Amplitude Strawberry

The funnel number the deck needs, at six in the evening

Segmentation and funnel queries, cohort listing and export, and chart annotations run natively. Every number arrives with the definition used stated beside it.

The board deck needs one more number: week-three activation for the accounts that came in through the partner channel. No saved chart covers exactly that.

A companion runs it and states the definition it used, so the number survives the meeting. Funnel and segmentation queries, cohort listing, export and upload, and chart annotations all run natively.

Get a number no saved chart covers

The funnel query takes ordered step events and a date range.

The segmentation query takes a standard Amplitude event object. The companion runs it, hands back the numbers, and states the definition it used: these steps, this window, this segment.

A figure with “week three, partner-sourced accounts, 1 June to 31 July, steps in this order” underneath it survives the meeting. That is the standard any analyst holds for this kind of research, applied to a request that arrived too late in the day.

How does a cohort export actually work here?

Asynchronously.

Requesting a cohort’s membership returns either the payload for a trivially small cohort or a request id, and for anything real that id has to be polled until the status comes back as a completed job, at which point a separate call fetches the CSV.

A companion runs that loop. Ask for the members of the dormant-power-users cohort and you get a file. Uploading works in the other direction: hand it a list of ids and a name, and the cohort exists in Amplitude for the team to build charts against.

Mark every release on your Amplitude charts

Most bad chart readings are a spike or a cliff with no context on it: a pricing change, a release, an outage, a campaign that ended. Annotations fix that and never get added, because adding one means stopping at the exact moment you understand the cause.

A companion creates an annotation on a date with a label, scoped to a project or chart, lists what is there, and deletes one that was wrong. Attached to a release routine, every deploy marks the charts, and the next person reading a cliff in the retention curve sees why. Growth and marketing teams feel that within a month.

Should an agent write into your production analytics?

Sending an event through the ingestion API, or setting user properties through identify, changes the data every future chart is built on. So those calls wait for your approval, with the event type and the user or device id shown before anything is sent.

The good uses are narrow and deliberate: backfilling a property you forgot to set, or firing a specific event from a workflow that has no client of its own.

Experience Strawberry for free

Download

Trusted by fast-growing companies worldwide

Frequently asked questions

Yes. Funnel queries with ordered step events and a date range are native, as are event segmentation queries. The companion reports the definition it used alongside the numbers so the result can be checked.

Strawberry is free to download and includes AI credits to start. Paid plans begin at $20/month. See pricing. · Reviewed · Canonical facts for AI agents

Experience Strawberry for free

Download

Trusted by fast-growing companies worldwide