When a client briefs CPA (or any cost-per-outcome model) on web and we can't commit with confidence, we don't decline. We propose a CPC phase first: install PF Pixel, learn real traffic and conversion signals, then come back with a CPA number we can stand behind. The client gets a live campaign now instead of a no, and we protect margin instead of guessing.
No pixel or tracking history on the client's site: nothing to model CPA against.
New client, no historical performance data on the account.
New or unproven product/landing page: no conversion baseline exists.
Client's target CPA is below category benchmark and they won't move on it yet.
Low traffic / new market entry: not enough volume to optimise toward CPA fast.
Brand-new product category we haven't run media for before.
"We want to get this right rather than guess and overpromise. For the first 3-4 weeks, we'll run on CPC with PF Pixel installed across your site. This gives us real signal on traffic quality and conversion behaviour. Once we've validated that against benchmarks, we move to CPA with numbers backed by actual performance, not assumptions."