The 3 Things D2C Brands That Survived Did Differently
Founder, BTB Audits. $150M+ in ad spend managed across Meta and Google
This is the second half of the UC Berkeley research. The first half quantified the damage: industries with iPhone-heavy audiences saw measurably higher business failure rates after Apple's App Tracking Transparency rollout, as covered in our paired post on the ATT business-impact study. The second half asks a different question: among the brands that survived, what did they actually do differently? Three traits emerged from the 210,000-advertiser, 700,000-campaign analysis, and they are the practical playbook for any D2C brand that wants to be in the survivor cohort during the next privacy round.
What happened
What most operators will get wrong
The popular take on this research is going to be: "Got it. Test more, CAPI is good, install the Pixel everywhere. We do all three. We are safe."
Almost no D2C account I have audited actually does all three. Most do half of one.
Take testing. The Tadelis finding is about brands that experimented "regularly." That means a documented test schedule, a hypothesis written down before the test starts, a measurement plan, and a decision rule. What I see in audits is brands that say they test creative and then point to the fact that they swap out one ad every other week. That is creative refresh, not testing. There is no hypothesis. There is no measurement. There is no learning that compounds. The discipline gap between "we swap out ads" and "we run hypothesis-driven creative tests" is the difference between surviving and not.
Take CAPI. Most accounts I audit have CAPI installed because their e-commerce platform shipped an integration. That counts as "having CAPI" in the registry sense. But when I run the Test Events tool, the events are not deduplicating. The Pixel and CAPI are sending the same purchase as two separate events, which means Meta is over-counting conversions and the optimizer is making worse decisions, not better. CAPI installed without deduplication is worse than no CAPI. The Tadelis finding is about brands with CAPI working, not brands with CAPI present.
Take Pixel coverage. The standard install puts the Pixel on the purchase confirmation page, because that is the conversion event. The brands Tadelis identified as survivors had Pixel events on every meaningful step: product page view, add to cart, checkout initiation, purchase. Multi-step coverage gives Meta's algorithm a much richer picture of the funnel and where users drop off, which means smarter optimization. Single-step coverage gives the algorithm a purchase signal and nothing else, which limits how well it can target.
The lesson is not that the three traits are individually hard. Each one is a known best practice. The lesson is that doing them with discipline is rare. Most operators have a checkbox version of each trait and assume they are covered. The Tadelis data shows the checkbox version does not produce survival outcomes. The disciplined version does.
The brands that win on this update are the ones who treat the three traits as audit subjects, not as completed projects. The brands that lose are the ones who hear "you need testing, CAPI, and Pixel coverage" and say "yes, we have all three" without verifying.
What you should actually do
Run this 3-step audit on your account this week. Each step takes 10 to 20 minutes. None require a developer. All three together identify whether you are in the survivor cohort or the at-risk cohort.
The full audit method for each of these checks lives in the 10-stage Meta ad audit method. The paired ATT business-impact post covers why the survivor cohort matters strategically.
How this changes the audit method
The audit method does not gain a new stage from this data. The Tadelis findings reinforce three existing stages that audit the survivor traits directly.
Stage 2 (signal layer) audits CAPI integration and Pixel coverage. The Tadelis data sharpens the check from "is CAPI installed" to "is CAPI deduplicating correctly." The Pixel coverage check sharpens from "is the Pixel installed" to "are events firing on every funnel step."
Stage 8 (measurement reconciliation) audits whether the brand has documented experimentation as part of how they make decisions. Before Tadelis, this was a soft check that often got skipped. After Tadelis, it is one of the survivor traits, which makes the check non-optional.
Stage 9 (creative and audience health) audits whether the brand has a testing program, not just a creative rotation. The Tadelis distinction between disciplined experimentation and ad-hoc swapping is now the operational definition.
This is the only change to the Meta ad audit method. The 10 stages stay in order. The checks at Stages 2, 8, and 9 update to align with the three survivor traits as defined by the research.
