How to compare and score PPC agency reviews explained

What to take away

If you run paid search for a Manchester retailer spending £30,000 a month, you have opened six tabs of agency review sites, found five-star ratings with no dates, and closed them again. This guide solves that.

Reviews are useful raw material, but only if you know what sits behind them and what to score yourself.

What do ppc agencies reviews actually measure?

Most published reviews measure client satisfaction at a point in time. That is a narrow signal. A client who joined three months ago and has seen cost per acquisition fall will rate highly. A client who inherited a messy account and is still fixing tracking may rate badly, even though the work is sound.

So read reviews as evidence of process, not outcome. Comments about communication, reporting cadence and who answers email tell you more than a star rating. Comments about results need a date, a spend level and a target to mean anything.

There is also a structural bias. Agencies ask happy clients for reviews. Unhappy clients leave quietly or post anonymously. Neither pattern is dishonest, but both distort the sample you are reading.

The review methodology used on this site sets out how we separate verified evidence from self-reported claims, which is worth reading before you build your own scoring sheet.

Where does review evidence come from?

Regulator rulings. The ASA rulings archive records ads that breached the CAP Code, including paid search copy. A ruling against an agency or its client is a documented fact rather than an opinion, and it tells you something about how that business handles compliance.

Platform documentation. Google's own explanation of conversion measurement is the benchmark for judging any agency claim about tracking quality. If an agency cannot describe conversion actions in those terms, its reporting is probably weak.

Market data. The IAB UK adspend research gives you the size and direction of UK search advertising, which helps you sanity-check claims about channel growth.

The data and sources behind agency reviews explains which third-party ratings come from independent research and which are commercial listings, which stops you weighing the two as if they were the same thing.

Reference calls. Two calls with clients who left the agency are worth more than twenty public reviews. Ask why they left and what happened in the final month.

One more check sits outside the review sites. Company filings show whether an agency is growing or leaning on one large client, and a change of directors shortly before you sign is worth a question. It takes five minutes and the record is public.

How should you score two shortlisted agencies?

Build a simple weighted sheet before you speak to anyone, so the conversation fills gaps rather than setting the agenda.

A workable weighting for a mid-market advertiser in England:

Score each out of five, multiply by the weight, and total. Adjust the weights to your own risk. A business with fragile tracking might put 35% on measurement and drop fee to 15%, because a cheap agency that reports badly is not cheap.

The exercise forces you to compare like with like instead of comparing two polished pitches.

The selection checklist for PPC agencies turns each row of that table into questions you can put in writing, which makes the answers comparable across bidders.

What does a fair fee look like in pounds?

Fees are the least transparent part of most reviews, so work from a labelled example.

A Leeds ecommerce team spending £40,000 a month on search might be quoted 12% of spend, or £4,800 a month, with a £2,500 monthly floor. A second agency quotes a flat £3,500 a month.

The flat fee is cheaper on that spend. It does not scale down if the team cuts spend to £15,000 in a quiet quarter, when it becomes 23% of spend.

A third common structure is a fee plus a technology or reporting charge. Ask for the total annual cost in pounds, not the headline rate. Then model it at your lowest expected monthly spend, not your average.

Watch for annual uplift clauses. Some contracts tie the fee to CPI or a fixed percentage each year, which changes the total cost you are comparing. Ask for the year-one and year-two cost in pounds before you score the fee row.

Agencies working with smaller budgets often apply a floor of £1,000 to £2,000 a month. Below roughly £5,000 monthly spend, percentage models stop working for the agency and the floor dominates. That is not unreasonable, but it should be stated up front rather than discovered in month two.

How do you verify claims in a review?

Verification is mostly about access and dates. Three tests cover most of it.

Ask who audits the tracking, too. A short independent measurement review catches duplicated conversion actions and double-counted leads, both of which flatter reported performance.

Where a claim involves personal data handling, for example audience uploads or call tracking, the agency should be able to describe its lawful basis. The Human Rights Act 1998 is the historical starting point for how personal data obligations developed in the UK, and it is a useful prompt when an agency claims ignorance of data rules.

