Marketing ROI Benchmarks UK 2026: Why the Benchmark Does Not Exist

If you have ever asked what a good marketing return looks like for a business like yours, you will have been given a number. Spend 5 to 10 per cent of revenue. Aim for a 4 to 1 return on ad spend. Keep acquisition cost under a third of customer value.

Those numbers are not measurements of UK small businesses. We went looking for the research behind them.

It does not exist.

Executive summary

The number everyone quotes, and where it actually comes from

The “spend 5 to 10 per cent of revenue on marketing” rule traces back to two surveys. Both are worth knowing about, because neither is describing you.

The Gartner CMO Spend Survey 2026 found marketing budgets at 7.8 per cent of company revenue. It surveyed 401 marketing leaders across North America, the UK and Europe between January and March 2026. The vast majority of respondents work at companies turning over more than one billion dollars.

The CMO Survey, run by Duke University with Deloitte and the American Marketing Association, found 9.0 per cent in Spring 2026. It surveyed 308 marketing leaders. It covers US companies only.

The UK data measures direction, not level

The UK does have serious marketing data. It just does not answer the question you are asking.

The IPA Bellwether Report is the best-known UK marketing budget tracker. Its Q2 2026 edition found a net balance of plus 6.9 per cent, with 23.8 per cent of companies increasing budgets and 16.9 per cent cutting them. But Bellwether measures whether budgets went up or down, not how much anyone spends. Its panel is around 300 marketing professionals drawn primarily from the UK’s top 1,000 companies.

The AA/WARC Expenditure Report put total UK advertising spend at £46.7bn in 2025, up 6.4 per cent, with search at £17.9bn and social media at £11.5bn. That is the size of the market, not the share of a small company’s revenue.

The only UK figures we found on small business marketing budgets come from LOCALiQ, which surveyed more than 500 UK businesses: 20 per cent spend under £1,000 a year, 21 per cent spend £1,000 to £4,999, and 13 per cent spend £5,000 to £9,999. Worth knowing, but LOCALiQ sells marketing services to UK small businesses and does not publish its fieldwork dates or method. And spend in pounds cannot be converted to spend as a share of revenue without turnover data the survey does not collect.

What is missing, and why that matters more than what exists

There is no recurring, published, UK-specific survey of marketing spend as a percentage of revenue for small and medium businesses. There is no UK customer acquisition cost benchmark with a stated sample and method. There is no UK return on ad spend benchmark with a stated sample and method.

The official statistics that would let you compare yourself to your own sector are not collected in usable form. The ONS Annual Purchases Survey, which would give advertising spend by industry, last published a reference year of 2018. The ONS E-commerce and ICT Activity bulletin, still the most recent of its kind, covers 2019 data and was published in February 2021.

Three things you actually can benchmark, for free

These are official, current, sample-sized and published by government. Use them.

If you are in the 74 per cent of micro businesses not analysing their own data, the honest next step is not finding a better benchmark. It is measuring anything at all.

Why no two benchmarks ever agree: attribution is a setting

Here is the part almost nobody explains to a small business owner.

Google Ads counts a conversion up to 30 days after a click by default, and you can extend that to 90. Meta counts 7 days after a click by default. In March 2026 Meta also changed its own definitions, narrowing click-through attribution to link clicks only and replacing engaged-view with a broader engage-through category.

So two platforms reporting on the same customer will disagree, by design, before anyone has done anything wrong. And Meta’s mid-year change means its own year-on-year comparison broke.

Shortening a Google attribution window from 30 days to 10 does not correct anything. It simply makes every conversion that happened on days 11 to 30 disappear from the report. The change applies to future conversions only. Your history is not restated.

The same spend, four different answers

Ruler Analytics published a comparison of the same channels measured four ways. Under last-click attribution, TikTok, Facebook and Instagram prospecting all returned a ROAS of zero. Under media mix modelling, Facebook prospecting returned 4.1.

Same money. Same campaigns. An answer of nothing, or an answer of four times your money back, depending purely on which method you picked.

Note who published that table: a company selling attribution and modelling software, showing last-click producing zeros for the exact channels its product is designed to credit. Read it as an illustration of how much the method matters, not as a benchmark.

What the peer-reviewed research says about measuring ads

Two studies are worth knowing, because they are rare things: large, controlled, and published in serious journals.

Gordon, Zettelmeyer, Bhargava and Chapsky, in Marketing Science in 2019, compared attribution-style measurement against randomised experiments using 15 large advertising experiments, 500 million user-experiment observations and 1.6 billion ad impressions. Their finding: the observational methods often fail to produce the same effects as the randomised experiments, even after controlling for extensive demographic and behavioural variables. This was work done with Facebook’s own data and cooperation, and attribution still did not reproduce experimental truth.

Blake, Nosko and Tadelis, in Econometrica in 2015, ran large field experiments at eBay. Ads bought against eBay’s own brand keywords produced no measurable short-term benefit. Frequent users, who accounted for most of the spend, were unaffected, producing negative average returns. Their conclusion was that returns from paid search are a fraction of conventional non-experimental estimates.

The mechanism there applies to any business bidding on its own name: you are paying to reach people who were going to arrive anyway, and your analytics will credit the ad for every one of them.

The method that answers the question is not sold to you

There is a rigorous way to measure whether advertising caused anything. You split the audience, show ads to one group, withhold them from the other, and compare. Google sells this as Conversion Lift.

Google’s own eligibility note reads: Conversion Lift is not available for all Google Ads accounts, and to use it you should contact your Google account representative.

