A Worthing-based accountant we spoke with last year was still running a Google Ads account built by a freelancer in 2015. Nine campaigns, over 400 exact-match keywords, most of them single words like "accountant" and "tax return". The account was spending £1,800 a month and generating four enquiries. That's £450 per lead for a service business where a decent lead should cost a fraction of that. Nothing in the account had been touched in years, because it had once worked, and nobody wanted to be the person who broke it.
That account is not unusual. It's the norm for a huge chunk of small business advertisers in the UK right now, and it's the reason this article exists.
Why 2026 Is Not 2016 for UK Small Business Advertisers
The single biggest shift in the last decade is that Google Ads now matches on the meaning behind a search rather than the literal words typed into the box. A decade ago, an exact match keyword only fired if someone typed that exact phrase, maybe with a plural or a typo. Today, an exact match keyword can trigger against a query that shares almost none of the same words, provided Google's systems judge the intent to be close enough. Tactics built around controlling literal wording simply don't behave the way they used to.
Plenty of local trades and service businesses across Sussex and further afield are still running account structures built somewhere between 2014 and 2018, because nobody has gone back in and rebuilt them. The account still spends, ads still show, so it gets left alone. Meanwhile the platform underneath it has changed almost completely.
Smart Bidding paired with broad match can now find converting searches that a human strategist would never have thought to type manually, phrasings, questions, local variations, all sorts of long-tail combinations. But that only happens if the account is structured and set up to let Google's automation do that job properly. Bolt broad match onto an old, rigid, manually-bid account and it does the opposite: it burns through budget on irrelevant clicks.
None of this would matter as much if clicks were still cheap. They aren't. Average cost-per-click in competitive local service categories such as electricians, dentists and solicitors has climbed steadily year on year, which means a wasted click in 2026 costs a lot more than the same wasted click did five years ago. Old habits that used to be merely inefficient are now genuinely expensive.
A quick way to see this for yourself is the Auction Insights report sat under the Insights and Reports tab in any campaign. Pull it for a competitive local trade campaign and you'll typically see five or six competitors overlapping on impression share, each one bidding the query up. Benchmark data for UK home services searches such as "boiler repair" or "emergency locksmith" now puts average CPC somewhere between £4 and £9 a click, more than double what the same terms cost in 2016. Run the numbers on that Worthing accountant's account from the opening example and the maths gets uglier still: £450 per lead on a service where a well-run local campaign should be landing leads at £40 to £80.
This article walks through five specific habits that used to count as best practice, why each one now works against you, and exactly what to replace it with.
Retire Single Keyword Ad Groups and Build Themed Ad Groups Instead
Single Keyword Ad Groups, usually shortened to SKAGs, had their heyday roughly between 2012 and 2018. The idea was simple and, at the time, genuinely useful: put one keyword in its own ad group, write an ad that echoed that exact keyword back word for word, and you got granular control over messaging and Quality Score. If someone searched "boiler repair Worthing", they saw an ad that said "Boiler Repair Worthing" and nothing got diluted by irrelevant traffic sharing the ad group.
The problem in 2026 is that this structure fights against the way the platform now works. Smart Bidding needs volume and conversion data at the ad group level to learn what a good click looks like. A SKAG structure, by design, starves each ad group of the impressions and conversions it needs to learn properly. You end up with dozens of ad groups each getting a handful of clicks a month, none of which ever accumulates enough signal for Google's algorithms to optimise anything.
It also multiplies your own workload without a matching return. Splitting traffic this thinly means more ads to write, more bids to check, more search terms to review, for a level of matching nuance that Google's own systems now handle automatically. Defining Digital Academy lists Single Keyword Ad Groups as one of the clearest examples of a tactic that has aged badly, and the reasoning holds up: you're doing manual work to solve a problem the platform has already solved better.
The replacement is a tightly themed ad group of somewhere between 5 and 20 closely related keywords, feeding into 2 to 3 Responsive Search Ads per ad group, each one built out with call assets, sitelinks and structured snippets. Group by the same underlying intent rather than by exact wording, and let Responsive Search Ads test headline and description combinations against each other automatically.
A worked example makes this concrete. A Worthing plumbing firm running a SKAG structure would have had "emergency plumber Worthing", "24 hour plumber Worthing" and "urgent plumber near me" sitting in three separate ad groups, each with its own ad, its own bid, its own thin trickle of data. Rebuilt properly, all three keywords sit in a single "Emergency Plumbing Worthing" ad group, feeding two or three Responsive Search Ads that reference urgency, availability and location across their headlines. The ad group gets three times the impressions to learn from, and the ads can rotate to find whichever combination of headlines actually converts.
