Most design ROI decks are built to win a meeting, not to survive the follow-up question. The move is almost always the same. Revenue rose 20 percent in the two quarters after the redesign shipped, so the headline slide says the redesign drove a 20 percent revenue lift. It reads great for about nine seconds, right until the CFO asks what else changed in those two quarters. That question ends the deck, because the real answer is that a lot changed, and the slide was built to pretend none of it did.
So here is the stance. Attributing a revenue lift entirely to a redesign is the fastest way to lose a CFO’s trust, and that trust is worth more than any single budget you were trying to win with the slide. A credible design ROI deck does the opposite of the confident one. It names three things on purpose: what design moved, what design did not move, and what you genuinely cannot separate from everything else. The third column is the one most decks delete, and it is the one that earns you the room.
I have built these decks from both sides of the table. I led an AI-first design transformation inside a multi-billion-dollar eCommerce org, ran weekly A/B tests while rebuilding a store on Shopify, and now I measure my own bets at Story Genie, the company I founded. The decks that held up under real questioning were never the ones with the biggest number on the cover. They were the ones that told the truth about where the number stopped being defensible.
The attribution game, and why finance sees through it
A redesign almost never ships into a quiet room. In the same window, marketing changed the ad spend, sales hired two people, pricing ran a promotion, a competitor stumbled, and the season turned. Every one of those moves revenue. So when a deck draws a straight line from a redesign to a revenue number, a finance partner does not see proof. They see a person claiming one variable out of ten and hoping nobody counts the others. The instinct behind the overclaim is understandable. Design has been underfunded for years, and a big number feels like overdue credit. But an inflated number does not fix that. It confirms the suspicion that design cannot be measured cleanly, and it takes the whole deck down with it.
It helps to be clear-eyed about the best evidence design has, because even that evidence is a correlation, not an attribution. McKinsey studied 300 public companies over five years and found the top quartile on its Business Value of Design index (opens in new tab) grew revenue 32 percentage points faster than their peers and delivered 56 points higher total returns to shareholders. That is a real finding and I cite it often. It is also a statement about which kinds of companies tend to win, not proof that a specific redesign caused a specific quarter. If I will not stretch McKinsey’s number into a claim about my project, I have no business stretching my own quarter’s revenue into one either.
Attributing a revenue lift entirely to a redesign is the fastest way to lose a CFO’s trust.
Column one: what design actually moved
The first column is the part you can defend without flinching, and the bar to get in is a clean line from the change to the number. The gold standard is a controlled experiment, because it is the only method that isolates one cause. Show the old design to half the traffic and the new design to the other half, at the same time, under the same ads and the same prices, and the only thing different between the two groups is your work. Whatever gap opens between them is yours to claim, and you can claim all of it.
The size of the number is not what makes it credible. The isolation is. Ron Kohavi and Stefan Thomke described a case in Harvard Business Review (opens in new tab) where a Microsoft engineer changed how Bing displayed its ad headlines, a small formatting tweak nobody expected much from, and the change "increased revenue by an astonishing 12%," worth more than 100 million dollars a year in the United States alone. A 12 percent revenue swing from a formatting change sounds like exactly the kind of claim a CFO would laugh out of the room. They believed it because it ran as an experiment against a control, so nothing else could have produced the gap. That is the whole point of column one. You are not selling a big number. You are selling a clean one.
This is also the discipline I run when the stakes are real. Rebuilding a store on Shopify, we tested weekly, one clear change at a time, so every win pointed at a single cause we could name. The serious version of this scales up rather than down. Booking.com, as Stefan Thomke documented in a later HBR piece (opens in new tab), runs some 25,000 tests a year, which is what it looks like when a company decides to actually know why its numbers move instead of guessing after the fact. You do not need 25,000 tests. You need the discipline to put nothing in column one that a controlled test, or a clean before-and-after with nothing else moving, cannot support.
Column two: what design did not move
This is the column that feels like career suicide and is actually the opposite. You volunteer the misses. The metrics that stayed flat. The feature you were sure would lift activation and it did not. The other forces running that same quarter, spelled out by name: the promotion, the ad push, the seasonal swing, the new sales hire. It is tempting to hide this column because it looks like weakness. Finance reads it as the first straight thing a designer has ever shown them.
The mechanism is simple and a little counterintuitive. When you name what design did not move, every claim in column one gets more believable, not less, because you have proven you are willing to report against yourself. A deck that contains only wins is a sales pitch, and everyone grades a sales pitch on a curve. A deck that reports its own flat lines is a measurement, and people trust measurements. I would rather walk in with three defensible wins and two named misses than five wins nobody in the room believes.
Column three: what you cannot separate
The third column is the one almost every deck deletes, and deleting it is the tell. Some of the most valuable design work does not survive a clean experiment. Brand, trust, the feeling a product gives someone in the first five seconds, the slow compounding of long-term retention. You cannot run a tidy two-week test on whether people trust you more, and if you pretend you can, you will invent a number and someone will eventually pull the thread. The straight move is to put this work in its own column and label it plainly: real, valuable, and not cleanly attributable.
You can still bring evidence here. It is just directional, not proof. Track repeat behavior, referrals, branded search, the words customers use to describe you when nobody is asking. At Story Genie, I let people read the entire finished book before they pay a cent, which runs against every rule of conversion design. The value of that choice is trust, and trust is the product, not a line I can isolate in an A/B test. So I say exactly that in the deck. Here is what I measured, here is what I believe is working that I cannot cleanly measure, and here is why I still think it matters. Naming the limit is what makes people believe the parts I do claim.
The credible deck is not a humble one. It is an accurate one, and accuracy is what compounds into trust.
Who this deck is for, and when it is the wrong one
This deck is for anyone whose budget, headcount, or seat at the table depends on a number a finance partner has to sign off on. If that is you, build the three columns and resist the urge to collapse them into one triumphant figure. The confident single-number deck wins the first meeting and loses the relationship. The three-column deck occasionally loses the first meeting and wins every one after it, because you become the person in the building whose numbers hold up under questions.
There is a case where this is the wrong deck, and I want to be clear about it. If you genuinely ran a clean experiment and the result is unambiguous, do not perform false modesty. Take the full credit, state it plainly, and move on. Accurate attribution is not a habit of shrinking every number. It is a habit of claiming exactly what you can defend, no more and no less, which sometimes means claiming a lot. The three columns are not there to make design look small. They are there to make design look accurate, and accurate is the only reputation worth having with finance.
It is also the wrong deck for an internal hype moment where nobody is scrutinizing the math, or for a pre-product-market-fit startup still changing so fast that a formal ROI case is effort spent measuring something that will not exist next quarter. Before there is a stable business to move, move fast and skip the accounting. The deck earns its keep once the numbers are real enough that someone with a spreadsheet is going to check them.
If you want the mechanics under all of this, I wrote the full method in how to measure the ROI of design and the presentation side in how I prove design moved the business. The short version is the one sentence I would put on the last slide of any credible deck. Design moved these numbers and we can prove it, design did not move these and here is what did, and this last set matters even though we cannot separate it cleanly. A CFO has heard a hundred decks claim everything. The one that also says what it cannot claim is the one they remember, and the one they fund again.