Conversion lag: judge last week's CPA on matured data or react to the drop?

Every week someone posts that their CPA doubled and asks whether to pause the campaign. Often the last three or four days just haven't converted yet. With e-commerce and a short purchase lag that's a small problem, but for B2B with a 3 to 6 week sales cycle the recent days in the conversions column can look badly underreported, so what looks like a drop can be nothing more than lag.

My view is to judge performance on a window that's had time to mature, something like the 14 days ending a week ago, and only react faster when spend itself went wrong, like a budget spike or a broken landing page. The catch is that a week-old window makes you slow to catch real problems. How do you handle this for longer cycles, a fixed lag window or just waiting it out?

I'm not sure "only spend problems justify reacting fast" holds. A broken form or tag doesn't look like a spend problem at all. Clicks stay steady while conversions drop, and you can catch that within days by watching click-to-lead rate against its usual level. Let the CPA mature before you act on it, but keep a fast lane on the funnel ratio.

Agreed on the fast lane. The ratio works best when you check the form separately, though. Clicks and sessions can hold steady while a broken form or tag quietly kills conversions. So if click-to-lead drops suddenly while sessions stay flat, look at tracking or the page first, not the ads.

Kai's split makes sense to me. I'd take the cutoff from your own data instead of guessing: open the conversion lag view in Google Ads and find the day by which about 90% of past conversions had landed. For a B2B account with a long cycle that might be day 10 or day 21. Anything newer than that is provisional, so don't read it as a trend, and judge CPA only on the mature window. Recent days are always undercounted, so comparing them to older weeks makes healthy campaigns look like they're dying.

The cutoff method is solid, but I'd check where the lag actually comes from. If you import offline conversions from a CRM, the lag view also counts however long the upload takes, so a 21-day number might be partly CRM delay and not the real sales cycle. Pull a sample of leads and compare their actual close dates against the upload dates to split the two before you set the window.