Automated campaign types work by testing combinations and reallocating toward what converts. That mechanism has a prerequisite that rarely gets said out loud: enough conversions, often enough, for the system to tell signal from noise.

Below that threshold the automation does not fail loudly. It spreads spend across placements, collects a thin scatter of data, and produces a report full of activity. The spend is real. The learning is not.

The question to ask before you turn it on

Not “what is the minimum budget?” but: how many real conversions will this campaign generate per week?

Take your realistic monthly spend, divide by your typical cost per acquisition, and divide by four. If the answer is a handful of conversions a week, automation has very little to work with. If it is under one or two, you are paying for exploration that will not converge.

For a business where one conversion is worth a lot and they come in slowly — legal, construction, specialist B2B — this maths is often unfavourable even at respectable budgets.

What low-volume advertisers should do instead

  • Start with search on tightly chosen terms. Fewer keywords, clear intent, manual control over where money goes. You will learn what actually converts, which is the input automation needs anyway.
  • Feed the system a better conversion. If your only conversion is a form submission, count qualified enquiries instead. Fewer, better-defined conversions beat a larger pile that includes junk.
  • Consider an upstream conversion as a proxy. If true conversions are too rare, a reliable earlier step — a pricing page view plus time on site, a call started — can give the system enough volume. Only do this if that step genuinely predicts the outcome.

If you are going to run it anyway

Give it the best inputs you can and then leave it alone long enough to mean something. Strong assets rather than the minimum, a clear exclusion list for the traffic you know is wrong, and at least a few weeks before judging. Changing targets weekly resets the learning you are paying for.

And check where the money actually went, not just the headline cost per conversion. If most of the spend landed on placements far from search intent and the conversions are suspiciously cheap, look closely at what is being counted before you celebrate.

The honest summary

Automation is a multiplier on a working account, not a substitute for one. If you already know which terms and audiences convert, it can scale that efficiently. If you do not know yet, it is an expensive way to find out — and a manual search campaign will teach you the same thing for less.