When to question your PPC agency's measurement and reporting

What to take away

Search measurement changed shape in the last two years, and the change was structural rather than cosmetic. Google removed average position from reporting, tightened how conversion actions are counted, and pushed advertisers towards value-based bidding. The IAB UK also refreshed its search creative guidance, while the Advertising Standards Authority kept ruling on claims made in paid search copy.

The practical effect for anyone buying paid search management is that the old scoreboard has gone. An agency can no longer be judged on position, and should not be judged on click-through rate alone. It has to be judged on what it records, how it attributes it, and how honestly it reports the limits of both.

This guide sets out the questions that separate a measurement-led agency from one sending a monthly slide deck.

Where agencies get measurement wrong

Most measurement failures are not analytical. They are administrative. A tag fires twice, a form thank-you page gets indexed, a phone number goes untracked, and reported cost per lead drifts from reality by a margin nobody notices for a quarter.

An illustrative tracking audit shows the scale. Suppose a conversion tag fires twice on a thank-you page that Google has indexed. The platform records 550 conversions a month. In reality, 250 paid form fills occurred, and 50 organic visitors triggered the same conversion. Reported cost per lead looks like £9 on a £5,000 media budget.

The true cost per paid lead is £20. Every budget decision built on the lower figure inherits the error.

Agencies that audit their own tracking before they audit their bids are rarer than they should be. The recurring patterns are collected in our review of common PPC measurement mistakes, which is worth reading before you sign anything.

There is also a cultural problem. Reporting tends to reward the metrics the agency controls. Spend, impressions and clicks are always available and always flattering in volume terms. Pipeline, margin and retention are harder to obtain and less flattering, so they appear less often.

Attribution windows cause a quieter version of the same problem. If one report counts a lead on the day of the click and another counts it on the day of the sale, the same campaign shows two different results.

Conversion tracking: what to verify

Start with the primary conversion actions in the account. Each one should have a named owner, a defined value, and a documented counting method. An agency that cannot explain its own counting settings is telling you something before you ask a hard question.

Ask for a test conversion from a clean device, with the timestamp recorded. Then reconcile that timestamp against the platform report and against your own customer relationship management system. If three numbers disagree, you have found the first thing to fix.

Ad Rank and the quality signals that replaced position

Google's explanation of how Ad Rank is calculated is worth reading alongside your agency's bidding rationale, because it sets out the auction inputs an agency can actually influence. Expect a clear account of expected click-through rate, ad relevance and landing page experience, and how each is being tested.

If the response describes bid adjustments with no mention of ad strength or landing page quality, the account is being managed as a spending machine rather than a performance channel. That is a legitimate choice for some businesses, but it should be a stated one.

Attribution: the choice that changes the number

Attribution is a modelling decision, and every model flatters some channels and penalises others. Last-click credit will always overstate the final search query. Data-driven models redistribute credit towards earlier touchpoints, which often means paid search looks less heroic.

What matters is not which model your agency picks, but that the model is named and held constant across periods. A model switched silently between months makes trend lines meaningless.

For a fuller comparison of the options, our guide to PPC attribution methods sets out where each model misleads. Read it before you agree a reporting specification, because changing attribution afterwards always looks like moving the goalposts.

What a good reporting dashboard contains

A dashboard is a decision tool, not a scoreboard. The test is simple: can a reader who did not run the campaign make a decision from it in under five minutes?

Good dashboards separate three layers. The top layer is commercial: leads, revenue, cost per acquisition, and performance against target. The middle layer is diagnostic: which campaigns, ad groups and queries moved. The bottom layer is technical: tracking health, disapprovals, budget pacing and the changes made that month.

Most agency dashboards stop at the middle layer because it is the easiest to populate. If yours does too, the fix is a specification rather than a new tool. A useful field list includes spend, clicks, conversions, cost per acquisition, revenue, return on ad spend, and budget pacing.

Refresh the commercial layer weekly and the technical layer after every site release. Our PPC reporting dashboard covers the fields and refresh cadence clients most often ask for.

Two further rules help. Every number should carry a comparison, whether against target, previous period or forecast. Every chart should carry a sentence explaining what changed and what the agency did about it. A dashboard without commentary is data, not reporting.

Cadence matters as much as content. A weekly check on tracking health and budget pacing often beats a monthly deck, because it catches breaks while they are still cheap to fix. Match the reporting rhythm to how fast you can act on the answer.

