Your dashboard shows what happened. This tells you why.
Most reporting stops at the chart. The AI insight layer reads your unified data every week, compares it against what's normal, and writes the part a human analyst would spend an hour on — what changed, why it matters, and what to do — ranked by dollar impact.
Revenue +12.4%, driven by Shopify. But SKU-D is selling well and barely profiting — margin eaten by ad spend and fulfillment. TikTok ROAS fell to 1.2×, the only channel below break-even this week.
Dashboards tell you what. Nobody tells you why.
Every platform reports its own version of the truth, and reading five tabs to find the one number that matters takes longer than it should — every single week.
Five steps, run automatically every week.
Read
Pulls the week's unified numbers from your BigQuery warehouse.
Compare
Checks each metric against its trailing baseline and normal range.
Rank
Scores every anomaly by estimated dollar impact, not just size of change.
Write
Turns the top findings into plain-English commentary and a recommended action.
Deliver
Sent to dashboard, email and Slack — before your Monday standup.
The interpretation is where the value actually is.
A clean dashboard is table stakes. These are decisions the AI layer surfaced that a static report would have buried.
£77 → £46. The insight layer flagged a channel converting far cheaper than the rest — budget was reallocated the same week.
AI review analysis traced a brand's 1-star reviews to its subscription model, not the product — a pattern nobody had connected by hand.
The layer flagged a metric moving outside its normal range, which traced back to a silent double-counting bug before it reached a report.
Built on top of the warehouse, not instead of it.
The AI insight layer reads from a unified BigQuery model — it's the interpretation step, not a replacement for clean data underneath it.
Data warehouse & pipelines
The foundation this layer reads from — one clean model in BigQuery, blended and deduplicated across every channel.
Automated dashboards
Where the AI commentary lives day to day, alongside the charts and KPIs your team already checks.
What people ask about the AI layer specifically.
Does the AI just summarize numbers, or actually analyze them?
It analyzes. It compares this week against trailing baselines, flags what moved outside a normal range, and ranks each finding by estimated dollar impact — the same first pass a human analyst would do, run automatically every week.
What happens if the AI gets something wrong?
It only writes from the numbers in your warehouse, so it can't invent facts — but interpretation can be off, which is why every insight links back to the underlying data and I review the model's output as part of the managed retainer, not just the pipeline.
Does this replace having an analyst?
It replaces the hours spent building the report by hand each week, not judgment. Think of it as a first draft that's already found the anomalies and done the ranking — you or I spend the saved time on the decision, not the data pull.
Can I see what the output actually looks like?
Yes — the sample report page shows a live example of the weekly AI summary, the SKU-level flags, and the ranked alert feed, using illustrative data.
See it running on real data.
15 minutes, no pitch. Bring the number you least trust and I'll tell you what's likely behind it.
Book a 15-minute call