Measure
Brief
Diagnose a drop in engagement
Paste three months of campaign metrics and what changed; get a diagnosis that separates evidence from inference, and three experiments.
The prompt
Help me work out why engagement dropped and what to do about it.
Context: metrics by campaign for the last [three months]: [paste name, date, audience, size, open rate, click rate]. What changed in that period: [new sender, new template, more sends, org news, seasonal factors]. What didn’t change: [list].
Audience: me, the comms lead, and eventually my manager.
Tone: analytical and honest; separate what the data shows from what you’re inferring.
Must include:
– Where the drop actually is (which campaigns, audiences, and metric) versus where it only looks like a drop
– Candidate causes ranked by how well the data supports them
– What you’d need to know to confirm the top two causes
– Three experiments for next month, each with the metric that would prove it worked
Constraints: don’t blame “email fatigue” without showing evidence for it; flag small-sample campaigns.
Output: a short diagnosis, then the experiments as a table.
Works in
Any AI toolHow to use
"What didn’t change" is as useful as "what changed" — it rules causes out.
Related Resources
page
Ask Insights Assistant "what are some interesting analytics changes in the recent data?" and it surfaces the shifts, outliers, and trends across your sends — no export needed.
The prompt
Help me work out why engagement dropped and what to do about it.
Context: metrics by campaign for the last [three months]: [paste name, date, audience, size, open rate, click rate]. What changed in that period: [new sender, new template, more sends, org news, seasonal factors]. What didn’t change: [list].
Audience: me, the comms lead, and eventually my manager.
Tone: analytical and honest; separate what the data shows from what you’re inferring.
Must include:
– Where the drop actually is (which campaigns, audiences, and metric) versus where it only looks like a drop
– Candidate causes ranked by how well the data supports them
– What you’d need to know to confirm the top two causes
– Three experiments for next month, each with the metric that would prove it worked
Constraints: don’t blame “email fatigue” without showing evidence for it; flag small-sample campaigns.
Output: a short diagnosis, then the experiments as a table.
