Paid social agencies review methodology: exercise and gates

A paid social agencies review methodology on paper only shows what an agency says it can do. A practical exercise shows what it produces. Give every agency the same synthetic, non-production case, then score only the outputs a reviewer can open, date and check against the brief.

What to take away

Test agencies on one synthetic, non-production case, not on case studies from unrelated work.
Score observable outputs onlypermission map, campaign build record, creative evidence file, measurement definition, incident response and exit pack.
Record each gate's requirement, evidence, owner, reviewer, status, limitation, expiry and retest trigger.
A written policy without a log is a hold, and an unreturned artefact is not evidenced.
A strong creative presentation cannot offset a missing account-control artefact.

Why a test case beats a deck

A synthetic case shows capability that a proposal can only claim. The method says not to infer capability from an unrelated case study, so the exercise replaces inference with outputs the reviewer can open, date and compare across agencies.

The case is non-live, so no real spend is committed, and every agency works within the same limits. That makes the outputs comparable on scope and responsibility rather than on how well each agency presents.

The exercise complements the paper review. The guide to reviewing agencies through evidence gates covers the evidence grid, while this method covers what happens when the agencies are asked to do the work.

Set up the case from the frozen brief

Build the case from the same frozen brief every agency already received. Nothing in the exercise should add scope, because a changed audience, platform, deliverable or media-payment arrangement is a departure that has to be marked.

Fixed brief elements

England audience and product boundary
Platform, account owner, media payer, spend authority
Creative formats, approval route, accessibility tasks
Tracking and audience-data design, outcome, reporting
Regulated-category, security, service, exit requirements

Issue the case to every agency with the same timetable, using the platforms named in the brief.
Ask them to map ad-account ownership and admin permissions.
Ask for a non-live campaign with stated audience, placements, exclusions and budget controls.
Ask for creative variants and a test plan.
Ask them to define tracking events, attribution and reporting.
Log every clarification for all agencies and freeze the accepted answers, attachments, source dates and proposal version.

Write the brief once and share it unchanged. A well-written brief, as covered in how to brief an agency without promising market growth, keeps the exercise about method rather than about optimism.

The brief elements every agency receives

The case is invented for testing and describes no real advertiser, so every agency answers the same facts. The table fixes each element; the buyer fills in the detail once and shares it unchanged.

Brief element	Synthetic case
Advertiser	A fictional online homeware retailer with one product range
Audience	Adults in England
Objective	Online sales of that range, with no live spend
Platforms	Meta and TikTok, as named in the brief
Accounts	The retailer owns the ad accounts; the agency states the admin rights it needs
Claims	One objective claim and one comparison in the ad copy, each needing evidence
Tracking	A purchase event, with consent obtained before any tag fires
Timetable and budget	One return date for every agency; one stated test budget, shown as a cap and never spent

The seven gates and the output behind each

Each of the seven gates has one output the reviewer can open, checked against the requirement written in the brief. Anything not returned is recorded as not evidenced, and it is never filled in from what the agency said in a meeting.

Gate	Output to return	What the reviewer checks
Account ownership and access	Permission map	The client's required admin access and a workable transfer or access-revocation route
Campaign build and delivery	Campaign build record	Targeting, exclusions, placements, budget controls and quality checks against the brief
Advertising compliance	Creative evidence file	Approval process, evidence for objective claims and verifiable comparisons
Creative testing and measurement	Creative evidence file and measurement definition	A documented test plan, defined events and attribution, a reporting approach that matches the brief
Lawful data handling	Measurement definition	Data flows, roles and tracking, with evidence of any consent required
Supplier security and incident response	Incident response	Controls for privileged access and a documented incident route
Exit and continuity	Exit pack	Account access, campaign records, creative and data assets, transfer steps and access removal

The measurement definition is the output most tied to data protection, because events, pixels and audiences each carry a data flow. Ask the agency to show where each one starts and who controls it.

For incident response, the buyer can pick one of the risk scenarios as the prompt, so that every agency answers the same event with the same documents.

Writing the gate record

Each of the seven veto gates in the table gets one line in the record, with eight fields: requirement, evidence supplied, evidence owner, reviewer, status, limitation, expiry and retest trigger. A line missing any field is an unfinished line, and it is completed before the review moves on.

Gate evidence and decisions

Gate

Identity and authority: Contracting name, company number, signatory
Account and finance: Client control, role matrix, media payer
Data and tracking: Purpose, roles, PECR position, transfers
Advertising: Claim file, comparison basis, disclosure
Security and continuity: Authentication, logs, leavers, recovery
Accessibility: Defined advert, landing, reporting tasks
Contract and exit: Scope, change, liability, renewal, exports

Minimum evidence

Identity and authority: Pass, hold or reject
Account and finance: Independent finance approval
Data and tracking: Independent privacy approval
Advertising: Independent advertising approval
Security and continuity: Independent security approval
Accessibility: Independent accessibility approval
Contract and exit: Independent commercial approval

Decision

Identity and authority
Account and finance
Data and tracking
Advertising
Security and continuity
Accessibility
Contract and exit

Require dated documents, not verbal assurances. A written policy without a log is a hold, and an unreturned artefact is not evidenced. Both statuses keep the gap visible in the record instead of hiding it behind a confident answer.

Expiry and retest trigger turn the record from a snapshot into a schedule. The expiry says when the evidence stops counting, and the trigger names the event, such as a change of platform or account structure, that reopens the gate.

Keeping reviewers independent

A gate is independent when its reviewer has no stake in another gate's outcome. Record the reviewer's name and the date on every line, and state who reviewed each output, so the record shows who looked at what.

Preserve defects as well as corrections. If an agency repairs its permission map after a first attempt, keep both versions. A strong creative presentation does not offset a missing account-control artefact, and the record should make that visible to anyone who reads it later.

Where outside rules apply

Platform approval is not UK legal clearance, so two gates need an external reference. For the advertising compliance gate, the ASA applies the CAP Code, and its misleading-advertising rules set the evidence test for objective claims and verifiable comparisons.

For the lawful data handling gate, tracking needs separate assessment under UK GDPR and PECR, as set out in the ICO's storage and access guidance. Record the reviewer and the artefact for each of the two gates.

Reading the results

First remove unresolved gate failures. Compare the agencies that remain on the same outputs, client effort, evidenced dependencies, whole-life cost and exit burden. If weighted scoring is necessary, publish the weights before opening bids and keep the underlying evidence.

No award is a valid result when no agency passes. The decision record should name the review owner, chosen scope, rejected alternatives, unresolved facts, conditions, review date and stop triggers.

Before you act

Write the case from the frozen brief, with no added scope.
Issue it to every agency with the same timetable and clarification log.
Ask for the listed outputs and no others.
Record one line for each of the seven gates, with reviewer name and date.
Mark anything unreturned as not evidenced.

Common questions

What should the practical exercise ask an agency to produce?

A permission map, a non-live campaign build record, a creative evidence file, a measurement definition, an incident response and an exit pack. Each is scored only on what a reviewer can observe, and each feeds a named gate.

How is a missing output recorded?

As not evidenced. A written policy without a log is a hold. The method says not to infer capability from an unrelated case study, so the gap stays in the record until the artefact arrives.

Can strong creative work make up for a failed gate?

No. A strong creative presentation does not offset a missing account-control artefact. Remove unresolved gate failures first, then compare the agencies that remain on the same outputs.