Imagine the rush of success you feel from sending an email promoting a new item, then seeing orders for that sandwich spike at certain locations. The email is working, right?
Then, after your victory dance, you find out that a menu item had gone out of stock at a few locations that same week, and guests were rerouting to whatever was still available. As much as you want to give your email the credit, the out-of-stock sandwich probably did the actual work.
That's the trap with restaurant marketing metrics: many are technically accurate and still tell the wrong story. Here are five metrics I see brands misread most often, and what to look at instead.
The problem with counting every loyalty signup
This is the one I'd flag first if a brand asked me where to start. It's common to see a brand hold onto its total loyalty count like a trophy: 26 million guests signed up, reported proudly to the board every quarter. Look closer, though, and maybe 8 million of those guests have actually ordered in the last three years. The rest joined for a signup incentive once and never came back.
That headline number feels good to report, but it doesn't say anything about whether the loyalty program is working.
Solution: Track active members instead of total signups, and define "active" in a way that makes sense for the brand. A fast-casual salad concept might define active as ordered in the last year; a higher-end dining brand, where guests can't visit as often, might set that window at two years. Total signups mostly reflect how good your signup incentive is, while active member count reflects reality.
The blind spot in attributed revenue
Attributed revenue looks like a clean number: this guest got an email, this guest ordered, therefore the email drove this revenue. The problem is in how "got an email" gets defined. Most reporting counts email deliveries, opens, and clicks within a set window, and credits their orders to marketing. Delivery is the loosest of the three. If a message simply landed in someone's inbox, it doesn't mean they ever saw it, and yet the order still shows up as marketing-driven.
The real issue is that a meaningful share of the guests on your list would have ordered that week whether or not the email went out. Attributed revenue alone can't tell the difference between an email that changed behavior and an email that happened to land near a purchase already in the works.
Solution: Pair attributed revenue with holdout groups. Hold back a percentage of your audience from a given send, watch how their ordering behavior compares to the group that received it. This gives you a real read on whether the campaign influenced anyone.
Open rate stopped being reliable years ago
Open rate used to be the go-to health check for an email program. Then Apple's Mail Privacy Protection started auto-opening messages for anyone using the Apple Mail app, whether or not a human ever looked at them. That inflates open rate and drags down click-to-open rate, because the "opens" in that denominator were never real people to begin with.
Plenty of brands are still leaning on open rate as their primary email metric anyway. It's familiar and easy to report, so most teams haven't rebuilt their dashboards around what actually changed.
Solution: Shift the focus to click-to-delivered rate and unique clicks on non-header links. Anyone clicking every link in an email is very obviously not a person browsing a menu, so looking at engagement with a specific call to action gives you a cleaner picture of who's actually paying attention.
Average frequency falls flat without differentiation
Most restaurants track average frequency—how often guests come in. But if you’re making marketing decisions with that number, you might need to get more specific.
Say your top guests order twice a month, your middle guests order once a month, and your low-frequency guests order quarterly. Average those together, and you might land on an average frequency of 45 days.
But, that 45-day figure doesn't actually describe any of those three groups. If you time a lapsed-guest email off that single average, you'll catch your top guests too late and your low-frequency guests way too early.
Solution: Keep the general average as a baseline health check, but build your actual marketing around frequency by segment. RFM segmentation (recency, frequency, monetary value) is one reliable way to split guests into groups and market to each on its own timeline, rather than one calendar for everybody.
What a "successful" offer redemption rate misses
A high redemption rate feels like a win, and often it partly is. But it’s critical to ask two questions: what behavior were we trying to encourage, and who did we actually want to redeem this?
If the guests redeeming your offer are the same top-frequency regulars who order every week anyway, you haven't created new demand. You've discounted an order that would happen anyway.
Solution: Set redemption goals by guest segment before an offer goes live, not just an overall percentage. Your top guests will always redeem at a high rate, and that's fine. What makes an offer worth the margin is whether it's also reaching new and lapsed guests, the groups that wouldn't have ordered without it.
Track it, then question it
Every metric we’ve discussed is worth tracking—the mistake is stopping at the surface number instead of asking what's actually driving it.
Go back to that sandwich email. If I'd stopped at "orders spiked, so the email worked," I'd have kept running that campaign and crediting it for something an inventory shortage did on its own, when I could have been spending time on something that would move the needle.
The good news: you don't need to overhaul your whole reporting setup. Start with holdout groups on your next campaign, and you'll know a lot more about what's actually driving guest behavior.
Want help putting this into practice? Olo's Professional Services team works with brands every day to build holdout groups, segment guests by frequency, and get more out of Engage. Connect with our team to see what your data is actually telling you.
