Most performance budgets are spent chasing people who are still deciding whether they want anything at all. By the time intent is real, the auction has already taken a cut that has nothing to do with the sale.
PubadHub is built around the opposite idea: buy the last hop — the click-out that happens when a shopper already has the card out — and price it against orders, not impressions.
Why the final hop behaves differently
A pageview tells you someone arrived. A click-out tells you someone chose. Those are not the same signal, and they should not carry the same price. When a shopper leaves a comparison page, a coupon listing, or a review round-up to reach a merchant, they have already done the hard part of the funnel on someone else’s property.
Intent is already priced in
Upper-funnel media pays to create intent. Click-out inventory pays for intent that exists. That single difference is why the same budget produces a different cost per order depending on where in the journey it lands.
What the auction hides
Broad-match search and display auctions bundle discovery traffic with purchase traffic and hand you one blended number. Once the blend is broken apart, most advertisers find that a small slice of placements carries the majority of the revenue. You can read more about isolating that slice in our note on reading a cost per order that holds up.
What you actually buy
Placement-level click-out inventory is sold as a hop, not a banner. The three things you control are:
- Placement — the exact module, list position, or page type the hop comes from.
- Price — a bid per click-out, set in real time or fixed for the month.
- Pass-back — the click ID that lets you tie the hop to a downstream order.
Setting a benchmark before you move budget
Do not shift spend until the baseline is boring. Five steps, in order:
- Freeze your current channel mix for a full purchase cycle so the baseline is not moving underneath you.
- Agree on one definition of an order, including whether returns are deducted and over what window.
- Instrument the click ID end to end and confirm it survives redirects, app handoffs, and consent walls.
- Pick a single category to test rather than spreading a thin budget across the whole catalogue.
- Run for at least three refund windows before you call the result.
What the numbers looked like
The table below compares a mid-size retail programme over one quarter, before and after moving fifteen percent of paid search budget into click-out inventory.
| Metric | Paid search only | With click-out | Change |
|---|---|---|---|
| Cost per order | $41.20 | $28.60 | −30.6% |
| Orders per month | 3,140 | 3,905 | +24.4% |
| Return rate | 11.8% | 11.2% | −0.6 pts |
| Time to first order | 4.2 days | 1.6 days | −62% |
Where this breaks down
It is not free money, and there are three situations where the model underperforms:
- Long consideration cycles where the “last five seconds” are weeks apart from the first touch.
- Categories with heavy offline fulfilment, where the order never lands in the same system as the click.
- Programmes with no refund window discipline, which flatter early results and disappoint later.
A rollout plan you can defend
- Week one: instrumentation and a read-only reporting dashboard.
- Weeks two to four: single-category test at ten to fifteen percent of budget.
- Week five: reconcile orders against finance, not against the ad platform.
- Week six onward: scale by placement, retire the ones that only look good pre-refund.
Final takeaway
Buy the hop that happens after the decision, price it against orders, and make the placement visible. If you want to see how this maps to your catalogue, book a walkthrough with the team.