Display ad costs fell hard in Q3 2026 — prospecting CPMs down 45% year over year in AdRoll’s Q3 report, covering 1 July to 8 September 2026 — but consumer confidence fell with them. Cheaper impressions are real. Cheaper customers are not. Buy the discount carefully, and do not let Q4 budget swings reset your bidding.
Why did display ad costs fall so sharply in Q3 2026?
Because advertisers pulled back before the holiday rush instead of piling in early. AdRoll’s Q3 2026 data, reported by Search Engine Land on 22 September 2026 and covering 1 July to 8 September, shows display prospecting CPMs down 45% year over year and 25.5% against Q2. Retargeting fell 29.1% year over year and 40.2% against Q2. Account-based campaigns went the other way, up 4.4%.
Worth saying plainly: this is US data. Canadian auctions are smaller and more concentrated, so the direction usually carries north of the border but the exact percentages do not. Treat the 45% as a signal that upper-funnel inventory got cheaper, not as a number for your own forecast.
Does cheaper inventory mean cheaper leads?
No — and this is the part most owners get wrong in October. CPM is what you pay to be seen; cost per lead depends on whether the person who saw you is in a buying mood. The same report pairs the price drop with a wobbly consumer: the University of Michigan sentiment index sat at 47.8 in September 2026, down 13.2% year over year, US inflation held at 3.4% in August, and Bank of America card data showed household spending up 4.5% year over year. People are still spending, just more nervously.
Weak sentiment with healthy spending usually means longer consideration and more comparison shopping. In practice that looks like a steady click-through rate, a falling conversion rate, and a quote request that takes three touches instead of one. Cheap impressions in front of a hesitant buyer produce traffic, not revenue.
| Metric | Q3 2026 reading | What it means for your budget |
|---|---|---|
| Display prospecting CPM | Down 45% year over year (AdRoll, US, 1 Jul–8 Sep 2026) | Cheap reach — worth testing, not worth betting the quarter on |
| Display retargeting CPM | Down 29.1% year over year | Cheaper to re-engage past visitors, if your audiences are populated |
| Account-based CPM | Up 4.4% year over year | High-intent targeting is still competitive — no discount there |
| US consumer sentiment | 47.8 in September 2026, down 13.2% year over year | Longer decision cycles; budget for follow-up, not just clicks |
| Smart Bidding learning | Up to ~50 conversions or three cycles (Google, 22 Sep 2026) | Low-volume accounts need weeks, not days, after a change |
What does Google’s new Smart Bidding guidance change for Q4?
It changes how long you have to leave things alone. On 22 September 2026 Google reframed its Smart Bidding learning guidance around conversion volume rather than calendar time: a bid strategy needs up to around 50 conversion events or three conversion cycles to recalibrate after a change, per Search Engine Land the same day. Accounts with strong history may settle faster.
Do the arithmetic on your own account first. A home services business getting 40 leads a month faces more than a month of learning after a bidding or conversion-goal change. Change your target in the first week of November because costs look cheap, and the strategy will still be calibrating when Black Friday lands.
The practical rule: make structural changes now, in late September and early October, so learning finishes before demand peaks. After that, move budgets — which Smart Bidding tolerates far better — and leave targets and conversion goals alone.
How should a small business actually set its Q4 budget?
Set it from your own cost per lead history and close rate, then let cheap inventory buy extra volume rather than a new strategy. The accounts that win Q4 are the boring ones: clean conversion tracking, a bidding strategy left to settle, and budget added in steps once the cost per lead holds.
That step-wise approach is what has worked in our own accounts. A GTA mortgage brokerage we manage started at roughly 200 leads a month at $13.38 each and, after restructuring Search and Performance Max and feeding cleaned lead data to Smart Bidding, reached 1,113 leads in August 2025 at $7.77 — budget scaled in steps as the cost per lead held, never in one jump. The numbers are in the mortgage brokerage case study. A cheap-CPM quarter does not change the method; it just makes each step cheaper. If you want that structure built before the quarter turns, that is what our Google Ads management work is.
What to do this week
- Count your monthly conversions and divide 50 by that number. That is roughly how many months a bidding change costs you in learning. Then finish all bidding and conversion-goal changes before mid-October.
- Check that your remarketing audiences are populated and your tags are firing. Discounted retargeting inventory is worthless if the list is empty.
- Test cheap prospecting inventory with a small, separate budget — 10 to 15% of spend — and judge it on cost per lead, not CPM. Kill it in three weeks if the leads do not arrive.
- Add one follow-up step for slow buyers: a callback within the hour, or a second offer at day seven. With sentiment this soft, the money is in the second and third touch.
Frequently asked questions
Should I increase my ad budget because CPMs are down?
Only if your cost per lead is already at or below target. Falling CPMs reduce what you pay for impressions, not what you pay for customers, and the AdRoll Q3 2026 figures are US data that may not match your Canadian auction. Raise budget in steps of 15 to 20% and watch cost per lead for a full week between steps. If it holds, step again. If it climbs, you have found your ceiling for the quarter.
How long does Google Ads Smart Bidding take to learn in a small account?
As of Google’s 22 September 2026 guidance, up to around 50 conversion events or three conversion cycles after a change. For an account producing 25 leads a month, that is roughly two months — far longer than the seven days many owners assume. Accounts with substantial history often settle faster. The safe approach is to make structural changes well before your busy season and treat budget adjustments, which disturb learning less, as your in-season lever.
Is now a good time for a small business to test display advertising?
Better than most, with the caveat that display rarely closes a sale on its own. Prospecting CPMs down 45% year over year mean a test costs less than a year ago, so a small ring-fenced budget is reasonable for a business that already has search working. If search is not yet profitable, fix that first — display in front of a hesitant buyer is an expensive way to build awareness you cannot convert.
Why are ad costs falling if consumers are still spending?
Because advertiser confidence and consumer behaviour are moving apart. Bank of America card data showed US household spending up 4.5% year over year in August 2026, while the University of Michigan sentiment index fell to 47.8 in September, down 13.2%. Advertisers reading the sentiment number held budgets back, thinning the auction and pushing CPMs down even as people kept buying. That gap is the opportunity — and the trap, if you assume cheap reach converts like expensive reach.
Want a second opinion on your Q4 plan before you commit the budget? Book a free 48-hour audit.
Sources: Search Engine Land on AdRoll’s Q3 2026 CPM data, 22 September 2026; Search Engine Land on Google’s Smart Bidding learning guidance, 22 September 2026.