One thing that trips people up when they finally do this consolidation: Quality Score data resets to some degree when keywords move into new ad groups, and it's common to see a two to four week dip in average position or cost per click as the account relearns. That dip scares a lot of business owners into reverting back to the old structure before the new one has had a chance to prove itself. In the plumbing example above, cost per lead actually rose slightly in week one, from roughly £58 to £64, before falling to £24 by week six once the consolidated ad groups had gathered enough conversion data for Smart Bidding to optimise properly. Judging a restructure after seven days is one of the most common and most costly mistakes in this whole process.
Call assets deserve particular attention for trades businesses. A large share of conversions in plumbing, electrical work, locksmithing and similar trades still happen by phone rather than through a web form, especially for anything urgent. An ad group with no call extension attached is quietly losing conversions that never show up anywhere in the conversion data, because the customer just rang the number off the screen without clicking through to the site at all.
Stop Manually Bidding on Broad Match and Let Smart Bidding Do the Heavy Lifting
If there's one habit that drains small business budgets faster than any other, it's running broad match keywords on manual CPC bidding. Industry analysts have described this combination as the fastest way to exhaust a small budget on queries with zero commercial intent, and after seeing enough Sussex-based accounts up close, that description feels generous rather than harsh.
Here's why it goes wrong. Broad match used to mean "loosely related words in any order". It now means Google's algorithm deciding, based on everything it knows about intent and context, what counts as a relevant search. That algorithm has become smarter and more dangerous in equal measure. Paired with a proper Smart Bidding strategy like Target CPA or Maximise Conversions, in a well-structured account, broad match can find converting queries a human strategist would genuinely never have thought to bid on. Used carelessly, on manual CPC, with no automated bidding to guide it, it just burns through the daily budget on searches that have nothing to do with buying anything.
The fix is not to abandon broad match, and it's not to pick a single match type and apply it account-wide either. A portfolio approach across separate campaigns works far better for most local businesses.
- Campaign 1 runs exact match on branded terms and your highest-intent, best-converting phrases, where you need guaranteed visibility and tight control over spend.
- Campaign 2 runs phrase match on your core service terms, giving Google some room to expand around those phrases without letting things drift too far from what you actually offer.
- Campaign 3 runs broad match purely for discovery, under Target CPA or Maximise Conversions, deliberately left looser so it can surface new search variations you haven't thought of yourself.
Each campaign is doing a different job, and each one is judged on different terms. You wouldn't expect the discovery campaign to have the same cost per lead as the branded exact match campaign, and it shouldn't.
Value-based bidding is what makes this whole structure work properly rather than just theoretically. Rather than telling Google to chase conversions as a flat number, you tell it what each type of conversion is actually worth to your business. A phone call lead might be worth roughly £15 to a local trade business, while a contact form submission might be worth £8, reflecting the fact that phone leads tend to convert into paying customers at a higher rate. Feed those values in and the algorithm starts chasing quality rather than raw volume, spending more aggressively where the better leads are coming from.
None of this works if your conversion tracking is broken or half set up. Before switching any bidding strategy, check that your conversion actions are firing correctly and that Enhanced Conversions is switched on and working, matching hashed customer data back to conversions even when cookies or ad blockers get in the way. Smart Bidding is only ever as good as the signal you feed it. Flip the switch on value-based Smart Bidding with faulty tracking underneath and you'll optimise the account toward the wrong thing entirely, confidently and expensively.
There's a specific edge case worth naming here, because it catches out almost every business that switches to broad match for the first time: brand and irrelevant-intent leakage. A Worthing based driveway and patio company we looked at moved its core service keyword to broad match without setting up a negative keyword list first. Within ten days, the campaign was showing for "how to lay a patio yourself", "driveway cost calculator" and even "block paving courses Sussex", none of which were ever going to convert into a paying job. At an average CPC of £3.20 and roughly 40 of those irrelevant clicks a day, that's over £4,400 wasted in ten days on searches that were never buying intent to begin with. The fix is a shared negative keyword list, built once and applied across every campaign in the account, covering job-seeker terms, DIY and "how to" phrasing, course and training searches, and competitor brand names you don't want to pay to appear next to. Google Ads lets you build and maintain this as a single shared list under Tools and Settings, rather than duplicating the same negatives into every campaign by hand.
It's also worth saying plainly that decisions like these shouldn't be left to run on autopilot. Switching bidding strategies and rolling out broad match at scale are exactly the kind of account changes that should stay under manual human review rather than being auto-applied through Google's own recommendation prompts. Accepting a recommended change to pause a dead keyword or add a sitelink is low risk. Accepting a recommended change to switch your bidding strategy or expand match types across the whole account, without checking what it will actually do to spend, is how budgets disappear in a weekend.
Three More Outdated Habits Worth Retiring
Restructuring ad groups and match types fixes the biggest sources of wasted spend, but three more habits from the same era are still quietly costing UK small businesses money every month. Each one made sense under the old rules of the platform. None of them make sense now.
