01
Evidence first
An insight links to the records it used and the time they were last updated.
AI insights that cite their numbers
The auditor looks for drops, anomalies, high frequency, missing new creatives and misallocated spend, and explains each finding with figures you can open and check.
Floating cards and figures are examples on sample data, not a customer account.
1The proof
Every number comes with its receipt.
Sample data, fictional brand. This is how AI insights shows a figure: the value, where it came from, the formula behind it and when it was last updated.
CPM jump+38% vs 7 days agoSource and formula3 ad sets, MetaUpdated25 min ago
Creative fatigueFrequency 4.1Source and formulaSpring UGC setUpdatedsee source
No new creatives21 daysSource and formulaProspecting campaignUpdatedsee source
Rule: every number shows its source, formula and update time. A missing source reads "not measured", never 0.
2Part of Measure
AI insights: what it does.
One of four areas of the product, sharing the same client record, roles and log.
02
Outside text is data
Comments, documents and web pages are read as data and never followed as instructions.
03
Suggests, never acts alone
A suggested change goes through preview and approval like any other change.
Works with
3 platforms, read-only first
Connections start read-only. Writing is switched on per account, only after you grant the rights.
At a glance
Stated up front
- things it does
- 3
- steps start to finish
- 4
- platform connections
- 3
- limits stated
- 2
3The data path
How the data moves. Source to screen.
Lines draw as you scroll. Platforms on the left are read-only first, AI insights sits on the one client record, and each output keeps its receipt.
Meta AdsBuiltGoogle AdsBuiltShopifyBuiltROASfinityAI insights
Evidence first
Outside text is data
Suggests, never acts alone
Sources are read first, the numbers are stored once per client, and every output shows its source, formula and update time.
How the work moves in Measure
The same four moves apply to every module in this area.
- 1 ReadConnected sourcesOrders, spend and results come from the store and ad platforms, read-only first.
- 2 ComputeFixed formulasMER, CAC and the rest are defined once, in the open, the same for every client.
- 3 ShowSource on every numberEach figure carries its source, formula and the time it was updated.
- 4 FlagNot measured, never zeroWhen data is missing the screen says so, instead of showing a made-up value.
4The routine
How it works, step by step.
From first connection to the number a client sees.
- 1
Collect history
Daily metric snapshots, rules and tasks form the evidence.
Step 1Collect historyexample - 2
Check
Scheduled checks compare against history with a minimum sample.
Step 2Checkexample - 3
Explain
Each finding states what changed, by how much and where to look.
Step 3Explainexample - 4
Hand off
Open a task or a proposed change with the evidence attached.
Step 4Hand offexample
5In use
Built for businesses like these. Example scenes.
Photos are stock images that show the kind of business this is built for. They are examples, not customers.
Stock photos and example labels, not customers.
6In practice
Common mistakes, and a short checklist.
General practice for this kind of work, not a claim about any one tool.
Treating an AI summary as a finding
A summary is a starting point written from numbers that may be incomplete. Open the source rows and check the figure before it reaches a client.
Asking a vague question
A question like 'what is going on?' gets a confident answer that may not fit the data. Name the metric, the period, the segment and the decision the answer will feed.
Not recording which insight led to a change
Without a record, nobody can tell later whether the insight helped. Log the insight, the change it caused and the result read back after the change.
A short checklist
Your ticks stay in this browser only.
7Honest limits
What it will not claim, and what it works with.
What it will not claim
- An explanation is a hypothesis backed by data, not proof of cause.
- Findings need a minimum sample, so very new accounts show fewer insights.
Works with
Connections are read-only first and go live only after the platform access is approved.
Built: the connector exists in the product and goes live per account once access is granted. Planned: not built yet.
Used instead of
Teams often use tools such as Madgicx for this. Our comparison pages are drafts being verified.
8Related modules
Same room, next door.
Modules that share this client record, these roles and this log.
MeasureSignalsAlerts for inactive accounts, rejected ads, CPM and CPA jumps, weak spend, broken URLs and stale connectors.Read more→MeasureBudget allocationScenarios for moving budget between campaigns, with limits and a record of expected versus actual.Read more→LaunchChange logEvery change with before and after, author, reason, time, API response and what happened next.Read more→
9Guides
Learn more. Short guides.
Short guides on the numbers and routines behind this module.
GuideA safety checklist before an AI assistant touches ad accountsAn assistant that can read your accounts is a convenience. One that can write to them is a decision. This checklist separates the two.6 min read→GuideMER vs ROAS: which number an agency should reportPlatform ROAS tells you how an ad platform scores itself. MER tells you how the business did. A client review needs both, clearly labelled.6 min read→
10FAQ
Questions about ai insights.
Can insights change my campaigns?
No. An insight can propose a change, and the change then needs a preview, an approval, a single send, a read-back and a log entry.
What if the data is missing?
The finding says what is missing and does not guess. A disconnected source is shown as not measured.
Which model writes the explanations?
Models are swappable behind an adapter, so the product is not tied to one provider.
Run the next client review from one screen.
We open one agency at a time, with your own accounts connected read-only first.