How to assess software for influencer-agency selection

When an influencer agency directory listing mentions software, treat it as a claim to verify during shortlisting, not proof of a tested service. Assess the proposed software as one part of the agency's service: choose a tool for one defined job, not as a substitute for campaign governance. Choosing channel strategy tools does not establish what a particular agency uses or prove that a workflow suits the buyer.

This desk review was researched on 6 September 2026 for decisions revisited in 2027. No account, feature, integration, supplier response, price, accessibility journey or campaign outcome was tested. Capability descriptions are attributed to current first-party pages. The manual route remains available throughout.

If the proposed job is still vague

If the proposal cannot be tied to one authorised job, pause selection and write the operating problem so evidence can answer it. For example, can the buyer maintain an authorised creator list, approved claim versions and withdrawal records for one UK campaign while keeping client ownership of data and accounts?

Specify the permitted population, source data, users, output, review period and decision owner. A buyer might need creator discovery, relationship records, content approval, payments, reporting or an agency service. These are different functions, and a product page cannot prove data completeness, classification accuracy or workflow fit.

Keep the lanes separate rather than combining them into one score. A useful search function cannot compensate for missing rights, insecure access or an unsupported advertising claim. Describe the operating problem in terms that evidence can settle, using the governance problem as the starting point.

• Rule of thumb: define one job and its evidence before comparing tools.

If supplier pages are the main evidence

If the proposal relies on public product material, treat each capability as a claim to verify, not as proof of suitability. The products below illustrate questions to ask about an agency's proposed software; their inclusion does not amount to approval, and public material cannot establish a campaign result.

CreatorIQ describes its influencer-marketing solution as covering discovery, programme management and measurement. Its legal document index lists dated standard contract, privacy, security and subprocessor records. Those documents do not establish the configured edition, UK availability, price, data coverage, result quality or contract offered to this buyer. Review the public supplier records as evidence to examine, not a substitute for a test.

Traackr's platform page groups its service into planning, activation and measurement. The dated EU terms identify Traackr UK Limited for customers using that agreement and describe dependencies on third-party sources. The buyer still needs the order form, exact modules, accessible-use evidence, an export test and a UK-specific data assessment.

Kolsquare's product FAQ attributes discovery, campaign management and reporting functions to its platform. Its legal terms identify a French contracting company and describe account conditions. This does not show which terms, functions or support apply to a UK customer.

None of the pages independently verifies performance or a campaign outcome. Public customer quotations, awards and vendor-created metrics are not buyer evidence. No account, feature, integration or supplier response was tested in this desk review, so claims still need checking against the proposed configuration and contract.

• Rule of thumb: a feature description is a question for testing, not a finding.

If the manual route could meet the need

If a no-purchase route can meet the same acceptance criteria, keep it in the decision. A manual baseline may use a permissioned source register, buyer-owned contact log, controlled document storage, rights and disclosure matrix, payment ledger and versioned exports, maintained by named people.

Set identical acceptance criteria for the manual and software routes. Record time in a buyer-supplied unit, definitions for errors and missing records, permission boundaries, accessibility tests, incident handling and exit work. Do not invent staff costs or assume a spreadsheet is secure just because it is familiar.

A credible baseline can show whether the actual need is software, an agency operating service, more internal capacity or a narrower process. It also gives the buyer a route to return to if a supplier test fails. A manual option should be judged against the same requirements, not dismissed by default.

• Rule of thumb: keep the manual option unless it fails the same stated tests.

If discovery uses personal or inferred data

If creator discovery depends on personal data, inferred fields, public content or platform-derived observations, require documentation before using it. Ask about sources, collection basis, update interval, geography, eligibility, suppressed profiles, inferred attributes, correction rights and the effect of platform changes on coverage.

Traackr's June 2026 influencer privacy notice says its market-research database includes publicly available information and information supplied by influencers. Kolsquare's April 2026 privacy policy describes its processing for users and creator profiles. These first-party notices raise questions; they do not select the advertiser's lawful basis or validate every profile.

Public visibility does not make data unrestricted. The ICO data-sharing code distinguishes controller sharing from processor activity. A named privacy reviewer must map the advertiser, agency, tool supplier, creator and platform by actual purpose. Direct creator outreach also needs separate privacy and PECR analysis under UK GDPR, the Data Protection Act 2018 and the Privacy and Electronic Communications Regulations, all regulated by the ICO.

