Blog

3 min read

AI Insights: the analyst that reads your reports before you do

Scheduled AI observations across traffic, revenue, devices, custom events, and running A/B tests, ranked by priority and shown on the dashboard, the reports, and one feed.

Nobody reads every report every morning. That's not a discipline problem, it's a math problem — there are twenty-four of them. AI Insights reads them for you and leaves notes.

What a run does

On a schedule, Badgerlytics compares your recent numbers to the prior period and writes short observations where something moved. Each one includes:

  • What changed — with the evaluation window it looked at
  • A plausible reason — not a certainty, a hypothesis worth checking
  • A next step — often "test this" or "look here"
  • A priority — high, medium, or low, so you can triage in thirty seconds

Report types covered: Website Performance, Revenue (or MRR and Usage Revenue on SaaS properties), Device Performance, Custom Events, and A/B Tests. You can switch any of them off in settings.

Where the notes show up

SurfaceWhat you see
AI Insights pageThe full feed grouped by priority, filterable by report type, plus a Recently Dismissed tab
DashboardA short highlights list on the Overview tab; test-specific notes on the Active Tests tab
Report pagesA collapsible strip at the top of any report that participates, so the note sits directly above the numbers it's about

How often it runs

Around 7am Central on run days:

PlanCadence
Starter3× per week (Mon, Wed, Fri)
Pro, Business5× per week (Mon–Fri)
Ultimate7× per week

Standard report insights compare the latest 30 days to the prior period. New properties use shorter windows until enough history builds up. A/B test insights cover running tests only, from the start of the current iteration to now.

Dismissals are a team sport

Acted on something? Dismiss it. Dismissals are team-wide for the property — it disappears for everyone, stops appearing on the dashboard and report strips, and stays hidden for 7 days. During that window scheduled runs skip regenerating the same (or a very similar) observation, so the feed doesn't nag. The Recently Dismissed tab shows who hid what and when it can return, with a Restore button.

There's no dismiss-all on purpose. Clearing one at a time keeps the shared feed predictable.

Getting more out of it

  • Write a property description. "Mid-market accounting SaaS, sells to bookkeepers" gives the model business context. It also helps the Chat Analyst.
  • Turn off report types you don't care about. Fewer, sharper notes.
  • Treat recommendations as hypotheses. Validate the big ones with an A/B test before shipping.

Plans and tokens

Included on all paid plans. Runs draw from your organization's monthly AI token allowance — usage per run is high because several reports get analyzed, but runs are paced by your plan and shouldn't exhaust a normal allotment on their own. Token details: Billing & Usage → Token usage.

Full docs: AI Insights. Insights are the scheduled half of our AI features; the on-demand half is the AI Chat Analyst, and the copy-paste half is Team Summaries.

Quick answers

What are AI Insights in Badgerlytics?
Scheduled, automatic observations about a property's analytics. Each run compares recent performance to the prior period and writes short, prioritized notes with plausible reasons and next steps across report types like Website Performance, Revenue or MRR, Device Performance, Custom Events, and A/B Tests.
How often do AI Insights run?
Around 7am Central on run days — 3 times a week on Starter (Mon/Wed/Fri), 5 times a week on Pro and Business (Mon–Fri), and every day on Ultimate.
Where do AI Insights appear?
On the AI Insights page (the full feed grouped by high, medium, and low priority), as highlights on the dashboard Overview and Active Tests tabs, and as a collapsible strip at the top of matching report pages.

Ready to see what your data has been hiding?

Join the teams using Badgerlytics to grow revenue, run smarter experiments, and finally understand their funnels.

Cancel anytime.