Planning Reads · Pre-launch pilots

Which stores should run the pilot, and how will you know it worked?

Every Payoff Read measures a change you've already made. This read brings the same method to a change you're about to make. Before launch, we help you pick pilot stores that represent your whole chain, build each store's twin from stores outside the pilot, and right-size the pilot so it can prove the change's ROI. After launch, each pilot store is read vs. its twin, so you know where to roll the change out, and what to adjust where it fell short.

12 Pilot stores
18 Stores the twins are built from

Heartland Markets is a fictional chain. Every number on this page is made up to show the format.

What your numbers can't show

Standard pilot reports show only what the pilot stores sold, mixed in with everything else happening at those stores. They can't show what those stores would have sold without the pilot.

What you have: sales at the pilot stores before and after launch, compared to last year or to the rest of the chain.
What standard reports can't show: whether the pilot itself caused the change in sales. Retailers often pilot at their best-run, busiest stores, or the ones closest to the office, and those stores would likely have beaten the chain's sales performance without the pilot. A few stores over a few weeks also can't show a small lift, and a pilot too small to show its effect looks just like one that failed.
What the read shows: before launch, which stores to pilot, each store's twin, and how long to run it. After launch, exactly what the pilot added at every store, and where the change should go next.

Every pilot has a floor

The floor: the smallest lift a pilot can reliably detect.
What sets it: how many stores run the pilot, for how long, and how much their sales swing from week to week.
Below the floor: a 3% lift and no lift come out looking the same in the final analysis.
So we find it first: before the pilot starts, not after it ends.

What lands on your desk

A pilot, as Seurat sets it up and reads it

Heartland Markets wanted to try ordering kiosks before buying them for the 30 stores that make food to order. Seurat helped pick 12 of the 30 as pilot stores, chosen so that together they look like all 30 on food sales, lunch traffic and size. Each pilot store got a twin built from the other 18 stores that make food to order, and the pilot ran 12 weeks from March 2, 2026.

The answer, in three lines:

Before launch: 12 weeks, so the pilot could detect a 4% lift in food sales. The 4-week pilot first planned could only have detected 7%, so a real 5% lift would have looked like no effect.
After 12 weeks: kiosks added 5% to food sales at the 12 stores, +$400 a week per store, statistically significant. Most of it came from the add-on the kiosk suggests with every order.
They pay at the 7 stores with a lunch rush, where they shortened the line. At the 5 without one, they added nothing you can measure, so the rollout goes to the 11 stores that look like the 7, not to all 30.

per pilot store, per week, the 12 weeks after the kiosks went in, vs. its twin

What changed	Result	Verdict
Food sales	+$400	Significant lift
Orders with an add-on	+7 points	Significant lift
Lunch visits, 11 a.m. to 2 p.m.	+22	Significant lift
Margin on the added sales	+$185	Significant lift
Kiosk lease and support	−$95	Cost of the kiosks
What the kiosks add to the bottom line	+$90	No significant effect

Why 12 weeks: with 12 stores and Heartland's week-to-week swings in food sales, 12 weeks can detect a 4% lift. The 4 weeks first planned could only have detected 7%. The kiosks added 5%, so the shorter pilot would have called them a failure.

By store type: kiosks pay at the 7 pilot stores with a lunch rush, where they shortened the line and suggested an add-on with every order. At the 5 without one, food sales didn't rise: there was no line to shorten.

by store type, per pilot store per week

Store type	Kiosks add	Confidence band	Verdict
7 pilot stores with a lunch rush	+$220		Pays for itself
5 pilot stores without one	−$90		No significant effect

What shoppers told us

7%of customers rank app or kiosk problems among their top 3 reasons to give up on a store.
75%say smoother operations make them prefer a store.

A kiosk pays where it shortens the line, not where it's new.

KeepFixStopWhat to do: keep the kiosks at the 7 lunch-rush stores, and roll them out to the 11 stores that look like those 7. Fix the other 5 by putting their kiosks in lunch-rush stores instead. Stop the plan to put kiosks in all 30.

The confidence band shows the 95% confidence interval. When the band spans zero, the result isn't statistically significant. Heartland Markets is a fictional chain. Every number on this page is made up to show the format.

Before the pilot

Not sure what to pilot? Start with a free Fit Check

A Fit Check shows which of your stores look like the places where a concept we've spotted works, and where to try it first. A Pre-Launch Pilot Read turns that list into a pilot you can read. It's free, you don't send any sales data, and the quick read takes about a minute.

Order aheadWhere to try ordering ahead first, store by store, each with a twin.Deploy self-checkout where lines formThe stores busy enough for self-checkout now, the ones to watch, and the ones where it would sit idle too much of the time.Salad bases with pricey add-onsWhere to try the salad first, store by store, each with a twin.

