driftbeacon is a growth analyst built for your company. It checks all your business data every day and digs into any change before anyone asks. Every number traces back to source, and every issue is followed through to a recheck. Your growth calls stop depending on who happens to be good with AI, or who has time to look.
Piecing together one customer’s story used to mean exporting CSVs from five dashboards and losing an afternoon to VLOOKUPs. Now everything lands in one database, joined by user: the ad they clicked, the day they started a trial, the day they paid. One line, start to finish.
Most events ship with one line of code. Key events get a spec, and anything that doesn’t match is flagged the moment it arrives.
Ads, the app and the App Store, connected for the first time.
track() call and it’s live. No tables to set up, no waiting on the data team.
Every day it goes through the numbers on its own. When a metric jumps or a data source goes quiet, it writes it up and pings the owner in Slack.
It knows how your team defines an active user and which table is missing a region. Every number in a conclusion traces back to the original query.
Evidence, ruled-out guesses, who’s on it and when to recheck all live in one file. Swap the person or swap the AI, and they just keep going.
Connect Claude Code, Codex or Cursor over MCP. Get halfway in one today, finish in another tomorrow. Same data, same file.
# Connect driftbeacon to Claude Code claude mcp add --transport http driftbeacon \ https://growth.your-company.com/mcp > Anything I should look at this week? ● 3 new issues, 1 due for recheck
Anything that changes a number needs someone to say yes. The AI’s notes keep every version, so you can always roll back.
Everyone on your team has an AI assistant now.
Your company should have one just for growth.
driftbeacon is a growth analyst that does its own legwork. Every day it goes over your data, and when something worth a look changes, it opens an issue and hands it to an owner.
From then on, who checked what, which guesses got ruled out and who signed off on what all goes on that track, until a recheck confirms it. Three weeks later, when your CEO asks, you can walk them through it.
Before any conclusion is saved, it has to pass one test. Every number in it must show up in the query results. If it doesn’t, it’s out.
Trial starts for Sep 8–14 were 2,176, up 38% from 1,577 the week before. The rise is concentrated in Japan (902).
| Region | Prior week | This week | Change |
|---|---|---|---|
| Japan | 410 | 902 | +120.0% |
| Other regions | 1,167 | 1,274 | +9.2% |
| Total | 1,577 | 2,176 | +38.0% |
Trial starts this week were 2,716, about 70% above the 1,577 of the week before.
| Check | In conclusion | In query result |
|---|---|---|
| This week total | 2,716 | 2,176 |
| Prior week total | 1,577 | 1,577 |
Chat-based BI answers one question at a time. driftbeacon goes looking for the questions, then sticks with them: spotted, investigated, decided and rechecked, all in one file. Asking in plain English is just one way in.
No. They’re great at behaviour, funnels and retention. driftbeacon pulls their data in next to orders, subscriptions and ads, and keeps working one business question across all of it.
No. The AI researches, prepares options and proposes changes for approval. People with the right access carry them out, and changes to driftbeacon’s own rules also need approval.
It can be deployed in your own cloud account, so the data stays in your environment. AI reaches it through interfaces you authorize, with read-only queries.
Yes. Pick one well-defined question, connect only what it needs, and expand once it’s working.
No need to buy the whole thing up front. We’ll pick a real question together, connect the data it needs and run one full recheck cycle. Then you decide whether to keep going.
Access requests are not open yet. The fields below preview the information we’ll ask for.