How to measure the lift of a limited-time offer across locations

To measure the lift of a limited-time offer across locations, compare customer counts at locations running the offer with a weighted twin of locations that aren’t running it yet. Do it over the offer window and a post-offer window of the same length. The net difference is the lift. LTO units sold, sales versus last year and sales versus the four weeks before are all easier numbers to pull. They all overstate it.

LTOs are built to be bought on a whim. Datassential’s research found that nearly two-thirds of LTO purchases are impulse buys (Source: Datassential, via White Hutchinson Leisure & Learning Group, 2022). That’s the point of them. It’s also the measurement problem, because an impulse purchase by a regular who was coming in anyway looks exactly like a new customer on the POS.

Why LTO units sold isn’t the lift

Your POS will tell you exactly how many of the new spicy chicken sandwich you sold. It can’t tell you how many of those buyers came in because of it. Every LTO unit lands in one of four buckets:

• Swapped. A regular who’d have ordered the usual picked the LTO instead. The item mix changed; the customer count didn’t.
• Pulled forward. Someone who’d have come next week came this week. The bump now is paid back by a dip later.
• Pinched. From a competitor, which is real, or from another of your own locations, which isn’t much use to the brand.
• New. A visit that wouldn’t have happened at all. This is the lift you’re after.

Research on supermarket promotions found the sales bump split roughly into thirds: switching between brands, purchases borrowed from other weeks, and genuinely new demand for the category (Source: van Heerde, Leeflang and Wittink, Marketing Science, 2004). A restaurant or retail chain isn’t a supermarket shelf, but the lesson travels. A big share of any promotion’s bump is borrowed.

How to set up an LTO so you can measure its lift

The biggest decision happens before launch: where the offer doesn’t run, at first. A national LTO with national media on the same day leaves nothing to compare against except last year, and last year had different weather, prices and competitors.

1. Stagger the launch. Hold the offer back in a set of markets for the offer window plus the post-period, typically four weeks plus four. Nobody loses the LTO; those markets just get it second. Pick them the way you’d pick geo holdout markets: pair similar markets and flip a coin, rather than asking which region is least excited.
2. Split the offer from the media. Run the offer with no extra local media in a third group: in-store signage and the app only. Comparing it with the offer-plus-media group tells you what the ads added.
3. Fix the outcome. Transactions or guest counts per location per week. Track average check separately. It tells you about upselling, not about new customers.
4. Freeze everything else. No price changes, other promotions or menu removals in only one group.
5. Plan the post-period. At least as long as the offer, so you can see the dip.
6. Check stock. If test stores run out of the LTO in week three, your test ended in week three.

A worked example: a four-week LTO across 60 locations

This example is illustrative. A 60-location fast-casual chain runs a four-week LTO: a premium bowl at a normal price, no discount. 24 locations get the offer plus $48,000 of local paid social and search ($12,000 a week). 18 get the offer with in-store signage only. 18 hold it back for eight weeks. Every location averages 3,000 transactions a week, and each transaction earns about $4.10 of contribution margin after food, packaging and variable labour.

The team’s first read: “Offer-plus-media stores were up 7.5% on the four weeks before.” That’s 72,000 × 7.5% × 4 = 21,600 extra transactions. Here’s what happens when you take it apart.

The first read of 21,600 drops to 12,384 once you take out what the season did on its own (72,000 × 2.0% × 4 = 5,760) and the dip afterwards (3,456). It overstated the lift by about 74%.

Now split it. Offer-only stores netted 264 extra transactions each. Offer-plus-media stores netted 516. The difference, 252 per store across 24 stores, is about 6,048 extra transactions bought by $48,000 of media: $7.94 each. At $4.10 of margin per transaction, the media earned back about $24,800 on $48,000. The offer itself was worth running. The local media behind it wasn’t, at this spend.

Check the range before you act. If the 95% range on the media’s share runs from 3,000 to 9,000 extra transactions, its cost runs from $5.33 to $16 per extra transaction. Even the best case is above the $4.10 break-even, so the conclusion holds across the range. The right reaction isn’t “kill LTO media”. It’s “next time, test half the spend, or a different channel, and read it the same way”.

Don’t stop at transactions: count the margin

Lift in transactions is where the money starts. An LTO moves three lines:

• Margin on the net extra transactions. Net extra transactions × contribution per transaction.
• The mix effect on everyone else. If regulars swap a $12 core item for a $14 LTO, average check rises on transactions you’d have had anyway. If the LTO is a discount, it falls. Price the swap explicitly.
• The offer’s own costs. Training, signage, new ingredients, and waste when the window closes.

Discounted offers need extra care. A $3 discount goes to everyone who buys, including all the regulars who were coming anyway, and on a popular deal that giveaway can dwarf the margin on the genuinely new customers. Our guide to baseline vs incremental sales covers that split in more detail.

What if the LTO already launched everywhere?

Then you can’t get a clean read, and anyone who hands you one is guessing. You have three weaker options:

• Accidental late starts. Supply problems, a franchisee who opted out, a market that started a week behind. These weren’t chosen at random, so treat them with care.
• Media weight. If local media weight varied by market, comparing heavy and light markets gives a rough read on the media, though not on the offer itself.
• A forecast. Fit a model on a few years of history and compare the LTO window with what it predicted. Better than year-over-year, but it can’t separate the LTO from anything else that happened those weeks.

Then plan the next one properly. LTOs come round several times a year, so each one is a chance to run a cleaner test than the last. To size the holdout so it can actually see the lift you expect, use the minimum detectable effect arithmetic.

Common mistakes when measuring LTO lift

• Counting LTO units as new customers. Most of them are swaps. Measure transactions, not items.
• Comparing with the four weeks before. Seasonality moves those numbers, and the weeks before a launch are often soft because the last offer was winding down.
• Comparing with last year. Different prices, weather, competitors, store count and promotions. Any one of them can be bigger than the lift.
• Ignoring the post-offer dip. It’s where the borrowed visits get paid back. Read at least as many weeks after as during.
• Measuring only stores with media. Without an offer-only group you can’t tell whether people came for the bowl or for the ads.
• Rolling up to the brand total too soon. Some markets respond and some don’t. Look at the spread by location first; our guide to restaurant marketing attribution by location shows how.
• Calling it after week one. Launch weeks run hot on curiosity and staff enthusiasm. Read the full window.

Questions people ask about measuring LTO lift

An LTO is one of the few marketing moves you fully control: what, where and when. Use that control to keep some stores back, and you’ll know what it did.

Written by The Countlift team at First Byte in The Woodlands, Texas. Questions or corrections? Email hello@countlift.com.