Most marketing channels make you take their word for it. Impressions, clicks, and view-through conversions all point at activity, then ask you to believe it turned into sales. Card-linked offers are different, because the reward fires on a verified transaction. That gives you something rare: a channel where you can measure real return instead of inferring it.
But a clean signal is only useful if you measure the right things. Plenty of brands run card-linked offers and report a return that looks great on paper while telling them almost nothing. This is a framework for measuring card-linked offer ROI in a way that actually holds up when someone asks whether the channel is working.
Start with the question you are actually answering
Before any math, get clear on what you want to know. 'Did this campaign make money' breaks down into three separate questions:
- Did the offer drive sales that would not have happened otherwise?
- Did it bring in customers you did not already have?
- Was the return worth the spend compared to other channels?
Each needs a different measure. Reporting a single blended ROI number without separating these is how marketers end up defending a channel they cannot actually explain. The framework below takes them in order.
The three measures that matter
1. Incrementality
Incrementality is the foundation, and it is where most measurement goes wrong. It is the sales your offer actually caused, not the sales that would have happened anyway.
The trap is easy to fall into. You run a cash back offer, sales come in attributed to it, and you count all of them as wins. But some of those buyers were going to purchase regardless. Rewarding them did not create a sale. It just moved margin from your pocket to theirs. That is not ROI. It is a discount you gave to existing demand.
To measure incrementality properly, you need a comparison. The cleanest approach is a holdout: a portion of your target audience that is eligible but does not receive the offer. The difference in purchase behavior between the exposed group and the holdout is your incremental lift. That lift is the number for your ROI calculation. Gross attributed sales are not.
If you cannot run a holdout, comparing sales before and after the campaign against a similar baseline can get you close, though less precisely. The principle stays the same: measure against what would have happened without the offer. Our Complete Guide to Incrementality Testing for Commerce Media walks through the mechanics in more detail.
2. New-to-brand rate
The second measure asks who the offer reached. A campaign that only rewards existing loyal customers is very different from one that brings in new buyers, even if both report the same attributed revenue.
New-to-brand rate is the share of offer redemptions that came from customers who had not purchased from you in a defined window, often the past 12 months. Because card-linked offers work from transaction data, you can see this directly rather than guessing.
This number reframes the whole ROI conversation. A high new-to-brand rate means the offer is doing acquisition work, and those customers carry future value beyond the first purchase. A low rate means you are mostly paying to reward people who were already yours, which may still be worth it for retention, but you should know that is what you are buying.
3. Verified-transaction attribution
The third measure is what makes the first two trustworthy. Card-linked offers attribute on the transaction itself, not on a click or a cookie that may or may not have led anywhere.
This matters because it removes the guesswork that undermines most channel measurement. There is no last-click debate, no cross-device gap, no view-through assumption. A reward pays out when a real purchase posts to a real card, so the sale you are counting is a sale that happened. For the fuller argument on why transaction data beats click-based tracking, see The Case for Pay-for-Performance.
The practical benefit is that your ROI inputs are grounded in verified events. When incrementality and new-to-brand rate are built on real transactions rather than modeled conversions, the number you report can survive scrutiny.
Putting it into a calculation
Once you have the three measures, the ROI math is straightforward. The key is to feed it incremental figures, not gross ones.
A simple version:
Incremental ROI = (incremental revenue − campaign cost) / campaign cost
Where:
- Incremental revenue is the lift you measured against your holdout, not total attributed sales.
- Campaign cost includes the reward funding plus any platform or media fees.
For a fuller picture, layer in the value of new customers. If your new-to-brand rate is high, a share of your acquired customers will purchase again, so a first-purchase-only ROI understates the real return. Bringing expected repeat value into the calculation, even conservatively, gives a truer read on what the channel is worth.
The discipline is to keep incremental and gross numbers clearly separated. Gross attributed sales are fine for a directional read, but the moment you are making a budget decision, incremental is the number that counts.
Measurement timeframes and attribution windows
The three measures above tell you what to count. Timeframes and attribution windows tell you when to count it, and getting this wrong quietly distorts every number.
An attribution window is the period after a customer sees or activates an offer during which a purchase still counts as driven by that offer. Set it too short and you undercount slow-moving purchases, making the channel look weaker than it is. Set it too long and you start crediting the offer for purchases it had nothing to do with, inflating your return. There is no single correct window. It depends on your buying cycle.
A few principles to set it sensibly:
Match the window to your purchase cycle. A quick-serve restaurant or a convenience purchase has a short cycle, so a window of a few days to a couple of weeks usually fits. A considered purchase like electronics, furniture, or travel has a longer cycle, so a window of 30 days or more may be appropriate. The right window reflects how long your customers actually take to buy, not a default someone picked for you.
Keep the window consistent across campaigns. If you want to compare two campaigns or two channels, they need the same window. Changing it between measurements makes the comparison meaningless, even when both numbers look precise.
Separate the measurement period from the campaign period. A campaign that runs for four weeks needs measurement to continue past the end date so purchases inside the attribution window still get captured. Cutting measurement off the day the campaign ends undercounts late conversions and understates ROI.
Give repeat value its own, longer horizon. First-purchase ROI and repeat value operate on different clocks. Measure the incremental first purchase inside your attribution window, then track repeat behavior from new-to-brand customers over a longer horizon, often a quarter or more, to see the fuller return. Reporting both, clearly labeled, is more honest than forcing them into one number.
The theme across all of this is consistency. Pick windows that match your buying cycle, hold them steady, and label what period each number covers. Precise-looking ROI built on shifting timeframes is one of the easiest ways to mislead yourself.
Common measurement mistakes to avoid
Counting all attributed sales as incremental. The most common error, and the one that inflates ROI the most. Without a holdout, you are almost certainly overstating return.
Ignoring new-to-brand mix. Two campaigns with identical ROI can have very different strategic value depending on whether they acquired customers or rewarded existing ones.
Comparing against the wrong benchmark. A card-linked offer's real competition is your other performance channels. Measuring its incremental return against, say, an affiliate program on the same incremental basis is far more useful than comparing it to a gross number from a channel that cannot isolate incrementality at all.
Judging too early. Repeat value takes time to show up. A campaign that looks break-even on first purchase may be clearly positive once repeat behavior from new customers lands.
Why this framework favors card-linked offers
None of these measures are unique to card-linked offers in theory. Any channel would love to report clean incrementality, new-to-brand rate, and verified attribution. The difference is that card-linked offers can actually deliver the data to support all three.
Because the channel runs on transaction data, you can build a holdout, see who is new, and attribute on real purchases. Most performance channels can do one of those at best. That is what makes card-linked offer ROI not just measurable, but defensible, which is exactly what you need when the budget conversation gets serious.
The takeaway
Measuring card-linked offer ROI well comes down to three things: isolate incrementality with a holdout, track your new-to-brand rate, and trust attribution because it runs on verified transactions. Feed incremental figures into your ROI math, layer in the future value of new customers, and compare against your other channels on the same basis.
