Anti-Adblock Inventory: Incrementality Before Spend

Anti-adblock inventory is mostly not a new audience. It is people your other campaigns already reach, counted again on the sessions where they happened to be blocking. So the first deliverable here is a measurement, not a media plan, and on more than one account that measurement has ended with us recommending no spend at all. If you want a channel that makes a slide look better, this is the wrong desk. The measurement: We run a geo holdout. Two comparable market groups, matched on baseline conversion rate and seasonality, one exposed to the recovery buy and one held out for a fixed window, read on total conversions in your analytics rather than on anything the platform reports about itself. Where geography cannot be split cleanly we use a ghost-bid split instead, bidding to win and suppressing the render for the control arm. The test is sized to detect a lift large enough to justify the channel, not the smallest lift that is technically visible, and we tell you that threshold before it runs so it cannot be moved afterwards. Four different things get sold as one number: Filter-list blocking at the network layer, cosmetic filtering that hides elements already loaded, DNS-level blocking through AdGuard DNS, NextDNS or a household Pi-hole, and browser-native shielding. Recovery technology reaches the first two. It does not reach the third or the fourth, and it never will, because there is no request to recover. Buyers are routinely quoted the union of all four as their addressable adblock audience, which is how a channel gets approved at three times its real size. Part of this engagement is telling you which fraction of your blocked sessions is actually purchasable in each market you care about. Where the money goes: Publisher-direct recovery supply through vendors participating in the Acceptable Ads framework, and in-page push served from publisher first-party domains. Static creative, no autoplay video, weight held under 150KB, and layouts written against the publisher's own markup rather than generic ad class names, which cosmetic filters catch within days. Frequency stays low. This audience opted out of advertising once already and the buy should behave like it knows that. What we will not build or buy: Circumvention scripts. Domain rotation to outrun filter lists. CNAME setups that dress a third-party tracker as first-party. Ad calls misdeclared as content requests. Every one of those works for a few weeks, dies when a maintainer files a rule, and leaves you with a rebuilt stack and no inventory. They are also a bad-faith answer to a reader who made an explicit choice, and that is the part we care about more. Attribution: Recovery traffic gets its own click ID space in your tracker and its own postback endpoint, so recovered conversions cannot quietly inherit credit from paid social or organic. We set attribution to click-based with a short window and no view-through. View-through on recovered inventory is a way of billing you for people who were coming anyway, which is the exact failure this whole engagement exists to detect. Not included: Publisher-side monetisation. We buy this inventory, we do not sell it, and we do not advise publishers on recovery stacks, because the same firm cannot honestly sit on both sides of that trade. We also do not build creative for other channels off the back of this work. Who this is not for: Advertisers in categories where consent posture on the publisher side cannot be verified in writing, health and financial services in particular. Teams who have already signed a recovery vendor and want a justification written after the fact. And anyone who needs the answer inside a week, since a holdout that short measures nothing but noise.