MARKETING
Auto-end losing Vercel A/B variants and log learnings to Notion
On a daily schedule, evaluates each running Vercel Edge Config experiment for statistical significance, removes any variant that is clearly losing, reallocates its traffic.
How it runs
The automated pipeline, trigger to output.
- TriggerDaily schedule fires the experiment review
- ActionRead active experiments and splits from Vercel Edge ConfigVercel
- LogicEvaluate each variant and flag confident losersOpenAI
- ActionRewrite Edge Config to drop loser and reallocate trafficVercel
- ActionAppend learnings entry to Notion experiments databaseNotion
- OutputPost ended-variant summary to marketing Slack channelSlack
What it does
Runs the full end-of-experiment decision once a day. For every active A/B test defined in Vercel Edge Config, it pulls the variant conversion numbers, checks whether a variant has reached a confident losing verdict, and if so retires that variant by rewriting the Edge Config split. The freed traffic is reallocated to the remaining variants, and a permanent learnings record lands in Notion so the result is never lost in a dashboard.
When to use it
Use it when you run several concurrent edge experiments and don't want to babysit each one. It removes the human delay between "the test is clearly over" and "the loser is actually turned off," which is where wasted impressions accumulate.
How it works
- 1A daily schedule fires the run.
- 2Read all active experiment definitions and current splits from Vercel Edge Config.
- 3For each experiment, evaluate variant performance and decide whether a variant is a confident loser.
- 4If a loser is found, rewrite the Edge Config to drop it and reallocate its traffic share.
- 5Append a learnings entry (hypothesis, result, traffic, verdict) to the Notion experiments database.
- 6Post a summary of ended variants to the marketing Slack channel.
Set it up
What you configure once, before turning it on.
- 1Connect VercelDeploys, runtime logs, analytics.
- 2Connect NotionPages, databases, comments.
- 3Connect SlackChannels, DMs, threads, mentions.
- 4Connect OpenAIModels, embeddings, files.
- 5Set each agent's modelWe leave models unset so you pick the tier — fast + cheap, or top-quality.
- 6Tune it to your dataEdit the prompts, filters, and field mappings so it matches how your team works.
- 7Test, then turn it onRun once against a sample, confirm the output, then enable the trigger.
More Marketing workflows
UTM Link Linter: Block Malformed Tracking URLs Before Launch
Validates every campaign tracking link queued in an Airtable launch table against your UTM naming rules.
Spin up a rush localization handoff from a Slack chat request
A marketer asks the CEO agent in chat to rush-translate a piece of approved copy into specific languages; the agent files the source in Google Drive.
Slack UTM Check: Paste a Link, Get an Instant Verdict
Lets marketers drop a tracking URL into a Slack channel and get an immediate threaded reply that says whether the UTMs are valid, what is wrong, and a corrected suggestion.
UTM Audit for Campaign Spreadsheets in Google Drive
Watches a Google Drive folder for new or updated campaign link spreadsheets, lints every URL inside.
Real-Time UTM Validator API for Link-Building Forms
Exposes a webhook that validates a tracking URL on the fly and returns a structured pass/fail JSON response.
Agentic UTM Taxonomy Steward: Detect, Explain, and Propose Fixes
An agent reviews the full backlog of campaign links in Airtable, reasons about taxonomy drift beyond rigid rules, drafts standardized replacements for malformed UTMs.
Run it inside a business
This workflow drops into a full company template. Import the org, and this is one of the playbooks its agents run.
Run this workflow in your colony.
14-day trial. No DevOps. No Sales call. Provisioned in under a minute.