Decision table: matching measurement approach to situation

Claims, evidence and the UK rulebook

The measurement conversation does not stop at the dashboard. Anything an agency writes in a search ad is a marketing claim, and claims carry evidentiary weight. The Advertising Standards Authority publishes guidance on claims, endorsements and testimonials that applies to paid search copy as much as to broadcast advertising.

Separately, the consumer protection regime prohibits misleading commercial practices, including advertising that misleads by omission.

Yet the rules that bind advertisers sit in the consumer protection framework and the CAP Code, not the Climate Change Act 2008. Under CAP Code rule 3.7, an advertiser must hold documentary evidence for any objective claim before publication. If an agency claims a performance figure in ad copy, it should be able to show the working.

That standard applies to agency marketing too. For example, a pitch claiming an 8:1 return on ad spend needs the period, the attribution model, the spend included and the revenue definition. Evidence would be a client-approved report showing the revenue and spend over a named period, with the model stated.

Other paid channels, such as LinkedIn advertising solutions, have their own advertising policies, but the measurement discipline is the same.

Creative testing and measurement quality

Measurement and creative are usually treated as separate workstreams. They should not be. Ad copy determines click quality, click quality determines the data the bidding algorithm learns from, and poor data degrades everything downstream.

The IAB UK guidance on search advertising creative best practice is useful here because it frames creative as a measurable variable rather than a matter of taste. Agencies that run structured creative tests, with a stated hypothesis and a defined read-out period, produce better learning data than those that rotate headlines on instinct.

One caution: platform automation now generates and tests variants at a volume no human team can match. That is fine, provided someone still owns the claim being made. Automation tests copy. It does not verify that the copy is accurate.

Benchmarks: useful, dangerous, or both

Clients ask for benchmarks because they want to know whether their numbers are normal. That instinct is reasonable, but benchmarks travel badly. A cost per lead that is healthy for a specialist consultancy is disastrous for a high-volume retailer, and averages across sectors hide more than they reveal.

Segment before you compare. A figure blended across an entire account usually mixes brand and non-brand campaigns, which have different cost and conversion profiles. One blended number hides the comparison that matters most.

Use benchmarks as a sanity check on order of magnitude, not as a target. Where a benchmark is quoted, ask for the sample, the period and the definition. Our PPC benchmark research for England explains how these figures are usually assembled and where the comparisons break down.

A better internal benchmark is your own trailing twelve months, adjusted for seasonality and any material change in budget or tracking. That comparison is unglamorous and far more useful for decisions.

Questions to put to an agency at pitch

Ask for the reporting pack from a comparable account, redacted. Then ask who writes the commentary, how long it takes, and what happens when a month goes badly. The answer to the last question reveals more than any case study.

Ask what happens when tracking breaks. Every account has a broken fortnight eventually. The difference between agencies is whether the client hears about it from the agency or discovers it in their own customer relationship management system.

Ask how the agency measures its own contribution. If the answer is spend under management, you have learned that the incentive is to spend, not to perform. If the answer is a commercial metric the client cares about, the alignment is real.

Finally, ask what the agency would stop reporting if it could. A candid answer here is worth more than a polished dashboard. For context on where the market is heading, our PPC trends and outlook for England in 2027 covers the shifts clients are already budgeting for.

Common questions

Should a PPC agency report on impressions and click-through rate?

They can, but as diagnostics rather than outcomes. Impressions tell you about reach within the auction, and click-through rate tells you about ad relevance. Neither tells you whether the campaign made money, so neither should sit at the top of a client report.

How often should measurement tracking be audited?

At minimum quarterly, and immediately after any site release, tag change or consent management update. For example, a business running a £6,000 monthly media budget with a single untracked form could misreport roughly a fifth of its leads for a full quarter before anyone notices.

Is last-click attribution ever the right choice?

Yes, in short-cycle, single-touch purchases where the final click genuinely is the decision point. It becomes misleading when the buying journey involves research, comparison or multiple devices, which is most B2B and much of retail.

What should a client do if reported results and their own sales data disagree?

Reconcile the definitions first: date range, time zone, conversion counting and attribution window. If the gap persists after that, pause target-setting discussions until tracking is verified, because every subsequent decision inherits the error.