Rajeev Nair, Co-founder and Chief Product Officer at Lifesight, had a Q&A with Greg Kihlström for The Agile Brand Guide’s Expert Mode series and this article is drawn from that interaction.
When third-party cookies started going away, first-party data got positioned as the replacement, and most of us budgeted like it was. Loyalty programs, logged-in experiences, CDPs, clean rooms. The reasoning was simple enough: if you own the customer relationship, you own the data, and if you own the data you can finally see what’s working.
Rajeev Nair, co-founder and Chief Product Officer at the measurement platform Lifesight, thinks that last step doesn’t follow. Owning the data tells you who to reach. It says almost nothing about whether reaching them changed anything. And he was appointed to the IAB’s Measurement Board this past June, which means he’s arguing this position inside the body writing the standards, not just from a vendor’s blog.
Owning the Data Solves Targeting, Not Causality
The gap Nair describes is the one between a good audience and a proven result. Asked to make the case that first-party data isn’t the answer marketers treated it as, he drew the line plainly.
“First-party data solves targeting, not causality. A retailer that knows a customer buys running shoes every six months can build a far more relevant audience without a third-party cookie, and that is a real, durable advantage. But knowing who to target isn’t the same as knowing whether the ad worked. If that customer buys running shoes after seeing an ad, the purchase may well have happened anyway since the runner needed new shoes. Marketers need to know whether their media actually drove the customer to buy the shoes to prove that spending on the advertising was worthwhile.”
The runner is the whole problem in one person. She was going to buy shoes in month six whether or not she saw the ad, and every system in the stack will credit the ad anyway. Nair also points to a second blind spot, which is arguably worse because it runs the other direction. First-party data is built from transactions and logged-in behavior, so it has nothing to say about the CTV spot that put someone in the market weeks earlier. “A purchase can look like it came from nowhere in a first-party dataset,” he notes, when it actually started somewhere the retailer’s data never touches.
For those of us setting budgets, that cuts both ways at once. The channels closest to the transaction get credit they didn’t earn, and the channels that created the demand get none. Shift money on that basis for four consecutive quarters and you’ve quietly defunded the top of your own funnel while your dashboards improved.
The $4 Million You Can’t Interrogate
The money here is not small or theoretical. Walmart reported its second quarter on August 20, and the split inside those numbers is stark: Walmart Connect grew 43% in the U.S. excluding VIZIO, while Walmart U.S. comparable sales grew 2.6% excluding fuel. Global advertising was up 38%. Membership income rose 17%. The retail business is growing at a normal retail pace and the ad business attached to it is growing more than fifteen times faster.
So what does a brand buying into a network like that actually get to see? Nair walked through it with a round number.
“Say a brand spends $1 million with a retail media network and the network reports $4 million in attributed sales. The brand can see the impressions, clicks, conversions and resulting ROAS. It may also see which products sold and which audiences converted.
What the brand doesn’t necessarily get is a clean view of the $4 million it would have generated without the $1 million investment. Some customers may have already been planning to purchase, some may have converted through another channel and some sales may have shifted from a different retailer or campaign. The $4 million is an attribution number, but the CMO ultimately needs to know how much additional revenue the $1 million created.”
That last sentence is the one to take into your next planning meeting. Attributed sales and incremental sales are different quantities, and a retail media network has no commercial reason to hand you the smaller one. Nobody is lying to you. The reported number is real — it’s just answering a question you didn’t ask, and the gap between the two is where renewal decisions get made badly.
Where Independent Measurement Gets It Wrong
Nair sells independent measurement, which makes the obvious rejoinder that he’s one vendor’s model grading another vendor’s model. Asked to argue against his own category, he didn’t dodge it.
“Incrementality testing isn’t automatically more trustworthy. A geo test can fail if test and control markets don’t behave similarly to begin with, or if the campaign’s too small to produce a clean read. Marketing mix modeling has different challenges. If a brand spends more on retail media every time it increases paid search, for example, it can be difficult to isolate which channel drove the outcome.”
His resolution is a sentence worth stealing: “The measurement methodology should match the decision.” Retailer attribution is fine for the questions it’s built for — which creative pulled better, which audience responded. Nair’s position is that a major budget decision needs an independent read on incremental revenue plus corroboration from more than one method before the money moves.
Practically, that means two tiers rather than one tool. Let the network’s own reporting run your week-to-week creative and audience calls, and don’t pay someone else to re-litigate them. Reserve the independent read for the decisions with a comma in them. It’s a less exciting answer than “replace your attribution,” and it’s cheaper.
Personalized Pricing Turns the Problem Into Arithmetic
The measurement gap just got a regulatory neighbor. On August 19 the FTC voted 2-0 to seek comment on a proposed enforcement policy statement on personalized pricing, with comments closing September 18. The Commission’s concern is disclosure — whether shoppers are told their data set the price — and it acknowledged it can’t ban the practice outright. That’s a consumer-protection question, not a measurement one. But Nair’s point is that a retailer answering the FTC still won’t know whether the pricing worked, and the test for that is not complicated.
“A retailer could take 100,000 customers eligible for a personalized offer and randomly give the offer to half of them, while the other half see the standard price, then compare conversion, revenue and margin between the two groups.
Suppose the personalized group converts at 8% and the control group at 7.5%. The retailer created a 0.5 percentage-point lift, but it also gave a discount to the other 92% of customers who didn’t convert. The business needs to understand whether the incremental sales and margin generated by the offer outweighed the revenue it gave up through discounting.”
Note what that arithmetic does to the usual case for personalization. A lift that reads as a win in a conversion report can be a loss on the P&L, because the discount went to everyone in the group and the extra sales came from a sliver of it. Most personalization business cases we see are built on the conversion number alone. If yours is, the margin column is the one to go look at first, and it’s a holdout design of exactly the kind Nair describes that produces it.
The through-line is that better data made us more confident without making us more correct. A CMO who wants to fix that on Monday doesn’t need a new platform. Nair’s suggestion is to pick the largest retail media investment that’s up for renewal, withhold it from a set of comparable markets for a few weeks, and compare. Then go back to the partner and ask them to “separate attributed sales from incremental sales and provide the methodology behind each number.” Both numbers, with the math shown. If the answer is that they can only produce the first one, that is itself the finding, and it’s worth knowing before the next renewal rather than after it.
