Rising CAC Is a Targeting Problem, Not a Budget Problem

Rising CAC is the symptom of three forces stacking at once: every advertiser in your category bids on the same broad lookalikes, ad platforms lost most of their cross-site tracking after cookie deprecation and ATT, and the pixel that once found in-market buyers now optimizes against a thinner, noisier signal. When the inputs degrade, the auction does what auctions do, it charges more for worse matches. You can throw budget at this, but you are paying a premium to reach the same crowded pool everyone else is reaching.

The teams who escaped rising CAC did not find a cheaper channel. They changed the input. Instead of asking Meta or Google to guess who is in-market from a degraded pixel, they fed the platform a list of accounts they already knew were in-market, resolved from their own website traffic and behavior. The auction stops guessing and starts matching against truth. That is the difference between renting a platform's deteriorating model and steering it with first-party signal you own.

Lookalike and broad-match targeting were built for an era when platforms could observe rich behavior across the open web. Post-ATT and post-third-party-cookie, the seed data is sparser and the modeled expansion is mushier. So a 1 percent lookalike in 2026 is a far blunter instrument than it was in 2020, and because the modeling is opaque, you cannot see that it has decayed until your CAC creeps up quarter over quarter. Everyone in your vertical is feeding similar seeds into similar models, so you converge on overlapping audiences and bid each other up.

There is a structural fix hiding in plain sight. Your own properties still generate first-party signal that no platform can take from you: who visited pricing, who returned three times this week, which accounts opened the product comparison. Tools like RB2B and Snitcher resolve a meaningful share of that anonymous traffic into named companies, and Koala or Warmly score the intent. That resolved, scored set is the seed your competitors do not have, because it is generated by your own audience and your own funnel.

Treat first-party signal as two ad layers, not one. The targeting layer is a custom audience built from accounts your system has resolved as in-market, enriched in Clay with firmographics and contacts, then pushed into Meta Ads, LinkedIn Ads, and Google Ads as a matched audience. Instead of spraying a 2-million-person lookalike, you spend against a few thousand accounts that already raised a hand on your site. CPMs may rise on a smaller audience, but match quality rises far faster, so cost per booked call falls.

The suppression layer is the half most teams skip and it is where the fastest CAC win lives. Upload current customers, open opportunities, and recent closed-lost as exclusions so you stop paying to re-acquire people you already have or already lost. Then suppress accounts that just converted so retargeting budget rotates to fresh demand. Suppression is pure efficiency: same creative, same budget, fewer wasted impressions. Most accounts find a double-digit CAC improvement before they touch creative at all.

A static list upload decays within weeks because intent is a moving target. The durable version is a loop: RB2B or Snitcher resolves visitors, Clay enriches and scores them against your ICP, your CRM (HubSpot or Salesforce) holds the canonical account record, and a scheduled sync refreshes the matched audiences and exclusion lists in each ad platform on a daily or weekly cadence. New in-market accounts flow into targeting automatically; closed and disqualified accounts flow into suppression automatically. The advertiser owns the pipes, no agency holds the keys.

This is what it means to treat marketing like infrastructure rather than campaigns. The creative and the bidding still matter, but they now sit on top of a signal layer and identity graph that compound over time instead of resetting every campaign. The same resolved-account signal that powers ads also powers outbound and content sequencing, so one identity graph drives allbound. That shared layer is the asset; the ad account is just one place it gets expressed.