The dashboard and the cash register now disagree by a measured amount
Common Thread Collective’s Q1 2026 Channel Mix Benchmark measured 299 DTC brands, $1.01 billion in revenue, and $231 million in paid spend from January through March 2026, comparing what each channel reported against what incrementality testing says each channel caused 1. The headline pair: Google Brand search reported a 19.07x platform ROAS and delivered 5.72x incremental, while Meta acquisition reported 1.83x and delivered 2.07x 1.
The platform ROAS vs incremental ROAS difference runs in both directions, and that is the finding that matters. Branded search overreports by roughly 3x. Meta prospecting underreports. If budget decisions run on dashboard ROAS, money flows to the channels best at claiming credit, not the channels best at creating customers.
The incrementality factor, in one line
The incrementality factor is incremental ROAS divided by reported ROAS. Below 1, the dashboard flatters the channel. Above 1, it undersells it.
Here is the Q1 dataset through that lens 1:
- —Google Brand search: 19.07x reported, 5.72x incremental, a 0.30x factor. Nearly 70% of the reported value was capturing demand that already existed.
- —Retargeting (Meta non-acquisition): 5.88x reported, 3.53x incremental, a 0.60x factor, on just 2.65% of aggregate spend.
- —Snapchat: 6.18x reported, 1.85x incremental. Less than a third of the dashboard number survived testing.
- —Meta acquisition: 1.83x reported, 2.07x incremental, a 1.13x factor. The dashboard understates the channel.
Read the last row twice, because it is the one operators get wrong. The channel with the least impressive dashboard number in the set is the one whose dashboard number is a lie in the flattering-to-you direction.
Why the gap exists, in both directions
Attribution claims conversions that would have happened anyway. The person who clicks a branded search ad was already searching for the brand by name; the person who clicks a retargeting ad was already on the site yesterday. Both clicks are real. The causation mostly is not, which is why the two worst factors in the dataset belong to exactly those two tactics.
In the other direction, broad prospecting drives conversions attribution never connects: the impression seen and never clicked, the cross-device purchase, the order that lands after the attribution window closes. Meta acquisition took 58.7% of all ad dollars measured, with a median per-brand allocation of 72%, and the dashboard still under-credited it at that 1.13x factor; Google non-brand held 25.6% of dollars and Google brand 7.7%, against a median MER of 4.23x across the portfolio 1.
The structural point sits underneath the numbers: the channels that look best in a dashboard are systematically the ones doing the least causal work, because ease of attribution and ease of causation are different properties. Branded search is trivial to attribute and mostly non-causal. Broad prospecting is causal and hard to attribute. Any reporting system built on click paths will rank them backwards.
The firm’s own data already showed both halves of this finding
The firm’s audit and holdout data made both of these arguments before the external benchmark existed, and the portfolio-scale numbers now say the same thing.
Across the accounts the firm has audited, 50 to 70% of credited retargeting revenue turned out to be non-incremental in structured pause tests. CTC’s 0.60x retargeting factor is the same finding, measured across 299 brands instead of one firm’s book 1.
Firm holdout tests on branded search defense put its incremental contribution at 18 to 32% of reported volume. CTC’s 0.30x brand search factor lands inside that range 1.
Neither read needed updating. What changed is that neither one rests on the firm’s sample alone anymore.
How to apply this without enterprise tooling
Read CTC’s factors as priors, not gospel. Your account’s factors will differ with brand strength, category, and channel mix; the stronger your organic brand demand, the more of your branded search clicks would have happened anyway, and the lower your factor sits. The direction is what transfers, not the decimals.
Then test where the priors say the dashboard flatters most, which for nearly every account means branded search and retargeting first. The method does not require a measurement vendor: geo holdouts and structured pause tests fit inside a normal month of operations, and running them is standing work inside the firm’s management engagements. A two-week branded search holdout in a handful of markets costs almost nothing and replaces a dashboard fantasy with your own number.
The output of that work is a per-channel multiplier you apply to reported ROAS before any budget meeting. Once it exists, reallocation decisions stop being arguments about whose dashboard to believe and start being arithmetic.
The uncomfortable version of this finding is not that dashboards are wrong. It is that they are wrong in a consistent direction, so every quarter of budget allocation run on reported ROAS quietly transfers money from the channels creating customers to the channels claiming them.