What should you compare beyond the review scores?

Two agencies can score identically and still be very different suppliers. The differences that bite are contractual.

Data ownership. Who owns the conversion data, the audiences and the historical reporting when you leave? Insist on client ownership in writing.

Notice period. Ninety days is common and can be costly if performance dips. Thirty to sixty days is more balanced.

Account continuity. Ask what happens if your account lead resigns. A named backup with a documented handover is a reasonable ask.

Reporting. Monthly dashboards are standard. Weekly written commentary on what changed and why is a better signal of engagement than a prettier dashboard.

Certification. Individual certifications show a baseline of platform knowledge. The LinkedIn Ads certification programme is one example of a structured route, and it is worth asking which certifications the named team hold and when they were last renewed.

A side-by-side product comparison of PPC agencies in England is useful here because it forces the same criteria onto every bidder rather than letting each pitch define its own success measures.

Worked example: scoring three agencies

A Bristol software firm spends £25,000 a month on search and has shortlisted three agencies. All three have strong public reviews. Here is how the scoring sheet separates them, using illustrative figures.

Agency A and Agency B tie. Agency C edges ahead on fee and contract terms despite the weakest continuity score.

On annual cost the spread is wider than the scores suggest. Agency A quotes 15% of spend: £3,750 a month and £45,000 a year. Agency C quotes 11% with a £2,000 floor: £2,750 a month and £33,000 a year.

That £12,000 annual difference is the reason to score fees explicitly rather than letting a review score decide. It is also the reason to stress-test the model at a lower spend: if the firm cuts to £12,000 a month, Agency C's floor means the effective rate rises to about 17%.

The firm chose Agency C, negotiated a 45-day notice period and a written data ownership clause, and kept Agency B's account lead as a named contact in case the relationship needed revisiting. None of that came from a public review.

Which contract terms cause the most disputes?

Three clauses generate most of the friction we see described in client accounts.

Rolling terms with long notice. A twelve-month initial term with 90 days' notice means you are committed for fifteen months in practice. Ask for the total minimum commitment in months, written plainly.

Performance clauses with vague targets. A clause promising "improved ROAS" without a baseline, a measurement window and a remedy is not a performance clause.

Data and access on exit. Confirm that conversion tracking, audiences and historical reports transfer to you, and that access is removed only after handover.

Ask for the clause you would rely on if performance fell short, and check that it names a remedy rather than an intention. If an agency resists any of these, that is itself information. A confident supplier will accept a clean exit clause because it expects to keep the client on results.

How do you keep reviewing after you sign?

The review process does not end at signature. Set a quarterly internal review with three questions: is spend within the agreed plan, is the measurement plan still accurate, and is the named team unchanged?

Book the first review date at signature rather than three months later. A quarterly rhythm is easier to hold when the date already exists in the diary.

Keep a simple log of every material change the agency makes and the reason given. After a year this gives a far better evidence base than any public review site. You can then renegotiate from data, not sentiment.

For context on where paid search is heading as you plan that cycle, the trends and outlook for England in 2027 covers shifts to build into your next review.

If you are comparing agencies for the first time, decide early which third-party ratings rest on independent research and which are commercial listings.

Common questions

Are PPC agency reviews trustworthy?

Partly. Reviews from named clients with dates, spend levels and specific outcomes are useful. Undated five-star ratings on directory sites are usually marketing. Cross-check anything important against a reference call or read-only account access.

How many agencies should I shortlist?

Three is usually enough to get a real spread of fee models without the process dragging on. Score them on the same weighted sheet so the comparison is like for like.

What fee should I expect to pay in England?

For example, a team spending £40,000 a month might pay 12% of spend, which is £4,800 a month, often with a floor of £1,000 to £2,500. Flat fees exist and can be cheaper at high spend but costlier if you reduce budget.

Can I ask for read-only access before signing?

Yes, and it is a reasonable request. If an agency declines for client confidentiality reasons, ask for dated, redacted screenshots instead and a named referee you can call.