How we measure results
See what every Esplanda action earned, in dollars
Before an action runs, Esplanda writes down what it should earn; afterwards it records what it actually earned, side by side.
The ledger comes with Esplanda. Esplanda POS and your website are free to start, with no card needed. See how Esplanda works.
Illustrative example
Reading the ledger
Three things to notice in that month
A dashboard reports what happened. A ledger also records what the software told you to do, whether it was done, what it expected that to be worth and what it turned out to be worth.
Expected and earned, side by side
The expected column is written before anything runs, so every dollar earned is read against a number set in advance.
Measured against a held-back group
The win-back text is scored against a matched group of customers who didn't get it, so its dollars are the extra margin it brought in.
Activity, kept out of the dollars
Review replies count as activity: tracked, and kept out of the dollar total.
The ledger
Six things written down for every action
How a result will be measured is decided when the action is prepared, never afterwards. That order is what makes the numbers worth reading.
Before it runs
What it saw
The signal behind the action, with the numbers as they stood: a best seller running low, a cost increase, a regular who stopped coming in.
What it did
The specific action, not a category: this order, this price, this offer to these customers.
What it expected
A single dollar figure, written down before anything happens, so it can't be adjusted to fit the result. Before you commit, what-if scenario planning software gives the likely range.
How it will be measured
The comparison, the length of the window and whether it counts margin or sales.
After the window closes
What it earned
Filled in from your own sales data, in dollars.
The gap, and why
Over, under or on target, with the reason the data shows.
Methods
How each kind of action is measured
Every method is published, and every row says how sure its number is.
| Action | How it is measured | Counts as |
|---|---|---|
| Campaign: email, text or loyalty offer | Margin from the customers who got it, compared with a matched group left out on purpose | Measured |
| Price change | Margin on the item over four weeks, adjusted for its category's own trend | Measured |
| Dead stock cleared | Cash recovered after the markdown, plus the carrying cost avoided | Measured |
| Paid ads | Margin attributed to each channel over a fixed window, after the ad spend | Measured |
| Prevented run-out | Units sold during the days the shelf would have been empty, at the margin you actually made | Estimated |
| Local search and AI visibility | Change in calls, direction requests and profile visits, shown as a trend | Reported as a trend, kept out of the dollar total |
| Review replies | Tracked reply by reply as activity | Activity, kept out of the dollar total |
A prevented run-out starts as a supplier order from purchase order software, a price change starts in competitor price tracking, a text or email campaign in email marketing for small business, and paid ads run from marketing automation for small business.
Four labels for how sure a number is
Every dollar row also carries one of four labels. The label says how firmly the action is tied to the dollars, and it sets how much the row counts in the monthly total.
Caused
The action produced the result.
Needs a fair, randomized test that ran to a clean finish. Nothing else earns the word.
Counts in full
Likely caused
Strong evidence, and the other explanations were checked.
Needs a comparison with a held-back group or a matched set, with every check passing.
Counts at four fifths
Associated
The number moved, and this action is a plausible reason.
Needs a comparison with what was forecast to happen otherwise.
Counts at half
Recorded only
Kept in the record, with the reason beside it.
Needs the reason stated: too little data afterwards, another event in the way, or an effect too small to see.
Kept out of the total
Every label is shown in plain sight. A result that fails a check moves down a rung with the reason beside it, and earlier results keep the label they were measured with, even after a stronger method arrives.
Reading the numbers
What every number tells you
Most software that says it made a store money compares what happened with its own forecast and stops there. Esplanda labels every line by how firmly it is tied to the dollars, so strong results read as strong and every line shows its method.
How to read the total
- It is a list you can audit. Each line has an amount, its method and its label. The total is weighted by label, with the unweighted sum beside it, so you can see exactly what the weighting took out and keep only the lines you trust.
- Nothing counted twice, and the price taken off. When a campaign and a price change both plausibly moved the same sales, the overlap comes out, and what Esplanda costs you is subtracted before the total is shown, so every total is net of what you pay Esplanda. See Esplanda pricing.
- It is conservative. The monthly number only counts what can be traced, so it is likely lower than the true effect.
- Results set its autonomy. Work that falls short of its forecast goes back to asking you first. More on why Esplanda is safe to trust.
- One set of numbers. With Esplanda POS, the ledger reads the same records the register wrote, with nothing matched in between. Compare the POS system with inventory management with our POS integrations.
Where the total comes from
- Actions, one by one. The monthly number adds up effects traced to specific actions, each with its amount, its method and its label.
