The ad platform will always tell a franchise partner the campaign worked. That number counts orders you had already won, gets claimed twice across two platforms, and knows nothing about your margin. This is how head office works out what digital local marketing really returned, using an advantage most businesses do not have: a few hundred near identical stores.
Every ad platform will hand a franchise partner a return on ad spend figure, and it will usually look good. It is worth being precise about what that number is actually counting, because a franchise board will eventually ask, and "the platform said so" is not an answer.
A regular customer who was always going to order this week, and happened to scroll past the ad first, is counted as a conversion. At a local level, where most of the audience already knows the store, that group is large.
Meta and Google both count a conversion they touched. Add the two dashboards together and you get more attributed orders than the store actually took.
The platform does not know your food cost, and it certainly does not know what the offer gave away. A discount-driven campaign can post a strong return and still lose money on every order.
None of this means the dashboards are useless. They are good for comparing one creative against another. They are not evidence that local marketing grew the business.
In that trade area, in those weeks, compared with what the store would have done anyway. Not compared with last year, which had a different price list and a different national plan.
The media spend, plus the margin given away on the offer, plus print if there was print. Against the gross profit on the extra orders, not their revenue.
Look at the weeks after the campaign stopped. A campaign that pulled orders forward from next week is a cash flow event. A campaign that recruited customers is growth.
This is the quiet advantage of a multi-site brand. You have many near identical units selling the same product at the same price, which is exactly the setup an experiment needs. A single site business has to guess. You do not.
| Method | How it works | What it costs you | How much to trust it |
|---|---|---|---|
| Matched store holdout | A group of similar stores deliberately left out of the campaign | Those stores miss a campaign | The best answer available |
| Staggered start | Half the network starts weeks one to four, the rest weeks five to eight | Nothing, every store gets it eventually | Good, if trading conditions are stable |
| Region split | One region runs it, a comparable region does not | Nothing extra | Fair, weaker if the regions differ |
| Before and after only | Compare the store with itself, last month against this one | Nothing | Weak, because everything else moved too |
Swipe the table sideways to see the rest.
The staggered start is the one most networks can actually get approved, because nobody has to be told they are the control group. Run it once properly on a flagship campaign and you have a number you can defend for the whole year.
Media spend is the easy number. It is rarely the whole number, and the parts people forget are the parts that decide whether the campaign was worth running.
Given away on every redeeming customer, including the ones who needed no persuading. Usually larger than the media spend.
Every dollar shown to somebody the store cannot serve. On a radius this is routinely a large share of the budget, and it converts at zero. See delivery area targeting.
Design time per store, and the cost of flyers and posters if the campaign included them. See print materials.
Brand team hours building it, and partner hours setting it up. This is the cost that a platform is supposed to remove. See running LSM at scale.
Harder to see, and often the biggest. Stores that launched nothing at all produced nothing at all, and that is a cost with no invoice attached.
If the local layer is showing the national offer to the national audience again, some of that return already belonged to national. See local vs national.
Industry benchmarks in a vendor deck are not evidence about your business. Your network is. For any campaign, the three comparisons that mean something are the same store before it ran, the median store across the network on the same campaign, and the stores that did not run it at all.
Amplaro reports spend, orders and results per store and per network, so those comparisons are the default view rather than a project. It also reports the uncomfortable one, which stores have launched nothing, because that number belongs in the same conversation. See reporting.
Anyone quoting you a single number has not seen your margins. A four times return on a full price product and a four times return on a heavily discounted one are completely different outcomes, and one of them can lose money. The useful benchmark is your own network: the median result across your stores for the same campaign, and the same store's own baseline before it ran.
Three reasons, and they stack. The platform counts orders inside a generous attribution window whether or not the ad caused them. Two platforms will both claim the same order, so the numbers do not add up to your total. And the platform has no idea what your margin is, so it reports revenue as if it were profit. None of that is dishonest, it just answers a different question than the one you are asking.
By comparing against stores that did not run it. Hold a matched group out, run the campaign everywhere else, and compare the change in sales between the two groups over the same weeks. That difference is the closest thing to a real answer you will get without a research budget, and a franchise network is unusually well set up to do it.
No. You need store level weekly sales, which you already have, and the discipline to leave some stores out. The analysis is the difference between two averages. The hard part is not the maths, it is resisting the pressure to give every store the campaign because it feels unfair to hold any of them back.
Long enough to cover a full trading cycle, so at minimum a few weeks, and ideally with a few weeks after the campaign ends as well. Stopping the measurement the day the campaign stops hides the two things you most want to know: whether the volume held, and whether you simply pulled forward orders from the following week.
Yes, and most reporting does not. If a campaign drove a thousand orders at a discount, the margin you gave away on all of them is part of what the campaign cost, including on the customers who would have ordered at full price. Leave that out and a deep discount will always look like your best performing campaign.
Match them on the things that actually drive sales rather than on geography. Trading history, store format, delivery mix, size of the trade area, how long they have been open. You do not need identical stores, you need two groups whose averages behaved the same way in the months before the test.
It is the least glamorous lever and usually the biggest one. Budget that reaches people outside the store's delivery area cannot produce an order at any conversion rate. Tightening the targeting to the real service area raises the return without changing the creative, the offer or the spend. See delivery area targeting.
Return reads differently when the decision takes weeks. Eg: gyms and childcare. More on franchise industries.