Google Ads in the United Kingdom: The 2026 Operator's Guide

The short answer

The single most consequential thing about running Google Ads in the United Kingdom is not cost, competition or format availability. It is that a meaningful share of what the account would normally learn is never collected, because UK visitors have to agree to advertising storage before it happens. Everything downstream inherits that: what your conversion count means, how fast audiences build, which search terms you can attribute to, and whether your numbers are comparable to anyone else’s.

That makes measurement design upstream of bidding, structure and creative in a UK account, which is the reverse of the usual order of work.

One thing I will not do is claim UK operating experience I do not have. My own accounts have run in other markets, and the UK specifics here rest on Google’s published country lists and policy pages, on the benchmark publisher’s own description of its method, and on cross-market practice that transfers. Where something could not be verified, it is flagged rather than filled.

What UK benchmark data actually exists

The UK is one of only two markets with a published paid search benchmark report built from observed campaign data — the other is the United States. That is genuinely unusual. Most markets, including large ones, have nothing that meets the standard at all.

The UK report is published by LOCALiQ UK as its Paid Search Benchmark Report, 2026 Edition, dated 8 April, drawn from campaigns running throughout 2024 and 2025 across 17 industries, covering average click-through rate, cost per click and conversion rate with year-on-year change. It is observed campaign data rather than keyword planning forecasts, which already places it above most of what circulates.

It is also considerably weaker than its US sibling, and the two are published by the same corporate family, which makes them look like equivalents.

Read the right-hand column as a whole. "Hundreds" of campaigns spread across 17 industries implies a few dozen campaigns behind each industry figure, against a stated floor of 52 per subcategory in the US report. Both are one agency reporting its own book of business, which skews toward that agency’s client mix — but only one publishes enough for you to judge how much weight its numbers carry.

Why there are no UK figures in this post

The report body sits behind a download form and I have not read it, so I cannot quote a UK cost per click by industry and tell you honestly where it came from. Rather than lift figures from a secondary summary and present them as sourced, there are none here.

If you want UK figures, download the report yourself and read its methodology page before its tables. The question to answer first is how many campaigns sit behind the row for your industry. If that is not disclosed, the row is a direction and not a measurement — and the same test applies to every benchmark table anyone shows you. The companion investigation into a market with no data at all shows what happens when nobody applies that test: figures get manufactured from forecasts and then cited as though they were measured.

Consent, operationally — and only operationally

I configure ad accounts, tags, consent signaling and audience lists. I am not a solicitor, this is not legal advice, and nothing below tells you what any law requires of your business. The UK regime here is UK GDPR alongside the privacy and electronic communications rules, overseen by the Information Commissioner’s Office, and the cookie rules were amended by legislation in 2025 with updated regulator guidance published in April 2026 — recent enough that most UK cookie material still in circulation predates it. Confirm your own position with a qualified UK solicitor. What follows is strictly what this changes about how an account is built.

The UK is inside Google’s own consent policy scope

Google’s user consent policy states its scope as the European Economic Area, the UK and Switzerland. Of the major English-language markets, the UK is the one where Google itself requires advertisers to obtain end-user consent for storage where legally required and for the use of personal data in ad personalization. The US, Canada and Australia sit outside that stated scope, so a consent configuration built for a North American account does not satisfy a UK one, and the reverse is over-restrictive.

The operational consequence of not implementing consent signaling is specific rather than vague: new UK users stop being added to remarketing and audience lists, and conversion modeling does not run. So the account does not merely lose some tracking — it loses the mechanism that would have partially compensated for the loss.

What happens to measurement when a meaningful share decline

When advertising storage is declined, the tags that would identify and attribute that journey do not set or read it. The visit still happens, the click was still paid for, and the conversion may still occur — it simply is not observed. Three things follow, and teams routinely conflate them.

• Attribution granularity degrades first. An unobserved conversion cannot be tied to a keyword or a search term, so your term-level reporting reflects a subset of outcomes rather than all of them.
• Bidding gets less signal. Automated strategies learn from conversions they can see, so a smaller observed set means slower learning and more volatility, particularly on lower-volume campaigns.
• Audience building slows. A visitor who declines does not enter a remarketing list, so lists grow more slowly than traffic and small advertisers can sit below the size needed to serve.

What modeling does, and what it does not tell you

Conversion modeling estimates the conversions that could not be observed, using the behavior of comparable journeys that were. It is a reasonable response to a real gap and makes aggregate reporting less wrong than leaving a hole.

What it does not do is tell you which click converted. A modeled conversion is an aggregate estimate, so it cannot be pushed down to a search term, an audience segment or a customer record. Two consequences: your search terms report describes only the observed portion of your outcomes, and any reconciliation against CRM records will legitimately fail to match your Google Ads conversion count. That mismatch is not a tracking bug, and I have watched teams spend weeks chasing it.

Why UK account data is not comparable to US benchmarks

This is the practical payoff of the whole section. The US benchmark figures are built from observed campaign performance in a market that runs on an opt-out privacy model, where the default is that measurement happens unless someone objects. A UK conversion count includes a modeled component that exists precisely because part of the journey was never observed.

So comparing a UK conversion rate to a US benchmark treats an estimate and an observation as the same unit. The comparison that works is a UK account against its own history, holding the consent configuration constant — and if you change that setup, mark the date on the chart, because your data changed shape that day rather than your performance.

The order of work in a UK account

Because measurement sits upstream, the sequence matters more here than in a market where the data simply arrives.

