Exact match in its own campaign, or folded into broad with negatives?

I'm not convinced by the "exact match doesn't matter anymore" takes. The usual split, exact in its own campaign and broad running with a big negative list, is really about control over which queries spend money. Take a store that sells one specific espresso machine brand in a clear price band. If exact still catches the high-intent brand and model terms while broad handles discovery, you can see what each one actually costs and earns. Merge them and Smart Bidding gets one bigger pool of data, which it likes, but you lose a clean read on the terms that matter most. Has anyone with a catalog of a few hundred SKUs merged them and seen the branded terms hold up?

Agreed on the control point, but exact match isn't as clean a filter as it used to be. Close variants mean a brand keyword also picks up misspellings, plurals and reordered words, so the "exact" line in your report can include queries you never meant to buy. I'd split by intent instead of match type. Put brand terms in their own Search campaign with its own budget, then add those terms as negatives in the discovery campaign so the two don't bid against each other. Say the brand gets a few hundred searches a month and a competitor term gets a few dozen. The brand campaign gives you a clean cost per sale, and the discovery campaign shows what broad finds on its own. Keep an eye on the search terms report in both, since the variants leak more than the match type suggests.

Negatives only protect the campaigns you add them to. If the store also runs Performance Max or Shopping, brand queries can still land there and bid against your brand campaign. PMax has brand exclusions you set at the account level, so open that list and confirm the brand terms are actually on it instead of assuming they are.

Also filter the brand campaign's search terms report for queries that don't contain the brand name at all. Those are usually generic or competitor terms that slipped in through broad-leaning matching. They belong in discovery, and until they're moved out, your brand cost per sale looks better than it really is.

Depends on brand search volume. With a few hundred brand searches a month, a separate campaign gives you a cost per sale you can actually read. Under a few dozen, the split just produces noisy numbers, so it's simpler to run one campaign and add brand negatives in discovery.

I don't think volume should be the deciding factor. Brand negatives in discovery stop the overlap, but a single campaign still bids on brand and generic queries against the same target, so the blended number hides what brand actually costs you. Low volume just means you wait longer for a readable split. That's a reason to keep the split and be patient, not to merge.

Fair on control, but measurement and structure are two separate problems. You can get a brand versus generic read inside a single campaign by filtering the search terms report for queries that contain the brand name, misspellings included, and totaling cost and conversions for that set. No budget split needed. It only works if the filter is clean, which is where Priya's variant point matters, so pull a sample of the matched queries and eyeball them before you trust the number.

One catch with that filter: the search terms report leaves out queries below Google's privacy threshold, so your brand and generic rows won't add up to the campaign total. Compare the two before you trust the split. If the hidden spend is small, the read holds. If it's a big chunk, the brand number could be off in either direction.

Agreed, and the gap is usually bigger on the brand side than people expect. Brand queries are high volume with a long tail of misspellings, so more of them get dropped by the threshold than you'd guess. A quick check is to compare the campaign's total conversions with the sum of the brand and generic rows, since conversions are often the first number to go missing.

I'm not sure brand is the bigger gap. The threshold works on individual query volume, and brand searches pile up into a few high-volume queries that clear it easily. The queries that get hidden are more likely generic or competitor terms, so reconciliation matters most in discovery.

The big brand queries probably do clear the threshold, but the misspelling tail is still worth checking, since those are often one-off strings. Compare hidden conversions, not hidden spend, because conversions are what the brand versus generic split is really judging. If they're more than about 10% of the campaign total, I wouldn't trust those numbers yet.