Single keyword ad groups — one keyword per ad group, in all three match types — were once the default advice for serious search accounts. The pitch was total control: if an ad group holds exactly one keyword, you know precisely which query triggered which ad, and you can align copy and bids perfectly to intent. That logic was sound in its era. It is now mostly obsolete, because Google spent the last several years dismantling the exact control SKAGs were built to provide.
As Store Growers put it, the SKAG era ended not because the logic was wrong, but because Google changed the rules underneath it, starting with the 2018 expansion of close variants. For B2B SaaS specifically, where conversion volume is thin and every structural choice affects how much data Smart Bidding gets to learn from, the case against blanket SKAGs is even stronger — and the case for using them surgically on a few key terms is worth understanding precisely.
What SKAGs were built to solve
In the mid-2010s, exact match meant something close to exact, and search ads were static text you wrote by hand. Under those rules, packing many keywords into one ad group meant your ad copy was a compromise across all of them, and your bids could not reflect the different value of each query. SKAGs solved both problems: one keyword per ad group let you write copy that echoed the exact search term (which lifted click-through and Quality Score) and set a bid that matched that term’s specific value. For high-stakes accounts, the maintenance overhead of hundreds of tiny ad groups was a fair price for that precision.
The whole structure depended on one assumption: that you controlled the mapping between query and ad. That assumption is what Google has spent years eroding. Close variants widened what exact and phrase match catch, responsive search ads replaced hand-written copy with machine-assembled asset combinations, and Smart Bidding took over auction-level pricing. Each change removed one of the levers SKAGs existed to pull, until the structure was maintaining control the platform no longer handed to advertisers at all.
Why close variants and automation broke the model
The decisive blow was close variants. Once exact match started matching reworded queries, synonyms, and same-meaning searches, a single query could legitimately trigger keywords sitting in several different SKAGs. Instead of clean one-to-one control, you got overlap: multiple ad groups eligible for the same auction, competing internally, with Google picking a winner by its own logic. The precise routing SKAGs promised became impossible to guarantee, so the main benefit evaporated while all the maintenance cost remained.
Automation finished the job. Responsive search ads mean you no longer write one ad per keyword — you supply assets and Google assembles combinations per auction, so hyper-specific copy per SKAG matters less than a strong asset pool. And Smart Bidding prices each auction using signals a static keyword bid could never capture, which makes per-keyword manual bids largely redundant. This is the same shift that reshaped how match types themselves behave: the account structures that made sense under manual control need rethinking under intent matching and automated bidding.
The hidden cost: fragmented conversion data
For B2B SaaS, the strongest argument against blanket SKAGs is not relevance — it is data density. Smart Bidding needs a meaningful stream of conversions per campaign to learn, and splitting your keywords into hundreds of single-keyword ad groups scatters conversions so thinly that no ad group ever accumulates enough signal. A niche software account that converts a few hundred times a month across the whole account cannot afford to fragment that volume; it needs to concentrate it. This is the same constraint that makes thin data so damaging in low-volume SaaS accounts, and SKAGs actively make it worse.
The modern answer is the single theme ad group (STAG): a cluster of 3-20 keywords that share one intent, with copy written to that theme. STAGs give up a little ad-to-query precision in exchange for pooling conversions where Smart Bidding can use them, which is exactly the trade a volume-constrained B2B SaaS account should want. The relevance you lose is small — responsive search ads and close variants already blur the per-keyword edge — and the data you gain is the difference between Smart Bidding learning and Smart Bidding guessing.
The modern default: theme-based ad groups by intent
The structure most B2B SaaS accounts should run in 2026 is theme-based ad groups organised around buying intent, not product features. Group keywords that share a stage and a message — for example, high-intent commercial terms in one ad group, comparison and alternative-to terms in another, problem-first terms in a third — and write responsive search ad copy to each theme. That keeps ads relevant enough to hold Quality Score while giving each ad group enough volume to matter.
Crucially, the ad-group layer should sit inside a deliberate account structure. We argue for organising campaigns around funnel and intent tiers so that budget and bidding targets can differ by how close a searcher is to buying — a bottom-funnel “pricing” theme deserves a different target than a top-funnel research theme. Theme-based ad groups are how you populate those tiers without fragmenting the data, and they pair naturally with the argument to consolidate enough that each campaign clears the learning threshold.
Where SKAGs still earn their keep
SKAGs are not extinct — they are specialised. Reserve them for the small set of terms where control genuinely pays and volume is high enough to survive isolation: your top commercial keywords that drive a disproportionate share of pipeline, competitor terms where messaging must be airtight, and brand terms you want to defend with exact copy. For these, the click-through and conversion-rate lift from perfectly aligned copy can justify the maintenance, and the volume clears the learning bar even in a dedicated ad group. Brand and competitor defence is a natural fit, connecting to how we think about competitor and brand bidding for B2B SaaS.
The winning pattern is a hybrid, not a doctrine. Run STAGs as the account-wide default, then carve out SKAGs surgically for the handful of terms that earn the extra control. Let the data decide which terms graduate: launch everything in theme-based ad groups, watch which individual keywords drive outsized revenue, and promote only those into dedicated ad groups. This keeps structure serving performance rather than ideology — the opposite of the 2010s reflex to SKAG-ify the entire account. Applied this way, single keyword ad groups remain a useful tool; applied everywhere, they are one of the more common Google Ads mistakes B2B SaaS accounts still make.