Sticking With Last-Click Attribution
For years, the default attribution model in Google Ads gave 100% of the credit for a conversion to the very last ad someone clicked before buying. That made a certain kind of sense when the customer journey was simpler and search behaviour was easier to track end to end. It makes far less sense now, when a typical purchase decision for anything above a low-value impulse buy involves several searches, several devices, and often a mix of branded and generic terms over a period of days or weeks.
Take a kitchen renovation company running search ads for a Worthing showroom. A typical customer searches something generic like "kitchen fitters near me" on day one, clicks through, browses, leaves. On day four they search the company's brand name directly and click a remarketing ad before finally booking a design consultation. Under last-click attribution, the brand search on day four gets 100% of the credit, and the generic ad from day one, the one that actually started the journey and did the hard work of winning attention against competitors, gets nothing. Look at that account through a last-click lens and the generic campaign looks unprofitable, a prime candidate for the chopping block, when it's actually the engine driving new customer discovery.
Google has been quietly retiring the older attribution models from the account settings menu, and data-driven attribution (DDA) is now the default for most new conversion actions, available to accounts of any size rather than being gated behind a minimum volume threshold as it used to be. DDA uses machine learning to look at all the paths that led to a conversion and all the paths that didn't, and distributes credit across the touchpoints based on how much each one actually influenced the outcome. Any account still manually pinned to "Last Click" in the attribution settings under Measurement should be switched over, and it's worth checking, because plenty of accounts set up years ago are still running on a model Google itself has all but abandoned.
Manually Setting Device and Location Bid Adjustments
Under the old bidding model, advertisers set explicit percentage adjustments for device type and location by hand: bid 20% higher on mobile, 100% lower on tablets, 15% higher within a five mile radius of the shop. These were sensible manual proxies for a simple fact, that conversion likelihood varies by device and location, at a time when the bidding system had no way of factoring that in automatically.
Smart Bidding now factors device, location, time of day, audience membership and dozens of other signals into every single auction automatically, adjusting the bid in real time for that specific searcher rather than applying a blanket percentage to an entire device category. A blanket manual modifier left over from the old system doesn't add precision on top of that, it actively fights it, capping how high the algorithm can bid for a genuinely valuable searcher just because they happen to be on a device the account owner deprioritised five years ago.
A common and expensive version of this mistake: a legacy -100% bid adjustment on tablets, set years ago when tablet traffic converted poorly, left untouched after the account moved to Target ROAS bidding. That -100% modifier doesn't get overridden by Smart Bidding, it excludes tablets entirely, full stop, regardless of how well they might convert now that most tablets are large-screen hybrid devices used for exactly the kind of considered browsing that leads to a booking or an enquiry. One Sussex based furniture retailer found, on auditing an old account, that tablet devices had been fully excluded since 2019, despite tablet traffic converting at a similar rate to desktop across the rest of their site.
The replacement isn't to remove all bid control and hope for the best. It's to feed the algorithm richer signals instead of blunt percentage caps: uploaded customer lists for remarketing and similar audiences, in-market and custom intent audiences layered on as observation, and, where the data supports it, audience-based bid signals rather than device-based ones. Check the account's bid adjustments under Settings and Devices for any modifier below -50%, particularly anything inherited from an account built before 2020, and question whether it's still doing the job it was originally set up for.
Treating Performance Max as an Untouchable Black Box
Performance Max campaigns get blamed for a lot of wasted budget, and sometimes fairly, but more often the problem isn't the campaign type itself, it's that the account never gave it anything useful to work with. Performance Max pulls in Search, Display, YouTube, Discover, Gmail and Maps inventory under one roof, and it needs asset groups, audience signals and, ideally, a product or service feed to have any real chance of finding the right customer in the right channel.
A common setup mistake looks like this: one single asset group covering every service a business offers, with a handful of generic images and headlines that could apply to almost any local trade, and no audience signal uploaded at all. Handed that little to work with, Performance Max defaults to chasing broad reach rather than qualified intent, and spend drifts toward the channels that are easiest to fill rather than the ones most likely to convert. The fix is to build a separate asset group for each distinct service line, kitchen fitting versus bathroom fitting versus general joinery, for example, each with its own images, headlines and, critically, its own audience signal built from existing customer lists or website visitors who've shown genuine buying intent.
The other edge case worth naming specifically: Performance Max campaigns cannibalising branded search. Because PMax can serve against a company's own brand name without that being obvious from the reporting, it's easy for a business to see a Performance Max campaign posting a strong return on ad spend and assume it's winning new customers, when a chunk of those conversions are simply capturing branded searches that would have converted anyway, for free, through organic results. Google Ads now allows brand exclusion lists to be applied at the account level specifically to stop this, under the Brand Lists section of account settings, and any account running both a dedicated branded search campaign and Performance Max should have that exclusion switched on, otherwise the two campaigns end up quietly bidding against each other for the same free traffic.
