For years, single keyword ad groups (SKAGs) were the mark of a well-run Google Ads account. One keyword, one ad group, one perfectly matched ad. Total control.
When we took over a UK home improvement group’s account last year, it had been built exactly that way: hundreds of tiny ad groups across two brands, more than £1m a year of spend, and around two-thirds of the business’s revenue riding on it.
It was also quietly holding the account back. This is the story of what we changed, why, and what happened to the numbers, along with a checklist you can use on your own account.
Where SKAGs came from
SKAGs made a lot of sense a decade ago. Advertisers set bids by hand, keyword match types were strict, and the easiest way to write a relevant ad and control your cost per click was to give every important keyword its own ad group. If someone searched “composite front doors”, they saw an ad that said “Composite Front Doors”, and you set a bid for exactly that search.
Agencies loved them because they looked meticulous. Clients loved them because they felt in control. And for a while, they worked.
What changed
Three things happened that turned SKAGs from best practice into a handicap.
Bidding moved to machines. Google’s smart bidding now sets a bid for every single auction, using signals no human can see: the device, the time of day, the location, the person’s recent searches and more. It learns from conversions. The more conversion data it has in one place, the better its decisions.
Match types loosened. Exact match hasn’t really meant exact for years. Close variants, rewordings and searches with the same meaning now all trigger the same keyword. The neat one-keyword-per-ad-group structure no longer delivers the control it promises.
Ads became responsive. Google now tests combinations of headlines and descriptions automatically. You no longer need a separate ad group to show a relevant headline, as long as the ad group is built around a single intent.
Put those together and SKAGs start working against you. You’ve taken a system that learns best from pooled data and chopped its data into hundreds of tiny pieces.
What we found in the account
When we audited the account, the symptoms were textbook:
- Fragmented data. Most ad groups had only a handful of conversions a month, far too few for smart bidding to learn from.
- Slow learning. Every change to a bid strategy or budget took weeks to settle, because the signal was spread so thin.
- Optimising for the wrong thing. The account was tuned for clicks and click-through rate, not for the enquiries and sales the business actually needed.
- Hard to manage. Hundreds of near-identical ad groups make it easy to miss what matters, and hard to spot wasted spend.
- A time-zone problem. The previous agency was in South America, so urgent questions from a fast-moving business took a day to answer.
None of this showed up as a crisis. The account was profitable. It just wasn’t as profitable as it should have been, in a market that was getting harder every month.
What we did instead
We rebuilt both accounts from the ground up around intent, not individual keywords.
1. Group by what people want, not the words they use
Instead of one ad group per keyword, we grouped searches by what the person was trying to do, such as researching a product, comparing costs or looking for a quote. Each intent got its own campaign or ad group, with enough search volume to give the bidding algorithm real data to work with.
2. Point the algorithm at the right target
An algorithm will optimise for whatever you tell it to. The old account told it to find clicks. We told it to find genuine enquiries and bookings, and we made sure those conversions were tracked properly and valued correctly.
3. Connect the CRM
This was the biggest single improvement. We connected the business’s CRM so that bookings and sales flow back into Google Ads as offline conversions. That means the account can now tell the difference between a lead that goes nowhere and a lead that becomes a high-value installation, and bid accordingly.
4. Fix what happens after the click
Around nine in ten leads came from mobile, so we tested landing pages built for phones: faster, shorter and with the enquiry form front and centre. A better landing page lifts every campaign at once.
5. Move money to where it works
With cleaner data, the patterns became obvious. We moved budget away from expensive regions, devices and days of the week, and towards the ones that produced the most efficient leads.
6. Report on revenue, not clicks
Monthly reports now lead with leads, cost per lead, sales and return on ad spend, backed by a live dashboard. And because we’re UK-based, questions get answered the same day.
What happened
The market didn’t help. Across UK home improvement, cost per click rose by 10–18%, cost per lead by around 10%, and conversion rates fell by around 15%.
The account went the other way:
- Conversion rate rose from 2.4% to 4.1%. The account now needs 42% fewer clicks to produce a lead.
- Annual spend fell by around £400k, from £1.59m to £1.16m, with cost per lead held flat.
- Return on ad spend improved to 3.88x, in a market where most advertisers were paying more for worse results.
- More than 20,000 leads and £4.65m in revenue across 2025.
Clicks got more expensive, but they worked much harder.
Should you scrap your SKAGs?
Not automatically. SKAGs, or something close to them, can still earn their keep if:
- your account is small, with very few high-value keywords
- you’re in a heavily regulated sector where ad wording must match the search exactly
- you rely on manual bidding for a good reason, and have the time to manage it
But if you’re spending serious money with smart bidding, and your conversions are scattered across hundreds of ad groups, it’s worth asking whether your structure is helping Google learn, or getting in its way.
A quick checklist for your account
Take ten minutes with your Google Ads account and ask:
- How many conversions does each campaign get a month? If most get fewer than 30, your bidding is probably starved of data.
- What are you optimising for? Clicks, or real enquiries and sales?
- Do you know which leads became customers? If sales don’t flow back into Google Ads, the algorithm is guessing.
- What happens after the click? Look at your landing pages on a phone. Would you fill in the form?
- Can you explain your reports? If your monthly report leads with impressions and click-through rate, ask where the revenue is.
If you’d like a second opinion, we’re always happy to take a look at your account. You can see how we approach paid media and lead generation, or read the full case study. We’ll tell you honestly what we’d change, and what we’d leave alone.
