Product metrics digest
By Toolspoke
Reads Amplitude each Monday and posts last week's active users, new users, retention and funnel to Slack, against the week before.
Needs a Toolspoke account. You choose which project it installs into.
- Amplitude + Slack
- Required integrations
- Runs on a schedule
- Mondays at 9am by default
- Post one message
- What its last step does
- Some actions can delete
- 1 of 29 actions in its tools
On this page
The short version
- What it is
- A ready-made Toolspoke agent that posts your product's weekly Amplitude numbers to Slack, each against the week before.
- What it reads
- Active and new users, weekly retention, up to five key events and one funnel, for last week and the week before.
- What it finds
- The numbers that moved by enough to count, and the Amplitude annotations dated near them.
- What you get
- One Slack message every Monday morning, readable in a minute.
- How often
- Every Monday once you switch it on.
- What it never does
- Look up an individual user, or say what caused a change.
- What you need
- A Toolspoke account, an Amplitude project's API key and secret key, and a Slack workspace.
What Product metrics digest will never do
Product metrics digest is told never to look up an individual user, never to say what caused a change, and never to count an event you did not choose. It reads the numbers and reports them.
Product metrics digest will never
- Call get_user_activity or get_user_search.
- Say a change was caused by a release, a campaign or an annotation.
- Count an event the person did not choose, or swap a chosen event for one that looks similar.
- Edit or delete a Slack message, or post more than once a run.
Why it will not look at a person
Amplitude can return one user's profile, location and every event they fired. A weekly digest needs none of that, so the agent is told never to call those actions, and reports people as counts.
Why it will not claim a cause
An annotation says something happened on a date. It does not say the number moved because of it. The agent places annotations beside the numbers and writes may be related.
Why it will not call noise a change
Small numbers swing. Below 100 people in the smaller week, it shows the number and says too few to call.
These are written into Product metrics digest's instructions. They guide the agent, but they are not a permission. What it is allowed to touch is set by the access you grant.
Product metrics digest reads Amplitude and posts the numbers. What they mean, and what to do about them, stays with your team.
What Product metrics digest does
Product metrics digest is an AI agent that reads one Amplitude project every Monday and posts the numbers a product team checks first: active users, new users, retention, your key events and your funnel. Every number sits beside the week before, so you see the change, not just the level.
Most teams have an Amplitude dashboard and a habit of meaning to look at it. The weekly number gets read when someone remembers, and a fall gets noticed a month late.
A good weekly read takes care. The two weeks have to be the same length, the funnel has to give both weeks the same time to finish, and a ten percent swing on forty people is not news.
Product metrics digest does that read for you each Monday. It uses only the events you chose, marks the changes big enough to count, and names any annotation your team left in Amplitude near the dates. It never says a release caused a number to move.
In practice, every Monday you have:
- Weekly active and new users, against the week before
- Second-week retention for the latest complete cohort
- Up to five key events and one funnel, step by step
- The annotations dated in those two weeks, next to the numbers
How it works
Product metrics digest works in four steps each week: it counts the users, reads retention and events, runs the funnel and posts one message. All of it comes from the agent's written instructions, which you can read and edit after you install it.
Step 1: Count the users
Active and new, for last week and the week before.
Step 2: Read events and retention
Your chosen events, and how many new users came back.
Step 3: Run the funnel
Your steps in order, with the same seven days to finish.
Step 4: Post one message
Each number with the week before, and the annotations nearby.
It uses your events, not guesses
At install you choose up to five events and a funnel of two to five steps. The agent checks each one against list_event_types, because Amplitude answers a misspelled event with an empty chart rather than an error.
It compares like with like
Both weeks run Monday to Sunday, and the funnel gives both the same seven-day window to finish. A funnel with a 30 day window would make last week look worse simply because it has had less time.
It needs enough people
A change counts at 10% or more, or 3 points for a rate, and only when the smaller week has 100 people. Below that it shows the number and says too few to call.
It reports, it does not explain
Annotations are listed beside the numbers they might touch, with the words may be related. Why a number moved is for your team to judge. The limits are listed under what it will never do.
