Before you shortlist branding agencies: decide whether to commission at all

Strategy work before you hire one of the branding agencies in a directory should begin with a decision that can be disproved: does the organisation need external branding work at all? An organisation in England needs to say what should change, for whom, compared with what, and which evidence would cause it to keep the current arrangement. Choosing a visual style or a supplier before answering those questions is premature.

This guide sets out a planning method for 2027, not a market forecast. Research closed on 5 September 2026. No agency was engaged, no client work was observed and no buyer survey or campaign was run. Official data can describe broader business populations and rules, but it does not reveal an England-wide appetite for branding work.

What to take away

Frame one decision

Write the decision in a form the accountable owner could sign. "Improve the brand" is not enough. A usable record identifies:

Keep aspiration separate. Purpose, values and a creative brief may matter later, but they do not replace a decision question. A team considering whether to change how one service is named has a different evidence problem from a group considering a corporate identity after a merger. Their audiences, risks, rights and acceptable measures cannot be assumed to match.

The Green Book from HM Treasury describes appraisal as comparing costs, benefits and risks across options for government objectives. It is mandatory in specified UK public-spending settings, not for every private branding commission. A commercial team can nevertheless borrow two useful disciplines: consider a sufficiently broad option set and document why choices advance or fall away.

Describe the audience as known, assumed or unknown

Use a documented research design rather than inventing a customer portrait. Every audience statement should carry one of three labels:

GDS guidance on planning user research advises teams to identify research questions and choose methods that answer them. The page concerns UK public services. It can inform a neutral research design, but it does not show what commercial brand buyers believe.

Do not quietly widen the geography. The ONS UK Business: activity, size and location 2025 includes industrial and regional tables for VAT or PAYE registered enterprises and local units.

Those data help describe a sampling frame, subject to definitions, but do not isolate brand owners, buying intentions, budgets, agency use or demand. Unregistered activity is outside that dataset's registered-business frame.

The baseline may instead come from lawful internal records. Define the unit before extracting anything. An enquiry, qualified opportunity, signed brief and completed commission are different states. Record the originating system, query version, owner, period, time zone, inclusion rules, exclusions, duplicates, missing fields and later corrections. If a definition changed during the period, keep the versions separate.

State the evidence gaps without filling them

List the facts that would alter the choice. Typical gaps might concern whether the eligible audience understands the current name, whether staff can apply an identity consistently, what rights exist in supplied assets, or whether procurement can support the intended arrangement. These are research questions, not findings.

For each gap, specify the least intrusive suitable method, recruitment route or documentary basis, responsible person, decision date and failure condition. A short authorised interview study may answer questions about recent buying decisions. A rights schedule addresses ownership. Neither can stand in for the other.

The ONS explanation of uncertainty and how it is measured notes that error may arise from sampling and from non-sampling sources such as non-response, inaccurate answers and processing. It is general statistical guidance, not a confidence statement for an agency plan. Apply the lesson by showing missing groups, weak records and uncertain inferences beside the decision they affect.

Build real alternatives

Compare complete routes, not supplier names detached from delivery. At minimum, include:

In branding, the routes map onto the type of work. Typical tasks and the routes that usually fit them:

Signs that external branding work deserves a place on the option list include a merger or demerger, a name that customers confuse with a competitor, an identity applied differently in every channel, or a move into a new market. A one-off logo for a stable business is often a freelance task rather than an agency commission.

Some options will be infeasible. Record the reason and evidence instead of deleting them from the history. For every surviving route, set out scope, expected mechanism, assumptions, dependencies, one-off work, recurring work, accountable owner, exit burden and unintended effects. No projected value should appear unless finance has approved its units, period, VAT or tax treatment, source, inclusions, exclusions and uncertainty.

Use the same criteria across routes. A polished proposal should not receive credit for details that the manual baseline was never asked to provide. Likewise, an internal option should include the economic cost of staff time and any capability gap rather than appearing free.

Define what could count as success

An acceptance measure needs a complete specification:

Leave a field blank if no defensible input exists. Do not turn a blank into an industry average. GDS guidance on measuring service success recommends choosing methods according to the question and combining performance data with other evidence. It is aimed at government services, so its measures are not ready-made commercial targets.

Planning an evaluation before delivery also reduces retrospective storytelling. The current Magenta Book distinguishes process, impact and value-for-money evaluation. In an agency decision, a process measure might show whether approved assets were delivered. It cannot establish that the branding caused sales. Any causal claim needs a design that supports incrementality and a qualified evaluation review.

Choose channels only after the evidence question

Owned, earned, paid and direct routes have different control and permission boundaries. Compare them on the same fields: eligible audience, intended action, message version, supplier dependency, terms evidence, rights, data collected, reporting definition, failure route and exit.

Owned media gives the organisation control over its asset, but not automatic attention or accessibility. Earned coverage depends on another party's editorial decision. Paid distribution depends on a contract, the selected service's current first-party terms and its reporting definitions. Direct contact involves facts about the recipient, address, purpose and channel.

Do not invent reach, price, conversion or platform behaviour. If a named service enters the shortlist, retrieve its current first-party product, advertising, terms, privacy, accessibility and billing pages. Record edition, geography, retrieval date, change notice and all unknowns. A supplier dashboard is evidence about what the supplier reports, not independent proof of a commercial effect.