Discipline mindset: before and after
| Aspect | Before | After |
|---|---|---|
| What "testing" means | Swapping out one creative for another every few weeks. Testing different audiences ad-hoc. | Documented hypothesis. Defined holdout or comparison. Written decision based on results. Learnings log that compounds. |
| What "having CAPI" means | CAPI integration is installed in the platform. Checkbox checked. | CAPI installed AND deduplicating with the Pixel correctly. Verified in Events Manager Test Events tool. Both sources recognized as the same conversion. |
| What "Pixel coverage" means | Meta Pixel is installed. Fires on the purchase confirmation page. | Pixel fires on every funnel step (home, product, add-to-cart, checkout, purchase). Each event has the right metadata. Funnel drop-off is visible to the algorithm. |
| What the audit checks across Stages 2, 8, 9 | Is the activity present at all? (Yes or no.) | Is the activity being done with the discipline that produces survival outcomes? (Verified, not assumed.) |
| Risk of misreading the data | Low. The three traits are well-known best practices. Operators reasonably assume they have them. | High if read as "I have these three things." The discipline gap between checkbox compliance and survival-grade execution is the entire point of the research. |
Frequently asked questions
Common questions
About the research
What does "significantly more likely to still be in business" actually mean?
The Tadelis study used statistical analysis of the 210,000-advertiser dataset to control for category, geography, brand size, and other variables. Brands exhibiting all three traits (regular experimentation, working CAPI, multi-Pixel coverage) had statistically significant higher survival rates years into the post-ATT period compared to brands that did not exhibit those traits, after controlling for other variables. The magnitude of the difference was meaningful enough to constitute a survival signal, not a noise-level effect.
Are these three traits causal or just correlated with survival?
The study design (observational data, not a randomized experiment) makes strict causal claims hard to establish. The correlation is real and the direction makes operational sense (better signal hygiene and disciplined testing should improve outcomes). The research team controlled for obvious confounders like brand size and category, which strengthens the causal interpretation. Treat the three traits as evidence-based best practices with a stronger empirical foundation than most marketing advice, not as proven causal levers.
Why these three traits specifically and not others?
The Tadelis team analyzed many variables and identified these three as the ones with the strongest and most consistent correlation with survival across the dataset. Other variables (creative quality, budget size, geographic targeting) also matter but did not show the same level of differentiation between survivors and non-survivors. The three identified traits are operational disciplines that brands can directly control, which is part of why they are useful as a playbook.
What to do next
I do not have time for disciplined testing. Where do I start?
Start with one test per month. One audience test or one creative test, with a written hypothesis (one sentence: "I think variant A will outperform variant B because X"), a defined comparison (same budget, same audience, same campaign structure except the test variable), and a written decision after 14 to 21 days (kill the loser, scale the winner, document what you learned). One disciplined test per month puts you ahead of the median operator. Three per quarter puts you in the survivor cohort.
My platform CAPI integration is installed automatically. Do I need to audit it?
Yes. The auto-installed integrations work correctly for most accounts but break in roughly 25 to 35 percent of audits I run. The break is usually subtle: event_id format changed during a platform update, custom checkout flow bypasses the standard integration, third-party app injected a script that interferes with the Pixel. The Test Events tool in Events Manager is the only way to verify it is working. Audit it once. Document the result. Re-audit annually or after any platform update.
What about brands that survived without all three traits?
Survival in business is rarely deterministic. Some brands without strong signal hygiene survived through luck, category tailwinds, exceptional creative, or strong organic word of mouth. The Tadelis finding is statistical: brands with the three traits had significantly higher survival rates across the population. Individual exceptions exist. The right read is that the three traits stack the odds in your favor, not that the absence of them guarantees failure.
The Tadelis survival research is the rare piece of marketing research that translates directly into an account-level audit. The 3-step diagnostic above is the same one that separated survivors from closures during the last privacy round. Run it this week. If you can show all three traits done with discipline, your account is positioned for whatever the next round brings. If you cannot, the fix is small and the downside of skipping it is not theoretical.
If you don't have four to six hours, or you want a second pair of eyes that's managed $150M+ across Meta and Google, the Free Quick Scan is what I built for that. I'll record a private 5 to 7 minute Loom walking through the leaks I find on your account using public data only. You'll have it in 48 hours.
Get Your Free Quick Scan →Keep reading on Meta survival and discipline
Aditya Chaturvedi is the founder of BTB Audits. He has managed $150M+ in ad spend across Meta and Google for DTC, SaaS, and lead-gen brands ranging from $10K per month to $500K per month. Industry Updates from BTB Audits cover platform changes and what they actually mean for operators, not what the headlines say they mean. Read more on the BTB Audits blog.