Browser Extension Partnerships Without Checkout Interception

Two numbers settle most arguments about an extension partner: the median time from click to conversion, and the URL the click landed on. Where half of a partner's clicks resolve inside ninety seconds and land on a cart, a booking summary or a checkout page, the partner is being paid for a decision the buyer had already made. We rewrite those terms or end the partnership, and we keep the extension placements that reach somebody before they have chosen the merchant. The diagnostic: Click-level logs pulled from the network, raw click exports out of Everflow, click logs out of Impact, whatever a network will genuinely release, joined to your own session data. What matters is the distribution of click-to-conversion time rather than its mean, the class of landing URL, the share of converting sessions carrying a direct or brand-paid touch within the previous thirty minutes, and the share of coupon sessions where no code was ultimately applied. Four weeks of logs normally ends the argument. Then a holdout. The partner is switched off for a randomised share of sessions or for one market, three to four weeks, measured on orders and bookings rather than on attributed revenue. This is the part advertisers dislike and it is the only part that answers the question they asked. Terms we write: A click has to be user-initiated and visible, an interaction rather than a fire on page load. No click registered on cart, checkout or confirmation URLs. A lower tier, or nothing, on sessions where no code was applied. Exclusion of sessions with a same-day direct or brand-paid touch. A ban on the partner bidding your brand inside its own in-extension search surface, which is where a good deal of the interception has moved. Click-level data delivered rather than summarised. And a right to audit with a clean test profile, exercised rather than merely written in. The placements worth buying: Extension inventory that appears before the merchant is chosen. Category and comparison surfaces inside the extension, stores-like-this recommendations, an activation prompt fired while the user is sitting on a competitor's site, price-drop alerts that bring back a lapsed buyer. Honey, Capital One Shopping, Karma, Coupert, Klarna's extension and the coupon layer built into Microsoft Edge are each a different negotiation, and Edge is the one you cannot drop, only shape. What is not included: No network migration, no creative for in-extension units, no legal action against a partner. Where a partner's behaviour is a contract breach rather than a bad deal, we document it in a form your counsel can use and then it is their matter. What we refuse: We do not work with a partner who will not release click-level logs; the summary dashboard is the product being sold and it is not evidence. We do not accept that these are the standard terms from a partner running custom terms with three of your competitors. And we do not run an incrementality test whose result the advertiser has already told us they will not act on, which is a fee for a document. Who this is not for: Advertisers who cannot carry a short-term fall in reported revenue while a holdout runs. The reported number drops before the real one moves, and somebody senior has to be ready for that conversation before the test starts, not during it. Advertisers whose finance function reads the network dashboard as revenue. And advertisers with a single extension partner and no room to renegotiate yet; the work pays when there are three. You end with the log analysis, the holdout result, rewritten terms with the partners worth keeping, and a written rule your team applies to the next extension partner that approaches you, so this does not become an annual engagement.