Binghott recommends full Meta Incremental Attribution switch for ad accounts

• A nine-campaign setup became one $1.5M CBO campaign, and binghott's main CBO thread says cost per conversion fell 5% while spend scaled.
• The more interesting proof was breadth: binghott's June Ads Manager screenshot showed $823,110 in spend, a top ad at $80k, the 30th ad still above $5,800, and 200+ ads above $1,000.
• His Incremental Attribution migration claim was all-or-nothing: his IA switch reply said overlapping old and new optimization types can distort the test.
• Cost caps were his manual control surface, not a rigid CPA target, and one cost-cap reply framed very high cost caps as close to highest-volume bidding.
• The caveats were concrete: his main thread said CBO optimizes to cost rather than profit, can fight guaranteed spend, needs volume, and resets learning during consolidation.

Meta's Andromeda engineering post explains the machine side: retrieval narrows tens of millions of ad candidates to a few thousand before later ranking stages decide what a person sees. A 2026 Incremental Attribution explainer describes IA as counterfactual modeling trained on Conversion Lift data, aimed at conversions caused by ads rather than every post-exposure conversion. In his Ads Manager post, binghott added the operator layer: $823,110 in June spend, 1,532 visible ad rows, and hundreds of ads left alive for Meta to choose from.

Binghott said he has spent 18+ years working on Facebook ads and has studied billions of dollars of ads, then gave a current account example: nine campaigns in March, one CBO campaign by the last 30 days, $1,508,114.36 in spend, and a $50,000 daily budget.

The operating claim was blunt: less manual media buying, more work on creative, product, offer, copy, CRO, and landing pages. Meta's Horizon ads documentation says ad sets enter a learning phase and points to 50 events over 7 days as the calibration target, which matches his argument that consolidation gives Meta more signal.

He also tied the setup to Incremental Attribution, saying IA tells Meta to optimize toward conversions it actually drove rather than every last-touch conversion it can claim in the same thread.

The June screenshot was the cleanest counterexample to the usual CBO complaint that Meta dumps spend into one or two ads.

His numbers:

• Total June spend: $823,110.02.
• Top ad spend: about $80,000.
• 30th highest-spending ad: $5,800+.
• Ads above $1,000 in June spend: 200+.
• Visible table count: 1,532 rows.

The three highlighted ads looked ugly in June, with about $30k in combined spend and high CPAs. In July, binghott said those same three ads had spent $94, because Meta throttled them without him turning them off in the screenshot thread.

That is the workflow reveal: his bench stays live, and Meta gets to stop spending on stale ads without deleting them from the option set.

Binghott's rollout answer was not gradual. He told one commenter that rolling IA out was "part of the problem" and said to do the full changeover in one go in a migration reply.

His reason was overlap. If one user can be shown ads from both old and new optimization types, the account is not testing IA in isolation. He put the same point more directly later: advertising is not happening in a vacuum, and the existence of one setup changes the other in a follow-up reply.

He also asked whether the account had been fully optimized for IA before judging the result in another reply. That leaves the test definition narrower than most rollout debates: pause old stuff, turn on the new IA setup, then judge the system that actually ran.

Cost caps were the steering wheel in the thread, but not in the usual "set the CPA target and pray" sense.

Binghott's mechanics:

• High caps: a very high cost cap behaves similarly to highest-volume bidding, according to his cost-cap reply.
• Walking caps down: one reply said CPA should fluctuate less and spend should fluctuate more as caps approach the desired CPA.
• Distribution control: his cost-cap framing described caps as a way to buff or nerf ad set distribution.
• Rigid caps: another reply said caps set too close to target can block spend needed to find new audiences.
• Bid caps: his bid-cap reply said he had not used bid caps seriously in years because they felt too rigid.

The clearest line came later: he said he does not relate cost caps directly to actual CPA and instead uses them as spend and prioritization levers in a cost-cap reply.

The CBO setup depended on creative supply. When one commenter asked about structure with too few assets, binghott's answer was that the account would need more ads fast in a creative reply.

The ad-set pattern was simple:

• Put a chunk of decent existing ads into one ad set per one setup reply.
• Use a new ad set per batch of new ads, with min budgets if needed per another reply.
• Keep adding as more ads arrive, as one reply described a setup with nine ad sets and counting.
• Let conversion volume decide density: his ad-count reply said low conversions mean fewer ads, while high conversion volume can support many.

He also said one lower-spend brand had more than 1,000 ads live in a reply about live-ad count. That number explains why the $823k screenshot was not just a budget flex. It was a creative-liquidity argument.

Binghott's own example was not a messy multi-SKU retail catalog. He said the account was financial services in one reply, and the main thread said the business had basically one main product, which made one-campaign consolidation cleaner in the CBO thread.

His caveat list for CBO:

• It optimizes cost, not profit.
• It can push spend toward cheap, low-value conversions when margins differ.
• It can fight guaranteed spend for a new product or geography.
• Low-volume accounts give it less to learn from.
• Consolidation can reset learning when an account leaves a fragmented structure.

For multi-SKU accounts, binghott said the strategy can still work when prices are similar, but cost caps may need to become ROAS bidding or separate conversion events by SKU/category in his multi-SKU reply. For clothing, his catalog reply favored catalog ads inside the CBO but separating existing customers for cleaner control.

Binghott repeatedly treated reported CPA and ROAS as partial views, not truth. He said in-platform and third-party CPAs can both mislead in different ways in one attribution reply.

The term he kept returning to was breakdown effect. In one reply, he said people call something "best performing" based on CPA or ROAS without understanding scale or volume in the breakdown-effect reply. In another, he called breakdown effect "a hell of a drug" when a commenter reacted to the spend table in a short reply.

That is why his high-CPA ad argument stayed probabilistic. He said some upper-funnel ads may bring new people into the funnel for other ads to convert, but he also admitted he cannot prove that hard because advertisers cannot see every on-platform interaction before a site visit in the June screenshot thread.

The thread's funniest evidence was not about Meta making mistakes. It was about media buyers making them.

Binghott listed ordinary operator errors: fat-fingered budgets, mislabeled ads, wrong URLs, wrong audiences, over-spend, and under-spend. Then he said he had recently seen experienced media buyers run a campaign they believed was targeting existing customers while it was excluding them, and label ad sets as Incremental Attribution while they were actually running 7dc1dv.

That fed his larger split: Meta can mess up, and visible account data can also lure humans into overconfident rules. He said Meta sometimes looks wrong while doing something useful in a reply about imperfect setups, which is exactly the discomfort his CBO workflow asks advertisers to tolerate.

The product pitch at the end was a monitoring tool for the stuff outside Ads Manager. Binghott said URL Love It tracks landing pages, website search results, and competitors for changes that quietly wreck or boost ad and business performance in the URL Love It post.

He tied that pitch to a specific failure mode: people stare at the ad account while the real performance change came from a landing page update, product launch, price change, broken checkout, competitor promo, holiday, or news cycle in the main CBO thread. The waitlist offer in the product post promised prelaunch pricing and a first month for $1.