A campaign that performs well at modest spend deteriorates as budget increases, and the effect is consistent enough to be treated as a law of the channel.
The best audience is reached first
Advertising platforms optimize delivery toward people most likely to take the desired action. At low budgets they serve the highest-probability segment almost exclusively.
That segment is finite. Increasing spend requires serving people the system already ranked as less likely to respond.
Cost per acquisition therefore rises with volume for mechanical reasons, before any change in creative, competition or market conditions.
Frequency erodes response
Extra budget spent on the same audience raises how often each person sees the advertisement. Response per impression declines as repetition increases.
The decline is not just fatigue. The people most likely to act on a message usually act early, leaving a residual audience that has already declined.
Creative refreshes recover part of this, which is why sustained programs need a production pipeline rather than a single campaign.
Auctions price attention against competitors
Placements are sold at auction, so cost reflects what other advertisers will pay for the same person. Increasing your bid to reach more people also raises the clearing price.
Seasonal periods compress this sharply, as retail advertisers bid heavily into the same inventory and costs rise for every category using it.
A business bidding on high-intent commercial terms competes against everyone else who values that intent, so the most efficient keywords are the most contested.
Attribution flatters small budgets
At low spend, much of the reported conversion comes from people who would have bought anyway, since the campaign is reaching those already close to purchase.
Reported efficiency at that stage overstates incremental effect, so the apparent deterioration when scaling is partly the illusion being removed.
Holdout testing across regions or audiences separates the two, and it usually shows true incremental cost above the platform-reported figure.
Structural limits set where the curve ends
Every market contains a finite number of people with the relevant need in a given period. No amount of budget creates demand that does not exist.
Growth beyond that point requires reaching a different audience, entering a new geography or changing the product, all of which reset the curve rather than extend it.
The practical consequence is that a plan built on holding early acquisition costs at higher volume will miss, and the shape of the curve is worth measuring before the budget is committed.