What a weekly digest looks like
A run of Product metrics digest posts one message to Slack. It leads with the changes that count, then the rest of the numbers, then what was too small to call.
Northwind
Channels
- general
- product
- growth
- engineering
product
20 to 26 October against 13 to 19 October. Weekly active users 18,420, up 6%.
Changes that count (2)
- Report exported: 1,140 people, down from 1,330 (down 14%).
- Funnel, created project to invited teammate: 29%, down from 34% (down 5 points). 2,050 people reached the step.
The rest of the numbers
- Active users: 18,420, up from 17,380. New users: 2,310, up from 2,260.
- Week 2 retention: 41% of the 6 October cohort came back, against 43% for the cohort before.
- Annotation, 22 October: "Export dialog redesign shipped". May be related to the fall in exports.
Too few to call
Upgraded plan: 64 people, against 51. Not counted as a change.
- Changes that count
Numbers that moved by 10% or more, or 3 points for a rate, with at least 100 people in the smaller week.
- The rest of the numbers
Users, retention, events and funnel steps that held steady, and any Amplitude annotation dated in the two weeks, placed beside the number it might touch.
- Too few to call
Numbers with too few people behind them to call a change. They are shown so you know they were read.
When nothing moved
If every number stayed within the thresholds, the digest still posts, says nothing moved and gives the numbers. A weekly report that goes quiet looks the same as one that broke.
If a chosen event has no data in either week, the digest says which and leaves it out, because a renamed event looks like a fall to zero. If Amplitude cannot be read, it says so once and stops.
Who uses Product metrics digest
Product metrics digest suits a team that tracks its product in Amplitude and wants the weekly numbers in Slack without opening a dashboard. The examples use made-up names.
A product manager in a Monday planning meeting
Priya Nair opens the digest before planning. The fall in exports, with the redesign annotation beside it, goes straight on the agenda.
A founder who wants one number a week
Tomás Rivera reads the first line and nothing else, unless something is marked.
A growth lead watching activation
Amara Okafor watches the invite step of the funnel. The digest marks it when it slips.
An engineering lead after a release
Jonas Berg checks the Monday after a big release that nothing fell over. The annotations show up next to the numbers.
What Product metrics digest connects to
Product metrics digest connects to two tools: it reads Amplitude and posts one message to Slack. You connect each one yourself, and Toolspoke stores the connection encrypted, so the agent never sees the credential.
Amplitude
RequiredQuery Amplitude event segmentation, funnels, retention, active users and annotations.
Authentication: API key and secret
Slack
RequiredSearch conversations and send messages to your workspace.
Authentication: Bot token
| Tool | What Product metrics digest does | What it cannot do |
|---|---|---|
| Amplitude | Counts users, events and funnel steps, reads retention and lists annotations (active_user_counts, event_segmentation, retention_analysis, funnel_analysis, list_annotations, list_event_types). | Change anything in Amplitude: every action the connector has is a read. It can look up a single user's profile and events, and the agent is told never to. |
| Slack | Posts one message to the channel you chose (send_message). | Delete messages: the connector has no delete action. It can edit its own messages and schedule messages, and the agent is told to do neither. |
That the agent cannot change your Amplitude project is a property of the connector, which only reads. That it never looks up a single user, and posts one message in Slack, is an instruction to the agent, not a lock. If you want a lock, a project can switch individual actions off. The access control guide explains how.
The Amplitude key and secret are scoped to one project, so the connection reaches only the project you connected. The digest reports people as counts and never puts a user ID, device or location in Slack.
When Product metrics digest runs
Product metrics digest runs every Monday at 9 AM, but only after you switch its schedule on. It arrives switched off, so nothing runs and nothing is charged until you decide.
- Each run compares last Monday to Sunday with the week before.
- It posts one message to the Slack channel you chose.
- It posts even when nothing moved, and says so.
- If an event has no data, it says which and leaves it out.
Each run uses model credit from your workspace balance, so it makes one run a week. The queries run in Amplitude and count against your Amplitude plan's API limits in the usual way. The Toolspoke tool calls themselves are free.