Modelling is not a safe substitute either. A Google research paper on media mix modelling notes that a typical dataset of three years of weekly national data contains only 156 data points, and gives a case study where five models all fitted the data well while their predicted sales differed by up to 50 per cent.

Every benchmark you find is measuring somebody’s customers

This is worth checking every single time, and it takes ten seconds.

• A conversion rate benchmark from an attribution software company is computed from businesses that already bought attribution software.
• An ecommerce benchmark from a platform is computed from merchants trading on that platform.
• An advertising click-through benchmark from an agency is computed from campaigns that agency manages.
• A cost-per-lead table from a cold email company tends to rank cold email as the cheapest channel.

In every case the businesses that tried the thing and left are not in the data. The benchmark is not the market. It is the publisher’s retained customers, which is a systematically better-performing group.

Even the professionals get the sum wrong

Two figures put the whole exercise in perspective.

The IPA reviewed the case studies submitted to its own effectiveness databank. Of 880 papers submitted since 1980, 39 calculated ROI correctly. The errors were relying on awareness shifts or response rates instead of financial payback, confusing total sales increases with incremental sales, and failing to control for things unrelated to the campaign. That is a 2009 audit of award entries rather than of all UK businesses, so read it precisely. But these were the industry’s showcase papers, written to win prizes, by people paid to do this.

Nielsen surveyed nearly 2,000 marketers worldwide, all managing budgets above one million dollars. 84 per cent said they were confident in their ROI measurement. 38 per cent measured traditional and digital marketing together.

Confidence and capability are not the same thing, and the gap is widest among the people most sure of themselves.

What to measure instead

If the benchmark does not exist, comparison to other businesses is the wrong goal. Comparison to your own last quarter is the right one.

• Enquiries, counted by hand, by source. Ask every caller how they found you and write it down. It is unglamorous and it beats every dashboard for a business under about ten staff.
• Cost per enquiry, not cost per click. Clicks are the platform’s metric. Enquiries are yours.
• Enquiry to customer rate. If this is low, more traffic makes the problem more expensive, not smaller.
• What a customer is worth over a year. Without this, no acquisition cost can be judged good or bad.
• The holdout test you can actually run. Turn a channel off for a month and watch what happens to enquiries. Crude, free, and closer to a real experiment than any attribution report.

Where this sits in the Whito framework

Benchmarking marketing ROI is a Scale-stage activity, and it is usually asked about at the Start stage.

If you are not yet recording where enquiries come from, no benchmark can help you, because you have nothing to compare. Start-stage measurement is one notebook and one question asked on every call. Build-stage measurement is cost per enquiry by channel. Scale-stage measurement is attribution, modelling and holdout testing, and it needs volume before it means anything.

Methodology

Every figure was read from the publishing organisation’s own page or paper on 19 August 2026, with sample sizes and fieldwork dates recorded where stated. Official UK statistics were taken from DSIT, DBT and ONS releases.

We excluded a large volume of widely repeated material that had no traceable source, including every UK customer acquisition cost benchmark and every UK return on ad spend benchmark we encountered. We also excluded figures locked behind membership walls or lead capture forms, on the basis that a benchmark you cannot check is not a benchmark.

Where a source has a commercial interest in its own finding, we have said so in the text rather than in a footnote.

Common questions

What percentage of revenue should a UK small business spend on marketing?

There is no UK research establishing a figure. The 5 to 10 per cent rule comes from US and global surveys of large enterprises, most with revenues above one billion dollars. It is not a small business benchmark and should not be used as one.

What is a good ROAS for a UK small business?

No UK dataset with a stated sample and method exists. The more useful question is whether the same campaign, measured a different way, would give a different answer. It usually does.

Why do Google and Meta report different numbers for the same campaign?

Because they count different things over different periods. Google Ads counts conversions up to 30 days after a click by default. Meta counts 7 days. Neither is wrong. They are answering different questions and presenting both as the truth.

How do I know if a benchmark is trustworthy?

Check three things: the sample size, the fieldwork date, and whether the publisher sells something that the finding makes look good. If any of the three is missing, treat the number as marketing rather than research.

What should I measure if I cannot benchmark?

Enquiries by source, cost per enquiry, enquiry to customer rate, and what a customer is worth over a year. Compare those to your own previous quarter rather than to other businesses. That comparison is real, and you control the data behind it.

Sources

Official UK statistics. DSIT UK Business Data Survey 2026, 4,450 businesses, published 18 June 2026. DBT Longitudinal Small Business Survey 2024, 8,396 SME employers. DBT Business Population Estimates 2025. ONS internet sales as a percentage of retail sales.

Marketing spend surveys. Gartner 2026 CMO Spend Survey. The CMO Survey. IPA Bellwether Q2 2026. AA/WARC Expenditure Report. LOCALiQ UK digital marketing statistics.

Measurement research. Gordon, Zettelmeyer, Bhargava and Chapsky, Marketing Science, 2019. Blake, Nosko and Tadelis, Econometrica, 2015. Chan and Perry, Challenges and Opportunities in Media Mix Modeling. Google Ads conversion windows. Google Ads Conversion Lift eligibility. Ruler Analytics ROAS by attribution method. Campaign on the IPA dataBANK ROI audit. Nielsen Annual Marketing Report 2024.

Related Whito research

For the spending side of the same question, see what UK marketing agencies actually charge, Google Ads and PPC costs and SEO costs. For the official picture of how UK businesses are set up online, see our UK website statistics.