Broad match with Smart Bidding: does a big negative list before launch still matter?

I'm setting up a broad match campaign on a modest budget and trying to work out how much pre-launch blocking it really needs. The usual advice is to load a long negative list before launch so Smart Bidding doesn't burn its learning period on junk queries. On a small budget, though, you're mostly guessing which terms are junk, and the list ends up shaped by a handful of clicks. Say a dental clinic adds "cheap" and "free" on day one, then later finds "affordable" converts well. You've blocked a good variant before the data had a chance to show it.

Has anyone run broad match with a light negative list and had it hold up, compared with starting tight and loosening later? I'd also like to know how many conversions it took before the search terms report felt trustworthy.

Blocking "cheap" as a phrase match won't stop "affordable" from showing, so that example is really about how wide the negative is, not about timing. For "free" and "cheap", phrase or exact negatives are the way to go, and a short list of clearly irrelevant terms like "jobs" or "DIY" costs you very little to start with.

Agreed on the phrase point. On a small budget, junk queries are mostly a cost problem, not a learning problem. Smart Bidding needs conversions to learn, not perfectly clean traffic, so a light negative list just means paying for a few irrelevant clicks while you wait. I'd keep "affordable" live and watch its conversion rate in the search terms report rather than block it blind. The obvious non-buyers like "jobs" or "course" are different, because no amount of data will make them convert.

How many conversions a month are you realistically expecting from this one?

I'm not sure junk queries are only a cost problem. On a small budget they also muddy what Smart Bidding learns from, because a few irrelevant conversions or near-misses can drag the target around before you have enough real data to outvote them. So I'd still keep the lightweight negative list, as long as it sticks to obvious non-buyers.