The Numerator Nobody Audits
Here is the thing every media buyer knows and few connect to their own ROAS: the attribution window and the view-through setting decide how many conversions your spend is allowed to claim. Set a 1-day click window and only conversions within a day count. Set a 28-day click-and-view window and a purchase four weeks later, from someone who merely saw the ad, gets pulled in too. Same campaign, same spend, same actual sales. The wider rule simply credits more of them to the ad.
So when your ROAS climbs after a settings change, an account restructure, or a platform default update, the first question is not "what did I do right." It is "did the numerator just get permission to reach further." Most of the time, that is exactly what happened.
A rising ROAS feels like a better campaign; often it is only a wider window and view-through credit pulling more of the same sales into the numerator.
Watch the Numerator Climb on Identical Traffic
Run one campaign through progressively wider counting rules and watch the credited conversions rise while the real business stays flat. I want to be clear this is an illustration of the mechanism, not a measurement of a specific account, but the shape is real for anyone who has ever changed a window and watched the dashboard reward them for it.
Illustrative: one campaign, one spend, one set of real sales, read through four counting rules; the reported ROAS more than doubles without a single extra unit sold.
The Math of a Borrowed Number
Now the part that turns a reporting quirk into a spending problem. If you widen the window and your ROAS jumps from 1.9x to 4.1x, and you treat that as a win, you scale. You pour more budget into a campaign whose real, same-day, click-earned return never changed. The borrowed conversions do not repeat at scale, because they were never incremental in the first place; they were sales that would have happened anyway, or impressions taking credit for other people's work. So your reported ROAS holds up on the dashboard while your actual blended margin quietly sinks, and you cannot see why, because the number you trusted was inflated by the very setting you never audited.
It gets worse when you compare. A buyer bragging about 4x on a 28-day click-and-view window is not describing the same achievement as a buyer reporting 4x on a 1-day click window. The second number is several times harder to earn. Comparing two campaigns, two trackers, or two months without matching the window is comparing two different measuring sticks and calling one of them better.
Is Your ROAS Real or Borrowed?
If a settings change moved the number, it was borrowed; if the window held and the credit is mostly click-through, it is a return you can actually scale against.
Why This Matters More in 2026
Two shifts have made the borrowed numerator easier to fall for. The first is that platforms keep changing their default windows and their attribution models, so your reported ROAS can move between reporting periods without you touching anything, and it is tempting to read that drift as performance. The second is that value-based and view-through optimization have both grown: platforms increasingly want to optimize toward a wider set of credited conversions, which means a generous window does not just flatter your report, it steers the algorithm. Feed the machine a numerator padded with late view-through conversions and it learns to chase the impressions that pad it, not the clicks that pay you.
The buyers who feel this most are the ones on impression-heavy sources, native, push, display, where view-through credit is abundant and the gap between reported and incremental return is widest. If you run tight, click-driven search or social with a short window, your reported ROAS is close to your real one. The wider your window and the more view-heavy your traffic, the more of your number is on loan.
What Good Looks Like
A ROAS you can trust has three properties. It is measured on a fixed window you chose to match your real sales cycle, so it does not drift when a platform changes a default. It separates click-through credit from view-through credit, so you know how much of the number required a real action versus a mere impression. And it is anchored to conversions your own tracking can verify end to end, not just the ones the platform decided to hand you. Pick a window and hold it, weight the credit you can verify, and treat every jump in the number with the suspicion it deserves until you have ruled out the counting rule.
So What Do You Do About It
Stop reading a rising ROAS as a win until you have checked whether the window moved. The number on the dashboard is a fraction, and the numerator is the easiest thing in your whole account to inflate without selling anything. The fix is to fix your measuring stick: choose an attribution window that matches your sales cycle, hold it steady across periods so your comparisons mean something, and keep click-through and view-through credit separate so you always know how much of the number you actually earned. That discipline is part of why I built ClickerVolt to anchor every conversion to a persistent click identifier and forward it server-side, so the credit you can verify stays distinct from the credit the platform assigns itself. See how click-anchored tracking keeps your ROAS honest.
Even if you never touch it, do one thing this week: take a campaign whose ROAS you are proud of, and pull the same date range on a 1-day click window with view-through off. Set that number beside the one you have been reporting. The distance between them is how much of your ROAS was borrowed from a longer lookback and from impressions nobody clicked.
This piece describes how attribution windows and view-through credit shape reported ROAS. The 1.9x-to-4.1x progression and the specific multiples are illustrative examples to show the mechanism, not a measurement of any specific account.