How the read works

This read starts before launch. Most store tests compare pilot stores with a matched group of other stores, so the result depends on how alike the 2 groups happened to be, and nobody knows until the end whether the test was big enough to show anything. Here, each pilot store gets its own twin, which isolates what the change did at that store, and the pilot is sized before launch, so you know up front the smallest lift it can detect.

Pick pilot stores that represent your whole chain. Retailers often pick pilot stores that are the best run or the closest to the office, so they'd likely beat the chain without the pilot. We don't do that. Instead, we pick stores that, together, look like the rest of your chain on sales, traffic and format, so the result applies to your entire chain.
Give each pilot store a twin: a predicted sales line built from stores outside the pilot, weighted so it matches the store for a year or more before launch. The twins are set before launch.
Find the pilot's floor. The smallest lift a pilot can reliably detect depends on how many stores run it, for how long, and how much their sales swing from week to week. We set the number of stores and weeks so the floor sits below the smallest lift that would make the change worth rolling out.
Read the pilot, evaluate what it's worth, then test the read. After launch, each pilot store is read vs. its twin and valued at your margins, with the change's cost taken off. A placebo test reruns the read on a made-up launch date before the pilot. Because there's nothing to find during this period, the placebo test should show no effect. If it shows an effect, the read isn't working properly, and we tell you.

Already ran a pilot? It can still be read, as long as you know which stores ran it and when. The twins are built from your history the same way. If the pilot turns out to have been too small to detect the lift you needed, the read tells you that, instead of calling the change a failure.

Can your pilot be read?

It depends on where the pilot is today. There are 3 routes.

1
A pilot you're planning
How the read works
We help you pick the stores, build the twins and set the length before launch
What it answers
Where to roll the change out, store by store, with a confidence interval
2
A pilot already running
How the read works
Each pilot store vs. a twin built from stores outside the pilot, from the launch date
What it answers
Whether it's working yet, and how many more weeks it needs to give a statistically significant read
3
A pilot that already ended
How the read works
The same read on the pilot's dates, with twins built from your history
What it answers
What the pilot added, or whether it was too small to tell

Pilot	Smallest lift it can detect
12 stores, food sales, 12 weeks	About 4%
12 stores, food sales, 4 weeks	About 7%
4 stores, food sales, 12 weeks	About 7%
12 stores, packaged drinks, 12 weeks	About 3%

Rough figures from a standard formula, before we see your data. More stores, more weeks and steadier sales all lower them.

We check your store list and the change you want to try on the first call, before you commit to anything.

Scope this read

What we need, and what stays with you

We need:

Weekly sales by store, split into food, drinks and everything else
What you want to pilot, and the stores it could go to
The launch date at each pilot store, and anything else that changed at those stores during the pilot
What the change costs per store, and your margins by category, as percentages
Transactions by store and hour, if your system keeps them
1 to 2 years of history

We never ask for:

Names, phone numbers, emails or anything else that identifies a customer
Employee names, shifts or payroll records
Your chain's totals or your P&L

How we handle your data →

How long it takes

About 2 weeks from the day your data lands, you get the pilot plan: the stores, the twins and how long to run it.
When the pilot ends, the read comes about 2 weeks after its last week of data, then a 60-minute readout with whoever needs to hear it.
We scope it to your chain on a short call: what you're piloting, where it could go, and when you need to decide.

Scope this read Book a call

Questions buyers ask us

Why not just pilot at our best stores? Because they'd beat the chain without the pilot, and the result won't hold at the stores the change goes to next. Pilot stores that, together, resemble your chain, each read vs. its twin, give you a result that applies to the rest of the chain.
Do some stores have to wait? Only for the length of the pilot. The stores outside it build the twins, and they get the change when the rollout reaches them.
Our pilot came back flat. Did it fail? Not necessarily. A pilot can be too small to detect the lift it was looking for. The read shows the pilot's floor next to its result, so you can tell a change that did nothing from a pilot that was too small to tell.
How is this different from test-and-learn software? Both compare pilot stores to similar stores. A Pre-Launch Pilot Read gives each pilot store its own twin, so you see where the change worked, store by store, and it runs on an export you already have, with no software to set up.
Can you tell us what to pilot? A free Fit Check shows which of your stores look like the places where a concept we've spotted works, and where to try it first. A Pre-Launch Pilot Read turns that list into a pilot you can read.
What if the pilot shows the change doesn't pay? A read with no significant lift is a private finding. You see every result first, and nothing goes anywhere without your approval.
Who does the work? We do. Your team runs the export, shares it with us, and comes back to hear our readout.

After this read

Every read shows where a change worked and where it didn't, so it naturally raises the next question: why? The next read usually runs on the same export you've already sent.
Reads cost less in sets of 3, and your first read will be credited toward a multi-read discount.

Try a free Fit Check →Scope several changes →Start with one read →How monitoring works →

Sources: Seurat Analytics, "Same Brand, Different Store Experience," a national survey of 1,200 U.S. shoppers, July 2026.