode
  • Data modeling and lineage controls

Objective: Data Visualizations

• Key Results:
  • Complete suite of charting options
  • External data for all insights

Objective: Extensible sources

• Key Results:
  • Repeatable and easy to build data import source pattern
  • Revised source configuration experience

Retro

Taking a sprint to do a bug bash

• Optimise external_data_sources endpoint (source) - @phixMe
• Paginating visualization table (it crashes if you go from data grid) - @EDsCODE
• Being able to set insight variables via URL parameters when linking between dashboards - @Gilbert09
• Instrumenting usage of editor
• Be able to edit source configs from the UI
• Be able to edit access keys/secrets of manual sources from the UI
• Reactive UI when using the manage sources page
• It says “Error” on new schemas when syncing a new source for the first time
• The status page of import sources flow doesn’t auto reload
• Display the name of insight variables when viewing insights (source)
• Links in HogQL don't look like links - e.g. no underline
• SQL Editor Visualizations formatting (conditional formatting takes up way too much space)
• Clicking table links from sources is broken
• Paginating visualization table (it crashes if you go from data grid)
• Anything that comes up from #proj-metabase-to-posthog

Follow on quest:

• Improving empty states for guiding
• Insight variables Insight Variables TODO meta#255
• Data modeling - Need to add scheduling
• plan for unrooting sql editor from insights

Priorities

Batch Exports folks are ramping back up on data warehouse

• release materialization. (Scheduling + UI) @phixMe
• pipeline error states @phixMe
• release sidebar folders @EDsCODE
• release insight variables (date variables) @EDsCODE or @Gilbert09
• Continuing on source pipeline issues. There's still gradual memory leaks. It's better than before but something is still wrong and preventing our very large user syncs from working @Gilbert09 + rest of team is following this
• Editor errors @rossgray or @tomasfarias