Seurat Analytics ongoing monitoring

Predictive Sales Monitoring

Minding the gap between each store and its twin

What today's KPIs say is your best store might actually be your worst one.

We build a twin of each store. The gap between the store and its twin shows how well the store is being run.

How a store gets judged

That store you were about to close... Not so fast!

Site 227 sits at the bottom of every list. Site 214 sits at the top. Here are the same 16 weeks of sales, judged 5 ways.

Reference line: your chain average, your data

Site 214Top of your chain. Best store.

Site 227Bottom of your chain. Worst store.

Solid lines are actual weekly inside sales. Dashed lines are the reference the view is judging them against. The sales data is identical in all 5 views.

Judged against the chain average, you'd close your best-run store and praise a site that's leaving $156K a year on the counter.

The 3 comparisons every operator already makes

Against competing stores. A peer set you don't choose, with formats and trade areas unlike yours.
Against your other stores. A ranking on sales mostly ranks trade area, traffic and format, not how each store is run.
Against its own last year. Everything that changed since is folded in. A store up 3% means little until you know what it would have sold had you not made your changes.

All 3 ask how a store compares, not whether it's doing as well as it should. A benchmark ranks your site against other sites, while a twin uses them only to measure conditions, then compares your site with itself.

How we build the twin

Each twin is refit on a set cadence from 3 things you already have.

Its own history. How the site's sales move by week and by season.
The rest of your network. Comparable sites the same change didn't touch, so a bad month everywhere doesn't count against a single store.
What was going on outside. Fuel prices, weather, promotions and seasonality, which every comparison site went through too.

The heavy work happens once per site, which is why this is a subscription. The full mechanism is at How It Works.

Before we switch anything on, we run it over your last 2 years as if it had been live. You see every store it would have flagged, and which of those turned out to be real problems.

Minding the actionable gap

Every store has good weeks and bad weeks. We only flag a store when a week falls outside its normal ups and downs.

The drift no single store can see

1Inside the band

At 1 store, a 1.5% drift is about $33,000 a year and hides inside the band. Across 30 stores it's about $985,000 a year, and read together it clears the band in about 9 weeks.Illustrative: stores selling $6,000 a day inside.

The leak is too spread out to see store by store, which is why we watch the chain. The band narrows over the weeks as well, so a shortfall hidden at the start shows up later.

Start monitoring Ask a question first

When a gap opens, there are 3 explanations

The site isn't running the lever, the lever doesn't work there, or it's noise. The confidence band rules out noise, and the register can't tell the other 2 apart.

Running it

Not running it

Actual sales at or above the twin

It's working. Roll it wider.

Something else moved sales. Check that before spending on the lever again.

Actual sales below the twin

The lever doesn't work at this site. We replace it on the board.

It was never run. We'll tell you which sites and from what date.

How we read execution without anyone filing a report

Out of the register. Whether the item rang at the promo code, from what date and at which sites.
Out of a photo. Staff photograph fixtures the register can't see as soon as the work is done, and the timestamp is the start date.
Out of a second photo, later. Repeat photos on a short list of levers show the fixture is still up.
Out of a visit. Limited in-store visits, 2 to 4 a year, where the register and the photos disagree.

What lands on your desk

A weekly exception list. Only the sites outside the band that week, ordered by dollar impact.
A monthly settlement. Every lever we sized, next to what it delivered.
A quarterly board update. Rows that settled, came off, or were worth more than we said.
Nothing to log into daily. If nothing arrives, nothing fell outside the band that week.

Why this comes from whoever did your prior reads

The bands, the measured effects and the comparison pools all come from your prior reads. A new vendor starts from zero and needs a quarter of history before it can build a twin of any store.

What it costs

$50 per site per month. Ongoing and billed monthly, so a 30-site chain runs $1,500 a month.
Limited in-store visits included. From 2 to 4 a year, scaled to your chain.
No new data ask. It runs on the feed your Lever Board already uses, plus photos from your front-line staff.

Start monitoring Ask a question first

Where this sits

We can get further without your data than you would expect. The dollar column is where that stops.

	You give us	You get
Store-to-Store Read	Nothing	Which of your stores are underperforming, and what your customers say is wrong with them
Calibration Read	One data export	What the changes you've already made actually contributed to in-store sales
Lever Board	2 years of sales data	Every proven lever ranked and suggested for each of your stores, with its lift contribution evaluated
Remodel Sequence Read	2 years of sales data and your capital plan	Which sites on your capital plan should get the money first, which shouldn't get it at all, and what would have to change for the rest
Monitoring	Nothing new	Verdicts as the reads settle, plus where a lever is not showing up in the data the way it should, which usually means it is not being executed the way it was specified