The ICO's direct marketing guidance explains UK GDPR and PECR considerations for marketing. Research participation, access to a public address and a previous service interaction are not interchangeable permissions. A named privacy and PECR reviewer must assess the real purpose, recipient and route.

Do not collapse the specialist gates

A strategy score cannot compensate for a legal, rights, safety or access failure. Run each relevant review on its own evidence.

Intellectual property and trade marks

The IPO's intellectual-property overview describes copyright, designs, patents and trade marks at a general UK level. It does not decide who owns work under a particular contract. Maintain an asset schedule covering creator, pre-existing material, proposed ownership or licence, territory, permitted use, approval and exit delivery.

A company-name check is also not a trade-mark clearance. The IPO provides a trade-mark search service, but a qualified trade-mark professional should decide the scope of searches, similarity analysis, classes, territories and response to uncertainty. Do not promise availability or registration.

Advertising and consumer presentation

ASA and CAP guidance on misleading advertising says objective claims need adequate evidence before an advert appears. Record the exact wording, audience, evidence file, owner and expiry date.

The CMA's current unfair commercial practices guidance addresses conduct prohibited in UK business-to-consumer trading under the Digital Markets, Competition and Consumers Act 2024. Neither source pre-approves a campaign. Advertising and consumer-law reviewers should assess the actual presentation.

Accessibility, equality and security

W3C's WCAG 2.2 is a global technical standard. Which legal, contractual and equality duties apply depends on the organisation and service. Assign an accessibility owner, test affected journeys with appropriate users, record defects and prevent release when a material barrier remains.

The NCSC small organisations guide provides UK cyber-security practices. It is not a certification. A security reviewer should examine the actual accounts, transfers, privileges, supplier evidence, incident route, backups and exit process. Brand work often crosses organisational boundaries, so responsibility should be written rather than implied.

Procurement, contracts and finance

Public bodies and regulated buyers may have procurement requirements general agency guides cannot settle: record the buyer's applicable route, approvals, conflicts and evaluation method.

A contract allocates scope, acceptance, changes, payment and rights. It covers confidentiality, data obligations, liability and exit. The ICO's controller and processor contract guidance explains required UK GDPR terms where that applies, but role labels depend on the facts.

Finance should review currency, VAT treatment, internal labour, supplier charges, contingencies and exit cost. HMRC's page on when to register for VAT states current UK registration rules and exceptions. It does not determine the tax treatment of an unidentified organisation or commission. Obtain qualified tax advice for the actual facts.

Use ninety days as gates, not a promise

A ninety-day planning horizon can create useful review points without predicting delivery.

During days 1 to 30, define the decision, baseline, hypotheses, constraints, options and evidence plan. Gate 1 fails if the audience or decision cannot be bounded, the baseline is unreliable, or required evidence cannot lawfully be obtained.

During days 31 to 60, conduct only authorised discovery and controlled checks. Keep observations, supplier claims, editorial inference and unresolved questions in separate fields. Gate 2 needs comparable option records and an outcome from each applicable specialist review. A hold in one queue prevents progression.

During days 61 to 90, the owner decides whether to continue the current arrangement, gather more evidence, prepare a separately authorised pilot or stop. Any pilot needs a limited exposure, synthetic or properly authorised fixtures, monitoring, incident route, rollback authority and another decision date. The passage of time never converts a missing approval into acceptance.

Sign a conditional decision

The final record should state the chosen route and rejected alternatives, evidence inspected, open assumptions, named owner, acceptance measures, stop events and next review. A defensible verdict may be "continue manually while researching the audience". It may also be "do not proceed".

Before any real use, named specialists must review the actual IP and trade-mark position, advertising and consumer claims, privacy and direct marketing, accessibility and equality, security, procurement and contracts, and tax or finance. Reopen the decision if the audience, source data, supplier terms, intended claims, costs or applicable rules change.

When the decision is to commission: from decision to shortlist

Only routes 3 and 5 above lead to an agency. If one of them survives the gates, build the long list from more than one source: agency directories and their reviews, the member lists of trade bodies such as the Design Business Association (DBA) and the Institute of Practitioners in Advertising (IPA), and effectiveness award entries such as the DBA Design Effectiveness Awards, whose case studies set out a brief and its results. Then carry the decision record into the search:

Before you act

Common questions

What should a usable decision record contain?

It names the organisation or service in scope, the eligible audience and its locating rule, the observable change sought, and the current route including doing nothing. It also records the owner with authority to choose, hold or stop, plus the final evidence date and next review.

Why should alternatives include continuing the current arrangement?

The article says to compare complete routes rather than supplier names detached from delivery. At minimum include continuing the current arrangement and monitoring the stated problem, improving the process manually, commissioning bounded external work, developing internal capability, or combining internal ownership with specified outside expertise.

Can official data show demand for branding work in England?

No. The article states that official data can describe broader business populations and rules, but it does not reveal an England-wide appetite for branding work. The ONS business dataset covers VAT or PAYE registered enterprises and local units, and does not isolate brand owners, buying intentions, budgets, agency use or demand.