Broad match with Smart Bidding on a small budget: open it early or wait for data?

Take a plumber on a $30 a day Search budget wondering whether broad match plus Smart Bidding is worth it. The case for broad is that Google matches intent better than it used to, so phrase and exact lists leave volume on the table. The case against is that a bidder with very little conversion data has almost nothing to learn from, so broad mostly buys searches like "free plumber course" and burns budget before the model figures out what good looks like. My view is that under roughly 30 conversions a month, you stay on phrase and exact with a tight negative list, and only open up broad once you have enough data to separate good queries from junk. Has anyone seen that threshold come down on a niche B2B account?

I'd push back on the 30 conversion threshold, because it treats match type as the thing that burns budget, and it isn't really. "Free plumber course" can show up on phrase match too, so staying on exact just means you collect data more slowly. The bigger risk is a target CPA or ROAS set on 8 conversions a month. The model is basically guessing at that volume, so it either bids too loose or starves the campaign. On a $30 budget I'd run broad with Maximize Conversions and no target, and clean out junk queries from the search terms report every week. You get more search data that way, and the negatives do most of the work.

Yeah, target CPA is where small accounts usually get burned. Bing's volume on local terms is often a fraction of Google's, so one junk query can eat a big share of your clicks. Because of that, I'd give the search terms report two weeks instead of seven days before adding negatives.

I don't buy the two-week wait. On $30 a day that's about $420 spent before anyone touches the negatives, and junk queries like "free plumber course" or "plumber salary" don't stop showing up while you wait. You can spot those within a few days, so blocking them doesn't need statistical comfort. Waiting really only helps with the harder calls, like whether a near-miss query converts at a decent rate, and a small account won't have the volume to answer that anyway. My advice would be to check the search terms report every few days, kill the clear junk early, and leave the slower judgment calls for later.

Agreed on the cadence, but I'd block by category instead of one query at a time. On a small budget, the expensive searches are usually intent you can name: DIY, jobs and careers, courses, "how to fix". Add those as campaign-level negative phrases so they're blocked across every ad group, not just the one where they happened to show up. Leave the grey ones like "cheap" or "emergency" off the list until you have conversion data, because that's where some of your paying customers hide.