Researching rival paid social agencies with a dated evidence register

Route each research question to the evidence that can answer it, then keep the date and limits beside the result. The same record rarely settles identity, public visibility, an agency's own claims and its operating ability. Use the branches below to decide what to check next, and leave unresolved points visible rather than scoring them as facts.

If a review could change the shortlist, ask what it can support

If ratings may influence selection, treat each record as a prompt for follow-up, not proof of service quality.

Note the platform, page, access date, displayed denominator, and any visible account of collection or moderation. Keep allegations, verified transactions and supplier responses in separate fields so one category is not mistaken for another.

Guidance on online reviews can explain duties and prohibited practices, but it cannot authenticate an individual rating or reconstruct a particular transaction. Preserve what the platform displayed when you checked, then identify what remains unknown. If a record raises a material concern, seek an authorised response before relying on it in selection.

Rule of thumb: Let a review prompt a question, not settle it.

If adverts are the evidence you need, define the search before looking

If public advertising is under review, decide the country, advertiser identity and observation window first. Save the date and record only what the interface makes visible, such as creative, disclosure or run dates. A missing result is an unknown, not evidence that the advertiser never ran a campaign.

Meta Ad Library and Google Ads Transparency Center provide limited visible advertising records, not a complete media plan, targeting details, efficiency, sales or proof of which agency was responsible. For advertising rulings by the Advertising Standards Authority (ASA), note the decision date, media type and outcome. A ruling applies to the advert and circumstances it names; no result does not establish compliance.

Rule of thumb: Record what was visible, without assigning performance.

If the next question concerns delivery, request permission-based evidence

If public material leaves operating questions unanswered, request evidence from the shortlisted agency through an agreed and authorised process. Do not attempt to enter a competitor's or client's account. Agree how information will be examined, and use a demonstration with synthetic data rather than access that has not been permitted.

Where a request involves data protection, the ICO's guidance on data protection impact assessments can inform the questions to ask. Answers still need supporting evidence and do not replace testing. Keep delivery controls within the permissions agreed with the supplier.

Rule of thumb: Examine only evidence you are authorised to see.

If the comparison keeps changing, decide who qualifies before searching

If the candidate set is unsettled, define its boundaries before interpreting results. Record the decision, England geography, research date, service boundary and entry rule. Track screened, included, unresolved and excluded names so the denominator remains visible, including where public information does not resolve a candidate's status.

Names can come from a Companies House search on SIC 2007 code 73.11 (advertising agencies), the agency member list of a trade body such as the IPA, or the Meta and Google partner directories. Use search results to find possible candidates, not to estimate market share. Personalisation, location, wording and paid placement can shape what appears. Save each query and date, then check names against appropriate records. Unresolved candidates should stay in view; comparison methods cannot compensate for a comparison set whose boundaries were never established.

Rule of thumb: Set the denominator before drawing comparisons.

If contracting identity matters, separate the entity from its trading name

If identity could affect selection, match the trading name to a legal entity before comparing offers. Record the company register result and the date checked, while distinguishing the registered company from trading names and group relationships.

Leave sole traders or overseas entities unresolved when the register used does not answer the identity question.

Companies House Find and update company information provides public company records. A filing describes what appears on the register; it does not establish operational capability or financial health. Recheck filing dates before procurement, and pass unresolved identity questions to a qualified UK legal or commercial reviewer instead of filling gaps with assumptions.

Rule of thumb: A filing identifies a record, not an operating capability.

If an agency page supports a claim, keep the source attached

If the agency's own page is the basis for an offer description, present the information as a self-published claim. Save the exact page and observation date, attribute every public statement to its source, and preserve missing details rather than inferring scope, location, fees or methods.

A client logo without a date does not establish current work, consent, scope or performance. If a page changes, retain a lawful snapshot or describe the change without unnecessarily reproducing protected material. Keep each statement attributed to the source that made it.

Rule of thumb: Do not turn an agency's wording into an independent finding.

A dated evidence register you can copy

Bring the branches together in one register, one entry per agency, with a date on every field. The entry below uses a made-up agency to show the layout.

Compare agencies by putting the same fields side by side, and leave an empty field empty.

If evidence points in different directions, keep the decision gates apart

If the collected material is ready to inform a choice, place each source beside the question it can answer and its remaining limits.

Legal identity, agency claims, visible adverts, rulings, review records and operational controls are different evidence types. Combining them into one score can hide which points are established and which remain open.

Finish with a review methodology that lists identity, attributed claims, public adverts, rulings, review records, authorised controls, unknowns and the next check. Assign a reviewer to each relevant area before research informs selection, and retain the date attached to each observation. Where evidence is absent, record the unknown and the next action rather than treating absence as a negative finding.

Rule of thumb: Make the decision across separate evidence gates, not one league table.