Operating brief: the summer slowdown and the Q4 ramp
CPMs soften every summer and every UA team half-forgets it. A working checklist for spending the quiet months on the tests you never get to run in Q4.
Every UA team knows CPMs soften over the northern hemisphere summer. Fewer of them plan for it. The auction gets quieter as e-commerce advertisers pull back through July and August, with entertainment and finance following, and mobile game buyers who keep spend flat through the period end up buying cheaper impressions without deciding to. That isn't a strategy. It's an accident with good timing; nobody chose it.
The useful question for late July isn't whether the slowdown is real, it's what to do with the room it creates before Q4 demand pushes CPMs back up. Three areas are worth the attention this quarter tends to starve. None of them is glamorous.
Through the summer slowdown, run the tests that need clean signal
Incrementality reads and geo holdouts need a stable auction to produce a trustworthy result, and so do creative portfolio experiments. A test launched in early November, when GTA 6's marketing spend and the usual holiday advertiser surge are both pushing CPMs around, will have a noisier baseline than the same test run in late July or August. If there's a channel mix question or a bidding strategy change under debate for Q4, this is the window to settle it with data rather than argue it in a planning meeting under time pressure.
Rebuild the creative bench, not just the top performers
Summer volume is lower, which means a given creative budget buys more distinct concepts tested to a meaningful sample size than the same budget would in a competitive Q4 auction. Teams that spend the summer refreshing only their proven top performers are optimising for a quarter that is already working. The bench that needs building is the following wave: the concepts that will need to be ready and tested, with a round or two of iteration behind them, by the time Q4 budget arrives, because there won't be spare auction room to discover them fresh in November.
Renegotiate before the advantage shifts
Rate cards are easier to discuss when a partner has more inventory than demand. So are minimum spend commitments, and so is any network-specific deal term. That advantage flips in Q4. Any conversation about volume discounts or new creative formats, or about access to beta placements, is better had in August than in the week before Black Friday, when every network's account team is managing a queue.
What not to do
Don't read a summer dip in blended CPM as a signal that a channel or campaign has genuinely improved. Isolate seasonal effect from campaign effect first. A campaign that looks efficient in August purely on lower auction pressure can disappoint badly once the market fills back in, and any conclusion drawn from it will feed a Q4 budget allocation. The teams that get burned every autumn are usually the ones who mistook a quiet auction for a working strategy.
Staffing the ramp, not just the budget
The other thing summer buys is calendar space for people, not just for spend. Creative production gets squeezed once Q4 volume arrives, as does QA, as does account management, and the requests that suffer first are the ones with no fixed deadline: a new market test, a secondary creative format, a review of tracking setup nobody has had time to audit properly. Scheduling that work for August, while it can still happen at a normal pace, is cheaper than scheduling it for November, when it competes with genuinely urgent campaign fires.
Setting the actual Q4 trigger points now
A ramp plan written in July should specify the signals that trigger a budget increase, not just the target date. Waiting for a calendar date to increase spend ignores that the auction moves on advertiser behaviour, not the calendar itself, and this year that behaviour includes at least one confirmed large media event later in the autumn. Agree now on what evidence will trigger the Q4 increase, whether that's a CPM move, a competitor's visible ramp or a platform signal, and the decision gets made against a plan rather than in reaction to whichever number lands on a Monday morning dashboard first.
