A layer is something you already have, like your weekly report. Tap one to see what you'd send, what Seurat does with it, and what you'd see. Heartland Markets, a fictional chain of 42 stores, shows each one.
Your PDI weekly sales report, the one your back office already makes, with sales, gallons and customers by store, so there's nothing new to build. The first time, include the 2 weeks before it, so Seurat can tell on Day 1 which gaps have held for 3 weeks. Delete any line you'd rather keep, such as cash over/short or margins.
| Store | Sales vs LY | Normal range | Flag |
|---|---|---|---|
| 13 · River Rd | −1.9% | −3.7 to +6.3% | · |
| 14 · Collins Rd | −2.1% | −1.2 to +3.8% | Below range |
| 23 · Hwy 6 E | −2.6% | −10.7 to −2.7% | Better than it looks |
Gives every store its own normal range, calibrated to its sales level and size, and adjusted for what changed nearby in the public record, social media and reviews. Any store outside its range gets a note, and a flag once it has held for 3 weeks.
Heartland, Day 1: Store 14 is down only 2.1%, less than Store 23, but it's Store 14 that's below its own normal range, for the 3rd week in a row.
Open it in the sampleOne export, once: your last 2 years of weekly sales by store and category, in units and dollars, with each store's format and town or zip code. Any spreadsheet or CSV works.
Drop it in. Rather not share dollars? Send an index instead, and gaps come back in percent.
Builds a twin for every store: a blend of your other stores that matched it week by week for 2 years. The twin shows what the store should be selling. Because the twin goes up and down with the season, the weather and the market, those swings cancel out, and what's left is the store's own gap. Each gap is then split into fuel, pump to store and basket, down to the category.
Heartland, Day 2: Store 14 has been 4.0% below its twin since the week of Aug 3, about $3,550 a week, mostly in prepared food.
Open it in the sampleShort answers to one question about one store, like “What changed the week of Aug 3?”, from its manager or your ops lead.
Seurat sends the question. You answer by email, text or a short call.
Asks every store the same 7 questions. Your sales data answers 3, public sources answer 3, and your team answers the last. Later layers add 2 more.
Heartland, Day 3: Store 14's gap traced to a new sandwich that nobody on its lunch shift had been trained on. Retraining is worth about $185,000 a year.
Open it in the sampleThe dates of your changes: launches, remodels, promotions, and which stores got each one.
Confirmed in a 30-minute call. Seurat finds most of them first, from your sales and the public record.
Shows where each rollout landed, store by store, in the Rollout Read. Then the Calibration Read measures each change separately, against similar stores that didn't make the change and against other chains that did. Then it recommends what to do with each one: roll it out, keep it with a fix, or stop it.
Heartland, Week 3: its 5 changes add about $20,200 a week. Roll out the burrito, keep 2 with a fix, stop 2.
Open it in the sampleSix months of item sales for prepared food, dispensed drinks, packaged drinks and snacks.
Drop it in, once. One export from your back office.
Finds the items behind each gap and each change, then turns what it finds into levers: tactical actions you can take at the stores where they fit, including ideas that worked at other chains. Each lever gets a yearly value and a confidence range.
Heartland, Week 6: 5 levers, worth about $1.4 million a year together.
Open it in the sampleA photo from the store when a lever starts there: the new warmer, the menu board, the shelf.
Text or email it from any phone. Each photo starts that lever's clock at that store.
Dates each lever at each store from its photo, then reads those stores against their twins every week. The read tightens week by week until it's final. Then the number stops changing, and the lever joins the Calibration Read's graded changes.
Heartland, Week 9: the levers are adding about $16,900 a week, about $880,000 a year measured so far.
Open it in the sampleThe public layers explain gaps before anyone has to ask. On Day 2, they explained 3 of the 4 stores outside their expected range, so only Store 14 needed a Closer Look.
Tap the layers you could send.
Nothing connects to your systems. We only see what you send us.
Mutual NDA first, never resold or shared, deleted 90 days after the work ends or sooner if you ask.