Why programmatic advertising reviews need a scoring method

What to take away

Why scoring beats opinion

Most buyers inherit a platform because someone liked the dashboard. That is not a review. A review is a repeatable method that produces the same verdict when two people assess the same vendor.

Start by separating evidence types. Inputs are things a vendor controls before a campaign runs: contracts, data processing terms, certifications, fee disclosures, supply chain documents. Outcomes are things that happen after launch: delivery against plan, viewability, invalid traffic rates, brand safety incidents, reconciliation accuracy.

Use a gate-and-performance model rather than collapsing inputs and outcomes into one average. Inputs show whether controls and commitments are evidenced; outcomes show whether the vendor delivered. Decide in advance which input criteria are gates and the required floor for each: a failed gate cannot be offset by strong campaign outcomes. Compare the remaining input and outcome scores separately. If you apply weights to other criteria, record them before scoring and use the same weights for every vendor.

The programmatic advertising review methodology offers supplementary guidance. For a working rubric here, score each criterion from 0 to 4 using the evidence available for the review period, and record the evidence and reason for each score.

What UK rules apply

Advertising claims, including claims made by agencies about their own performance, sit under the CAP Code. The ASA rulings database shows how enforcement has landed on advertisers and platforms, and it is worth reading a handful of recent decisions before you accept a vendor's comparative claims at face value.

Data use is the second rulebook. When reviewing an agency or platform that handles personal data, assess its responsibilities under the UK GDPR, the Data Protection Act 2018 and, where relevant to direct marketing, the Privacy and Electronic Communications Regulations 2003 (PECR). Test three things: lawful basis, proper consent collection, and honoured suppression lists.

The ICO's direct marketing checklist is a practical due diligence reference for ad ops teams, and it takes about an hour to work through against a real vendor.

Supply chain transparency is the third area. The IAB UK Gold Standard letter sets out what signatories commit to on transparency and quality. A certificate is not proof of good practice, but a refusal to sign, or an inability to explain what the commitment means in practice, is a useful negative signal.

Building the scorecard

Use no more than eight criteria. More than that and reviewers start scoring inconsistently. A workable set for a UK buyer:

Use the same evidence rubric for every criterion: 0 means no evidence or an explicit refusal; 1 means an unsupported assertion; 2 means some relevant evidence, but it is incomplete, out of date or does not match the review scope; 3 means current, complete evidence supports the criterion; 4 means that evidence is complete and can also be corroborated through an audit trail or system records. Record why the evidence fits the score, rather than awarding points for presentation quality.

Verifying the corporate record

Before scoring commercial terms, confirm who you are actually contracting with. Trading names, group structures and white-label arrangements mean the entity on the insertion order is often not the entity you pitched.

Use the Companies House register to check incorporation date, filing history and any insolvency notices. A vendor that has filed late accounts for three consecutive years is a different risk from one with a clean record, and the difference rarely appears in a pitch.

Check the registered address against the address on the contract. Check the named directors against the people in the room. If the sales lead is not a director and has no signing authority, ask who does.

This step takes ten minutes and has ended more negotiations than any scorecard criterion. It also protects you if the vendor is part of a group that later restructures.

Reading an agency review

Agency reviews are harder than platform reviews because the product is people. The same agency can perform well for a retail client and badly for a regulated financial one, so a review without a stated client context is close to useless.

Our guide to programmatic advertising agency reviews in England explains how to read a review written for a different sector and decide which parts transfer. The short version: check whether the review names the trading entity, the period covered and the measurement source.

When you assess an agency yourself, ask for three named client contacts and a written scope for each. Then check whether the scope matches what you are buying. An agency that runs managed service for one client and a self-serve licence for another is not offering one product, and the review should reflect that.

Platform comparison traps

Platform comparisons fail in three predictable ways. First, feature checklists treat a shipped feature and a working feature as the same thing. Second, pricing comparisons mix rate card, auction dynamics and minimum spend. Third, reviews compare platforms at different scale, so a tool that works for a £5,000 monthly budget is judged against one built for £500,000.

Fix the first trap by asking for a live walkthrough of the specific feature, not a slide. Fix the second by normalising all pricing to a single illustrative monthly spend, for example a team paying £40,000 a month in media, and asking each vendor to model it.

Fix the third by stating your budget band in the review brief. For a structured way to do this across shortlisted vendors, see our programmatic advertising product comparison, which sets out how to normalise fees, minimums and measurement before scoring.

Evidence from training and certification

Certifications are weak evidence on their own, but they are useful for testing whether a vendor invests in its team. A certified individual is not a certified company, so ask how many people hold the qualification and whether it is refreshed.

LinkedIn's Ads Certifications are one example of a platform-run credential that signals baseline competence in paid media. Treat it as a hiring signal rather than a quality mark for the agency's programmatic operation.

Better evidence is a named team with defined roles: who sets up campaigns, who checks them, who has authority to pause delivery. If one person does all three, the review should note the concentration risk.

The supply chain check

Ask for the vendor's inclusion and exclusion lists, and ask when they were last changed. A list that has not moved in two years is either very stable or unmanaged, and the vendor should be able to tell you which.

Ask how invalid traffic is measured and what threshold triggers a credit. Ask what happens when a publisher is removed for quality reasons and whether historical reporting is restated.

Ask for a sample log-level file, even if redacted. Vendors that can produce one usually have the plumbing to answer harder questions later. Vendors that cannot will struggle when a brand safety incident needs a forensic answer.

Checklist for a review

Use this as a working document. Each line should have an owner and a date.

For how these checks map onto day-to-day campaign delivery, including handover and reconciliation, our programmatic advertising operations and delivery guide covers the operational side in more detail. It is the natural next read once the commercial review is done.

Turning scores into decisions

A scorecard is only useful if it changes what you do. Before scoring begins, set and record the minimum overall total and any criterion-specific floors, applying the same thresholds to every vendor in the review. There is no universal pass total: choose thresholds to fit the brief and risk, and do not let a strong overall score conceal a failure on a gate criterion. A vendor can score well overall and still fail on data protection, which should be disqualifying rather than a deduction.

Record the decision and the reasons. When the contract comes up for renewal, the original scorecard shows whether the vendor improved on the criteria you flagged. That is far more useful than a fresh impression formed in a renewal meeting.

If you are running a first selection rather than a renewal, start from our programmatic advertising selection checklist so the review criteria match the brief you issued to the market.

Common questions

How many criteria should a programmatic review score?

Six to eight works for most UK buyers. Fewer than six and you miss a material risk area such as data protection or supply chain documentation. More than eight and reviewers start applying the scale inconsistently, which undermines comparison across vendors.

Should a review score the platform or the agency?

Score both, but separately. The platform review covers technology, measurement and fees. The agency review covers people, process and the commercial relationship. Mixing them hides which party is responsible when delivery goes wrong.

How often should a vendor be reviewed?

At least annually, and always before renewal. A review is a snapshot of a period, so a score from eighteen months ago tells you little about current delivery. Shorten the cycle if the vendor handles regulated data or a large share of your media budget.

What if a vendor refuses to provide evidence?

Score that criterion zero and record the refusal. Missing evidence is itself a finding. If the refusal covers data protection terms or the contracting entity, treat it as a reason to stop rather than a reason to negotiate.