1. Fix the conversion definitions before anything else. One real outcome, counted once, no form views counted as submissions, no imported goals double-counting native tags. Consent complexity on top of a broken conversion setup produces numbers nobody can reason about.
2. Get the consent platform and the tag configuration talking to each other properly, including a defined default state for the period before a visitor has decided. A delayed drop is not a consent mechanism, it is a slower one.
3. Instrument the split. Record what share of sessions grant advertising storage, and track it weekly. This is the single most useful number in a UK account and almost nobody has it on a dashboard.
4. Only then tune bidding. Automated strategies inherit whatever signal quality the first three steps produced, and tuning them before that is tuning against noise.
5. Re-baseline after any consent change. Treat the change date as a break in the series rather than a data point in it.

Step one is the prerequisite most often skipped. The mechanics are in the conversion tracking guide for GA4 and Tag Manager, and the wider sweep is in the Google Ads audit checklist. In a consent environment this matters more, not less — the smaller your observed dataset, the more damage each miscounted event does to it.

Shopping requires a Comparison Shopping Service

This is the most commonly botched point in UK paid search writing. Google states that Comparison Shopping Services must be used to place Shopping ads and free product listings on behalf of merchants in a defined list of countries, and the United Kingdom is on that list. Three things follow.

• The requirement survived Brexit. The UK is named on a list Google maintains today; it did not lapse when the UK left the EU.
• The list is not an EU list. It includes Norway and Switzerland, neither of which is an EU member, and excludes several EU states. It tracks Google’s own program design.
• No merchant is ever blocked, because Google Shopping is itself a CSS and bids on behalf of the merchants it represents like any other. The decision is which CSS represents you, and merchants may work with several — though Google prohibits partial switching, where some sub-accounts are unlinked to move to another CSS.

Now the part that says no. A specific percentage CPC saving from third-party CSS partners is quoted constantly in UK agency marketing, and I could trace it to no primary or independent source. There are legitimate reasons to choose a third-party CSS — service, reporting, commercial terms — but a fixed-percentage performance uplift is not evidence. Ask for the sample, the period and the method behind any such number, and treat the absence of an answer as the answer.

The 2% nobody budgets for

Google applies a 2% jurisdiction-specific fee to UK Google Ads costs, in force since 1 November 2020 and still applied as of July 2026. It is a standing uplift on media, not a one-off, and it is routinely left out of UK budget models — which quietly understates every UK plan before a single auction runs.

Several comparable markets carry no such fee at all, and one removed a similar surcharge during 2025 with fees credited back. Two advertisers buying identical media in different markets can sit several percentage points apart on platform surcharge alone — larger than most bid-strategy gains. It belongs in the budget model as a line, not a rounding assumption.

Local Services Ads: available, and narrower than it looks

Local Services Ads run in the UK, which puts it in a small group of countries where the format exists at all. The coverage is home services — around 28 categories nationwide — plus a professional services group that is currently available in Greater London only, including several legal categories and estate agents.

Two things before planning around it. There are no health care, personal care, education, pet or financial-planning categories in the UK list, so several verticals that use LSA heavily elsewhere simply cannot. And some surrounding tooling — lead management, booking tracking, lead credits — is not available in the UK even though the format is. Plan for a thinner operational surface than US material describes.

Where Bing actually sits in the UK

Tracked search referral data for June 2026 puts Google at 91.19% of UK search referrals and Bing at 6.22%. On mobile alone, Google is at 97.60% and Bing at 0.49% — a collapse of more than an order of magnitude. Whatever Microsoft Ads is worth to a UK advertiser, it is a desktop proposition, and a mobile-weighted account should expect very little from it.

The same period puts the UK device split at roughly 51% desktop, 45% mobile and 4% tablet — the highest tablet share among comparable markets and the one place a device-bid conversation about tablets is arguably worth having. Read all of these as tracked referral shares rather than a census: the sample is pageviews rather than people, and it sees only searches that produce a click through to a site.

What I could not verify, and will not guess

Google’s tax-by-country documentation did not render a UK section across repeated attempts, including the UK variant of the page. So I am not going to state the UK billing entity or the VAT treatment of a UK account, even though the commonly assumed answer is easy to find repeated. Open your own billing settings in a live account and take the treatment question to your accountant. Where a source could not be reached, the gap is stated rather than filled — a guide that quietly guesses on billing is worse than one that admits it does not know, because a reader will plan against the guess.

Where UK accounts actually fail

• Consent implemented as a banner only, with no signal reaching the tags. The site looks compliant and the account behaves as though nothing changed, which is the worst of both outcomes.
• Modeled and observed conversions read as one number, so a degrading measurement setup is invisible until a CRM reconciliation forces the question.
• Bidding strategies tuned aggressively on a thin observed dataset, producing volatility that gets blamed on the algorithm rather than on signal volume.
• Remarketing strategies designed for a market with no consent gate — many narrow segments, long membership windows — which never reach serving size and are quietly written off as ineffective.
• US benchmarks used to set UK targets, which sets an unreachable goal measured in a different unit and makes a sound account look like a failing one.

What this post is, and what to verify yourself

This reflects the position as of July 2026. Platform availability, tax treatment and privacy rules change, and the UK cookie guidance changed within the last twelve months. The benchmark report is an annual publication and a newer edition will supersede what is described here. Treat every specific as a claim with a date attached, and check the primary source before it enters a client plan.

On the legal material: I have described only what these rules change about account configuration, because that is my competence and the boundary of it. Nothing here is legal advice, none of it tells you what any law requires of your business, and anything with consequences goes to a qualified UK solicitor before it goes into production.

Where to go from here

Put one number on your dashboard this week: the share of sessions granting advertising storage. Then split your conversion reporting into observed and modeled and watch both. Those two will tell you more about a UK account than any benchmark table, because they determine whether every other number means what you think it means. For the market where this resolves differently, the US guide covers a market with far better benchmark data and a privacy model that runs the other way round.

That is the standard I hold the work to at Arcetis, the growth systems practice I run: no number enters a client plan without its source, its sample and its date attached.

Related reading