Skip to content

article

AI-Powered Insights

Hand a metrics summary to the suite AI and get a structured read: a summary, ranked insights with severity, and recommendations.

What AI insights do

AI insights let you hand a metrics or data summary to the StretchSuite AI layer and get back a structured analyst's read rather than a wall of text. The response includes an overall summary, a list of insights (each with a title, a detail, a severity, and the metric it relates to), and a set of concrete recommendations.

Crucially, AI insights are advisory only. The AI reads and interprets your numbers — it never changes your data, and it never takes action on your behalf. You stay in control of every decision.

What you get back

A successful insights run returns:

  • summary — a concise plain-language read of the data.
  • insights[] — each with a title, a detail, a severity (low, medium, or high), and the metric it concerns.
  • recommendations[] — concrete next steps you might consider.

Step-by-step: generate insights

  1. Assemble the data you want interpreted — typically a metrics summary from your overview, a KPI set, or an Explore result.
  2. Open the AI insights action in StretchAnalytics.
  3. Submit your metrics summary as the input.
  4. Review the returned summary, scan the insights ranked by severity, and read the recommendations.
  5. Decide what, if anything, to act on — the AI suggests, you choose.

Example

An operator pastes a summary: page views 8,000; CTA clicks 1,200; form submissions 260; open support tickets up 40% week-over-week. The AI returns a summary noting healthy top-of-funnel traffic but a weak CTA-to-submission path, a high-severity insight flagging the support-ticket spike as a possible product issue, a medium-severity insight on CTA conversion, and recommendations to review recent releases and A/B test the CTA copy. The operator uses it as a second opinion, then verifies each point in Explore before acting.

Governance: what the AI will and won't do

StretchAnalytics is deliberately governed. Today the AI is allowed to produce trend summaries, anomaly hints, KPI rankings, cohort summaries, and data-quality flags. More advanced capabilities — forecasting, next-best-action, automated report narratives, and budget or staffing recommendations — are treated as dry-run only and are on the roadmap. The AI never mutates records autonomously, and any persisted AI output is auditable.

Availability

AI insights are flag-gated. They are available when your workspace has the analytics AI capability enabled and the underlying model service is ready. If the feature isn't enabled for your workspace, the app safely reports that insights are unavailable rather than erroring — the rest of StretchAnalytics works exactly the same.

Tips

  • Give the AI a focused summary — a handful of clearly labeled metrics produces sharper insights than a data dump.
  • Treat severity as a triage aid: start with high, then work down.
  • Always verify an AI insight against the raw data in Explore before making a decision.
  • Feed it period-over-period numbers (from Compare) so it can reason about direction, not just level.

FAQ

Does the AI change my data or send anything automatically? No. It reads and interprets only. It never mutates records or takes action.

Why did I get a response saying insights are disabled? The capability is flag-gated per workspace and requires the model service to be ready. Contact your administrator to enable it.

Can I trust the recommendations blindly? No — treat them as a well-informed second opinion. Verify against your own data before acting.

Was this helpful?

Help us improve this article

Use these controls to share whether this answer solved the issue. Feedback helps prioritize updates to StretchSuite Support.