Do not infer a creator's age, protected characteristic, location, audience composition or suitability from a supplier label. Require source-level evidence, a correction route and human review. Treat automated relevance, brand-safety and sentiment outputs as unverified classifications until a defined sample has been tested against buyer-owned labels.

• Rule of thumb: do not use discovery data until its sources, basis and correction route are clear.

If content, claims, rights or payments enter the workflow

If the proposed software handles content or commercial activity, keep advertising approval and rights clearance as separate decisions. A platform may move briefs and approvals, but it cannot approve a product claim by itself; evidence for each decision must remain available beyond an opaque platform score.

Under CAP Code Section 2, recognition of a marketing communication is a separate question. Section 3 concerns misleading advertising and substantiation. Record evidence for both gates, and do not treat a workflow status or automated score as approval.

Record who represents the creator and who can license each asset. The government's copyright overview says use must rest on permission, ownership or a relevant exception. Retain source files and the terms for media, territory, duration, edits, paid amplification and withdrawal.

If a tool manages contracts or payments, establish the legal entity providing the function, who holds funds, supported currency and transaction unit, which records export, and how disputes or reversals work. Do not infer tax treatment from a software field; the tool's presence does not settle the underlying commercial or rights questions.

• Rule of thumb: keep claim substantiation, disclosure, rights and payment evidence distinct.

If control, access or exit is uncertain

If the buyer cannot demonstrate control over accounts and records, do not import data or commit yet. Map client-owned accounts, individual roles, privileged actions, authentication, logs, recovery and revocation, including how agency access can be removed.

The NCSC's supply-chain risk principles advise buyers to understand suppliers, subcontractors, protected information and access. A certificate is one artefact, not proof that the configured workflow is safe. Establish how the proposed arrangement handles each of those areas rather than relying on a general assurance.

Ask for an accessibility statement or conformance report for the exact product version, known defects and remediation dates. Test the buyer's actual tasks with relevant disabled users and assistive technology. No public document reviewed here establishes that any named product is accessible for this buyer.

Test exit before import. Define complete export fields and formats, identifiers, attachments, approval history, creator objections, deletion process, residual backups, subprocessor treatment, contract survival and account disconnection. The buyer needs usable evidence of recovery, not merely a download button.

• Rule of thumb: prove access, accessibility and usable exit before any real import.

If the price cannot be compared on the same basis

If there is no like-for-like commercial evidence, do not infer a UK price from public records. No current public record reviewed here supports a comparable UK price across CreatorIQ, Traackr and Kolsquare; ask for a dated quote and order form for the exact edition.

The quote should specify modules, users, creator or campaign units, storage, support, implementation, currency, VAT basis, commitment, renewal, overage and exit. Record these items separately so that a bundled offer does not conceal which service or risk sits inside the price.

Keep agency service charges separate from software licence, creator consideration, gifted products, paid media, internal labour, specialist review and tax. A bundled price can still be recorded, but mark which work and risk it includes and which remain outside it.

Use blank thresholds until the buyer supplies its requirements. Examples include maximum permitted unresolved records, required export fields, restore-test acceptance and review turnaround. Publishing guessed figures would turn procurement criteria into fiction.

• Rule of thumb: compare dated offers by unit, scope and exclusions, not headline price.

If a route passes the evidence gates

If a route meets the buyer's requirements, record how each requirement was tested before implementation. Send the same evidence request to every supplier and the manual owner, and reject any route that fails a non-compensating legal, privacy, security, accessibility, rights, commercial or exit gate.

Use synthetic creator and campaign records in a controlled environment. Test roles, approvals, exports, deletion, error correction and recovery. Do not upload real creator or customer data until privacy, contract and security decisions are complete. Do not activate platform integrations until the account owner can revoke them.

If a route passes, run it beside the manual baseline for a buyer-defined period. Reconcile record counts, missing fields, changed versions and exports. This is an operational acceptance test, not evidence of campaign performance, and the buyer should record the results against the criteria set earlier.

Where an agency proposes to operate the software, test client access to a supplier-managed workspace separately from a client-owned workspace with delegated agency users. Record who controls the tenant, licences, creator records, platform connections and exports in each arrangement. A service contract cannot rescue a design that prevents the advertiser recovering its evidence.

The selection record may conclude no purchase or insufficient evidence. Reopen it after a product, subprocessor, data source, platform permission, contract or control changes. Keep publication and procurement on hold until named specialists approve the actual supplier dossier and configured use.

• Rule of thumb: approve only the tested configuration, with a recoverable manual route beside it.