- Labels that mean something. Only a fair randomized test earns "caused", a held-back or matched comparison earns "likely caused", and a comparison with the forecast earns "associated".
- Your own sales data. Every earned figure is filled in from your register's records, in dollars.
- The whole business beside it. Your profit and loss report shows how the store moved; the ledger shows which dollars Esplanda's actions earned.
Tests sized to give a real answer
Sized first
Two subject lines, a small list
Say your email list is under two thousand people. To reliably see a five-point difference in opens, each version would need well over a thousand people. Esplanda shows you that arithmetic and the smallest difference your list could detect, then sends the line it thinks is better and records it as a judgment call.
The ledger keeps its evidence for comparisons big enough to show a real difference. The bigger options: test an offer against no offer, pool several weeks of sends into one longer test, or grow the list first.
Runs
An offer big enough to detect
- A fair split: each customer is assigned by a stable rule, so the same person always lands in the same group.
- Limits set before launch: a margin floor, an unsubscribe ceiling and a stop if one side is clearly doing harm.
- A held-back group that gets neither version, so you also learn whether doing anything beat doing nothing.
- Safe to check every day. If it stops early on a clear result, the figure reported is the corrected one, not the flattering one.
Never tested. Prices are never set by who the customer is: a price test separates by day or by location, so everyone in the store at the same time sees the same shelf price. Age checks, allergen and license disclosures and debt-related messages are never tested at all. These rules live in the system, not in a policy.
Quarterly review
The quarterly review, in plain dollars
Each quarter, the review reports what Esplanda was measured to be worth, in dollars, part by part. It is drawn straight from the ledger, so every figure traces back to an action. The review is part of POS system support, alongside free email and chat.
Every part, by name
Ordering, pricing and marketing each show what they earned, in dollars, so you see exactly where the quarter's result came from.
Clean attribution
When another event overlaps an action, like a competitor's promotion during your campaign, the action is marked and kept out of the total.
Built from measured results
Every figure in the review comes from measured results in your ledger.
Advice to add more only on evidence. A suggestion to add a paid tool or give a capability more freedom needs three things your account shows: you approve most of what's proposed, the measured result is positive, and you keep reaching your limits. A low approval rate produces no suggestion, whatever your sales.
Questions
Measuring results: questions owners ask
How does Esplanda measure results?
Every action Esplanda prepares or takes goes into a ledger with what it is expected to earn, in dollars, and how the result will be measured, both written down before the action runs. When the measurement window closes, Esplanda records what the action actually earned next to the prediction.
What is the Esplanda ledger?
The ledger is the record of everything Esplanda has done for your store: what it saw, what it did, what it expected to earn, how that was measured and what it actually earned. It lets you check the software the way you would check an employee's work.
How is this different from my analytics dashboard?
A dashboard reports what happened. It does not record what the software told you to do, whether you did it, what it expected that to be worth, or what it turned out to be worth. Recording all four is what makes software checkable.
How do you measure the ROI of a marketing campaign?
Esplanda compares the customers who got a text, email or loyalty offer with a matched group that was left out on purpose, over a window fixed before anything is sent. The result is the extra margin from the customers who got the offer, after the cost of the discount. When a list is too small to show a real difference, Esplanda shows you the arithmetic and records the send as a judgment call.
How do you separate Esplanda's effect from a busy season?
With a comparison group wherever one is possible, and by adjusting for the category's own trend where it is not. Each result then carries one of four labels: caused, only for a fair randomized test; likely caused, for a held-back or matched comparison that passes its checks; associated, for a comparison with the forecast; or recorded only. Weaker labels count for less in the monthly total, and a recorded-only result is kept out of it.
Does the ledger work if I keep my own register?
Yes. Esplanda measures from Esplanda POS or from a connected Square, Clover, Lightspeed or KORONA register. On Esplanda POS the ledger reads the same records the register wrote: one set of numbers, nothing matched in between.
How do I see whether Esplanda pays for itself?
The ledger shows you every month: what Esplanda's actions earned, in dollars, next to what you pay Esplanda, with the method and label on every line.
How soon does the ledger fill in?
The first rows are filled in within days, as the first actions finish their measurement windows. After a month or two, with more actions measured and forecasts settled to your store, it guides your bigger decisions.
Hold Esplanda to its own numbers
Start free today, or book a demo and we'll walk you through the ledger on a store like yours.
Esplanda POS and your website are free to start, with no card needed. Add paid tools only when you want them. See Esplanda pricing for every line.
