3 ways Ofcom and IAB UK data get misread as paid-social benchmarks

What to take away

Why Ofcom and IAB UK data are not campaign benchmarks

A benchmark is a comparison that holds everything except the thing you are testing constant. Most UK agencies borrow a published figure, drop it beside a client's result and call it a benchmark. That move fails on the first question a client asks: compared with what, and for whom?

Ofcom publishes research about communications markets and audiences. It tells you how many people use a platform, how often, and on what device. It does not tell you what a click cost, what a purchase was worth, or how a creative performed in a specific auction.

Those are commercial outcomes, and Ofcom is not in the business of publishing them.

IAB UK is a trade body for the digital advertising industry. Its output is standards, guides and market commentary. Useful for how to buy, how to assure quality and how to describe a placement. Silent on what your campaign should return.

Confusing the two turns a process document into a performance promise, and that promise will not survive a quarterly review.

The deeper problem is population. A national audience survey covers everyone in the country. Your campaign covers a defined audience, in a defined period, bought at a defined price. The gap between those two populations is where most benchmark claims quietly break.

This is the same trap covered in our note on why ONS, Ofcom and Google Ads figures make poor paid-social benchmarks.

There is also a legal and regulatory layer. The Advertising Standards Authority enforces the CAP Code, which requires evidence for comparative claims. If you tell a client their cost per acquisition beats a sector average, that average needs a source and a method.

Borrowing a headline figure from a survey is not evidence. It is a coincidence dressed as a standard.

Procurement differences sharpen the point. A campaign bought by a central team in Greater London, a devolved public body in Scotland and a manufacturer in the West Midlands will face different approval routes, different VAT treatment through HMRC rules and different reporting expectations. One national average cannot describe all three.

What Ofcom datasets actually measure

Ofcom's core consumer research tracks take-up, usage and attitudes across broadband, mobile, landline, television and online services. Its Online Nation and Communications Market Report series are the ones agencies most often cite. Both are about behaviour and access, not about advertising return.

Take a typical claim: our audience is highly engaged on social platforms. Ofcom data can support a statement about how many UK adults use social media and how often. It cannot support a statement about how many of those adults will respond to a paid social ad for a specific product at a specific price.

Engagement in Ofcom terms usually means reach or frequency of use. Engagement in media planning means an action tied to a commercial outcome. The same word, two different measurements. When the two get merged in a deck, the client hears the second meaning and the analyst meant the first.

Ofcom also reports on device ownership and connectivity. Useful for format planning, for example whether a video-first placement is sensible for a given audience. Not useful for setting a target cost per thousand, because it says nothing about auction dynamics or inventory supply.

Regional splits are another source of trouble. Ofcom publishes figures for England, Scotland, Wales and Northern Ireland, and sometimes for regions such as the North West or Yorkshire and the Humber. Those splits describe populations. They do not describe media costs, which vary by competition for inventory rather than by nation.

Use Ofcom for context, sizing and audience description. Quote it as context in the same sentence, so nobody mistakes a usage figure for a performance figure. If a client asks what Ofcom says your campaign should cost, the honest answer is that Ofcom does not say anything about that.

What IAB UK standards and guides actually cover

The IAB UK has published industry standards since the late 1990s and describes itself as the trade body for the digital advertising industry in Our story | IAB UK. Its remit is the health of the market: measurement conventions, ad quality, supply chain transparency and buyer guidance.

The Gold Standard is the clearest example. The IAB UK Gold Standard letter | IAB UK sets out requirements for ad quality and supply chain certification. It tells an agency what good practice looks like when buying. It does not tell an agency what a good result looks like once the buy is live.

Programmatic buying guidance is similar in shape. The Back to Basics Guide to Programmatic | IAB UK explains how programmatic trading works and what buyers should check. It is a map of the buying process. It is not a set of expected returns.

This distinction matters for agency pitches. Standards compliance is a claim about method. It says the agency buys in a way the industry recognises. It says nothing about whether the campaign will hit a target return on ad spend, because no standards body can promise that for an unknown product and price.

Guides also age. Platform features change, auction rules change, and privacy signals change. A guide written for one buying environment can be quoted years later as if it described today's. Check the publication context before you cite it, and say which version you used.

Where IAB UK material is genuinely strong is in definitions. If you need to agree what counts as a viewable impression or how to describe a placement, the trade body's language is a reasonable common ground. Use it to align terms, not to set targets.

The distinction between method and outcome runs through our guide to paid social measurement with reconciled sources.

The ONS series agencies quote out of context

The Office for National Statistics publishes the UK's official economic statistics. Two parts of its output get pulled into media decks more than any others: the digital economy material and the inflation and price indices.

The digital economy pages cover the shape and size of the UK's digital sector, including business activity and trade. They are useful for market context and for describing how large a sector is. They are not a measurement of advertising performance, and they do not track campaign costs.

Read them as background on Digital economy - Office for National Statistics.

The inflation and price indices are the more dangerous borrow. Agencies sometimes use a consumer price index to inflate an old cost per acquisition, or to justify a year-on-year rise in media cost.

The Inflation and price indices - Office for National Statistics series measures changes in prices across a basket of goods and services. Media auction prices are not that basket.

Auction prices move with demand, competition, inventory and platform changes. A general price index moves with the cost of the goods and services it samples. The two can point in opposite directions in the same quarter. Adjusting a media metric by a general index produces a number that looks rigorous and means nothing.