Your copy is an ordinary agent in your project. You can pause its schedule there, run it once on demand, change the day and edit its instructions. The agent's page lists recent runs and their cost.
How to set up Product metrics digest
To get your weekly numbers, add Product metrics digest to a Toolspoke project, connect Amplitude and Slack, choose your events and funnel and switch on the schedule. It takes a few minutes, and nothing runs until the last step.
Step 1: Create a Toolspoke account or sign in
Choose Use this agent. If you are new, you will be asked to sign up first, and you come straight back to this agent afterwards.
Step 2: Add it to a project
Pick the project it should live in. You get your own editable copy, and you can see the two tools it brings before anything connects.
Step 3: Connect Amplitude
Paste your project's API key and secret key from Amplitude's project settings. Use the EU host if your account is in Amplitude's EU region.
Step 4: Choose your events and funnel
List up to five key events and the funnel's steps in order, spelled exactly as Amplitude shows them.
Step 5: Connect Slack
Sign in to Slack and invite the bot to the channel the digest should post in.
Step 6: Switch on the schedule
The schedule arrives switched off. Run it once, check the numbers against Amplitude, then switch it on.
Make Product metrics digest fit your product
Product metrics digest's instructions are plain English and you can edit them after you install it. Installing copies the agent, so your edits stay with your copy. Open the agent in Toolspoke and tell the builder what to change, and it rewrites the instructions.
Paste any sentence below into the builder chat, or say the same thing in your own words.
Also split active users by the user property plan, and mark a plan whose users fell by 10% or more.
Compare last week with the average of the four weeks before, not just the week before.
Count a change at 40 people instead of 100, and say the numbers are small.
Also count the event purchase_completed, with totals as well as uniques.
Run every Friday at 4 PM, covering Monday to Thursday.
This changes when the agent wakes up, so the builder edits its schedule as well as the instructions.
None of these edits needs more access or touches the Never rules. They change what Product metrics digest is told, not what the connections allow.
How it compares to other options
Product metrics digest beats the alternatives when you want the same handful of numbers read carefully every week, and loses to them when you need to explore or explain.
| Compared on | Product metrics digest | An Amplitude dashboard | A scheduled chart email | An analyst's weekly note |
|---|---|---|---|---|
| Arrives in Slack every week | Yes | If someone opens it | By email | Yes |
| Compares with the week before | Every number | If you build it | Chart by chart | Yes |
| Ignores changes too small to trust | Yes, by threshold | No | No | Yes |
| Places annotations beside the numbers | Yes | On the chart | No | If they look |
| Explains why a number moved | No, it reports | No | No | Yes, with judgement |
| Cost | A run a week | Included | Included | Part of a salary |
An analyst is the better choice when you need to know why a number moved, with segmentation, follow-up questions and judgement. Product metrics digest is for the team that wants the same numbers read the same way every Monday.
An Amplitude dashboard is where the numbers live, and the agent reads the same data. The difference is that it reads it for you, compares it with the week before and tells you which changes are big enough to matter.
Frequently asked questions
Can Product metrics digest change anything in Amplitude?
No. Every action the Amplitude connector has is a read. It cannot edit a chart, a cohort or an event.
Does it put user data in Slack?
No. It reports people as counts. Amplitude can return a single user's profile, and the agent is told never to ask for one.
How does it know which events matter?
You choose them at install: up to five key events and a funnel. It checks the spelling against Amplitude and never swaps in an event that looks similar.
Will it tell me why a number changed?
No. It marks what moved and lists any annotation your team left in Amplitude near those dates, with the words may be related. The why is your team's call.
Does it work with an EU Amplitude account?
Yes. Set the region host to Amplitude's EU address when you connect. A key from one region is unknown in the other.
Does it work with PostHog or Mixpanel?
Not as shipped, because it reads Amplitude. Product funnel detective reads PostHog, and you can ask the agent builder for a version on another toolkit.
Is Product metrics digest free?
It has no price of its own, and every Toolspoke plan includes agents. Tool calls cost nothing, but each run draws on your usage balance. Pricing has the details.
Your product's week, in one message
Connect Amplitude, choose your events and switch on the schedule. Your first digest lands on Monday.