The second common misuse is timing. Index figures are published monthly and revised. A figure quoted from an early estimate can change. If a benchmark depends on a number that is later revised, the benchmark moves without any change in the campaign.

The third is geography and population. National indices describe the national picture. A campaign aimed at a single city, a single sector or a single age band is not the national basket. The denominator is wrong before any arithmetic begins.

Use ONS material for economic context in a client conversation: what is happening to household budgets, how a sector is growing. Keep it away from the performance line of the report. If a client wants to know whether media costs are rising, the answer comes from their own account history, not from a national price index.

Three specific misreadings and what each one distorts

These are the three patterns we see most often in UK pitch decks and quarterly reviews. Each one is a real misuse of a real source, and each one distorts a different part of the plan.

Misreading one: audience size treated as addressable reach

An Ofcom usage figure is quoted as the size of the audience a campaign can reach. The distortion is in the word reachable. A national usage figure includes people outside the targeting criteria, outside the buying platform's inventory, and outside the client's serviceable area.

What it distorts: budget. If the planner believes the addressable audience is larger than it is, the plan spreads spend too thinly and frequency never builds. The fix is to define the addressable audience from platform planning tools and first-party data, then use Ofcom only to describe the wider market.

Misreading two: standards compliance treated as performance assurance

An agency cites its Gold Standard certification as evidence that campaigns perform. The distortion is category confusion: a quality and process standard is presented as an outcome guarantee.

What it distorts: supplier selection. Clients end up comparing agencies on certificates rather than on measurement practice, incrementality testing and reporting honesty. Compliance is a floor, not a differentiator. Ask what the agency measures after the buy, not only how it buys.

Misreading three: a general price index treated as a media cost index

An ONS inflation figure is applied to last year's cost per acquisition to produce this year's target. The distortion is unit mismatch. The index measures consumer prices, not auction prices for advertising inventory.

What it distorts: targets. A target set from the wrong index either lets a weak result pass or fails a good one. The correct base is the account's own cost history by channel, placement and audience, with platform changes noted alongside it.

The table below sets out the three side by side, with the source each one misuses and the correction that holds up in a client meeting.

How to state a benchmark with a defensible denominator

A defensible denominator is a stated population, a stated period, a stated spend basis and a stated counting rule. If any of those four is missing, the number is a claim rather than a benchmark. Write all four into the sentence, every time.

Worked example. A deck claims a sector average cost per lead of fifteen pounds. That claim is unusable. The same figure becomes checkable when you add the population, the period, the spend basis and the counting rule around it.

Here is the full version. Across 42 lead-generation campaigns for UK financial services clients, run between January and June, with two million pounds of combined media spend, the median cost per qualified lead was fifteen pounds. A qualified lead means one accepted by the client's sales team within 30 days. Same headline number, now reproducible.

The four parts do different work. Population says who the number describes. Period says when it was true. Spend basis says whether the figure includes fees, platform charges and VAT. Counting rule says what event was counted and who confirmed it.

Build the benchmark in five steps:

A checklist for reviewing any benchmark before it goes into a deck:

Two more habits help. First, publish the range, not just the midpoint. A median with an interquartile range tells a client far more than a single figure. Second, date the benchmark and set a review point, because platform changes will move it. A benchmark without an expiry date becomes folklore.

This is also where internal alignment matters. Giving every paid-social metric an event, a denominator and an owner stops the same word meaning three things across strategy, buying and analytics. The benchmark conversation gets shorter when the definitions are already agreed.

Replacing borrowed benchmarks with first-party baselines

A first-party baseline is built from the client's own account history: past campaigns, past creative, past audiences, past prices. It is narrower than a national dataset and far more useful. It describes the actual system you are about to change.

Start with what exists. Pull twelve months of campaign data where available, and where the account is younger, pull everything and say so. Group it by objective, channel, audience and creative format. You are looking for the natural spread, not a single average.

Then set the baseline at the level you will actually manage. If you plan by audience and placement, build the baseline by audience and placement. A baseline at the wrong grain is just another borrowed number with better branding.

Where history is thin, borrow carefully and label it. A platform's own planning estimates, a sector figure from a client's trade body, or a figure from a comparable client account can all serve as a starting assumption. Mark each one as an assumption with a review date, and replace it as soon as your own data arrives.

Keep a written record of changes. Platform reporting definitions shift, attribution windows change, and consent rates move under UK GDPR and ICO guidance. When a baseline breaks, you want to know whether the campaign changed or the measurement did.

Finally, treat the benchmark as a planning tool and the baseline as a management tool. Benchmarks help you decide whether a plan is plausible before launch. Baselines tell you whether it is working after launch. Mixing them up is how agencies end up defending a number they never controlled.

For agencies working across the UK, the market itself is uneven. Client budgets, sector mix and buying maturity differ between Greater London, the North West, the West Midlands and the devolved nations, and public sector procurement rules differ too. That unevenness is why a single national figure rarely fits.

Our analysis of England's paid social agency market makes the same point about scale claims.

Build the habit into the reporting cycle. Every quarter, refresh the baseline, restate the denominator, and retire any external figure that no longer describes the account. A benchmark you can defend is worth more than a benchmark that sounds impressive, because only one of them survives the follow-up question.

Common questions