Leading advertising agencies: what the word leading is based on, and how to verify it yourself

The agencies described as leading are usually leading on one of four measures, and the measure is rarely stated on the list you are reading. Those measures are billings or revenue, creative awards, client roster size, and paid placement on a directory. None of them predicts whether an agency will do good work on your account, because the agencies at the top of revenue and awards tables are competing for budgets far larger than most advertisers hold. The agency that is leading for you is the one with proven results in your category, at your budget, with the specific capability your plan needs, and with the senior people who pitched still on the account afterwards. This page shows how the public rankings are built, so you can read them properly, and gives a scoring sheet for deciding it yourself.

This page covers how to evaluate and shortlist agencies. For running the relationship once it is signed, see a page on briefing an agency and judging the work.

How the published rankings are actually compiled

Four different kinds of list circulate, and they measure four different things.

List type	What it ranks on	What it is good for
Trade press revenue and billings tables	Reported or estimated money handled, often at holding company level	Seeing who has scale and global reach
Awards league tables	Juried creative and effectiveness awards won over a period	Finding craft, and effectiveness categories are the ones to read
Review and directory platforms	Client reviews, verified project data, and in many cases a paid tier that affects position	Building a long list and reading written client feedback
Editorial roundups and blog lists	The author's judgement, sometimes affiliate arrangements	Discovery only, once you know who wrote it and why

Read any list with three questions. Who compiled it, what did they measure, and could an agency pay to appear on it? A directory that sells sponsored placement is not dishonest, and it is not a ranking either. Where a list does not publish its method, treat it as a directory.

Four kinds of agency wear the same name

Most disappointing agency relationships are capability mismatches, agreed in a room where nobody said which job was being bought.

• Creative agencies make the idea and the assets: positioning, campaign concept, film, copy, design. Judge them on the work and on whether the idea survives contact with a media plan.
• Media agencies plan and buy the space, negotiate rates, and own measurement and attribution. Judge them on planning rationale, buying terms and transparency of fees.
• Performance agencies run paid search, paid social and shopping to a cost per acquisition target. Judge them on account structure, testing discipline and how they report incrementality.
• Full service and integrated agencies do several of these under one contract. Judge them on the weakest of the disciplines you need, because that is the one you will feel.

Write down which of the four you are buying before you contact anyone. If the honest answer is two of them, say so in the brief and ask each agency which part they would subcontract.

The scoring sheet

Score every agency on the long list out of five on each line. Weight the lines that matter to your plan, and shortlist the top three.

• Category evidence. Work in your category or an adjacent one with the same buying behaviour, with the result stated and the client named.
• Budget fit. Their typical account size sits near yours. Being the smallest client in a large agency and the largest in a small one both create problems.
• Who does the work. The names on the pitch, their actual hours on your account per month, and who you call when something breaks.
• Commercial clarity. Retainer, project fee, percentage of media spend or performance based, written plainly, with what is in and out of scope.
• Data and account ownership. You own the ad accounts, the analytics property, the tag manager, the creative files and the audience lists. Confirm in writing before signing.
• Measurement they propose themselves. An agency that volunteers how it should be judged, before you ask, is the signal most worth noticing.
• Reference quality. Two current clients and one former client. The former client conversation is the most useful twenty minutes of the whole process.
• Reporting cadence. What arrives weekly, what arrives monthly, and whether a human talks you through it.

Questions that separate the shortlist

Ask all three the same questions and compare answers side by side.

• Which of our current activities would you stop first, and why?
• What would you need from us to do this well, and what usually goes wrong on the client side?
• Show us a plan that did not work and what you changed.
• How is your fee affected if our media spend goes down?
• What will you report in month one, month three and month six?
• Who exactly will be in the weekly call in month four?

The fourth question is the sharpest one in a media conversation. An agency paid a percentage of spend has a built in reason to recommend more spend, which is manageable once it is visible, and a flat fee or a performance arrangement changes that balance. Ask it directly and get the answer in the contract.

Warning signs worth walking away from

Be careful with an agency that promises a specific ranking, a specific cost per lead or a specific revenue figure before seeing your data. Google states plainly that nobody can guarantee a number one ranking, and the same logic applies to a promised acquisition cost in an auction nobody controls. Other signals to weigh: case studies with no client name and no number, a pitch team you never meet again, refusal to let you own the ad accounts, long notice periods paired with vague deliverables, and an unwillingness to share raw platform access.

Run it as a paid pilot

The strongest version of this process ends in a short paid project rather than a twelve month retainer. Give the chosen agency a defined piece of work with a deadline, a budget and a success measure, and give them the access they need to do it properly. You learn more about an agency from six weeks of real work than from any pitch, and both sides enter the retainer knowing what they are signing.

Are the big network agencies better than independents?

They are different. Networks bring scale, international coordination, research resources and bench depth. Independents bring senior attention and faster decisions. For a single market budget, senior attention usually produces more, and for a multi market launch the coordination is worth paying for.

How much should an advertising agency cost?

Fees are structured as a monthly retainer, a project fee, a percentage of media spend or a performance element, and the right level depends on the hours your scope genuinely needs. Ask each shortlisted agency to show the fee as a team plan: who works on the account, how many hours each, and at what rate. That converts a number you cannot judge into one you can.

Should I hire separate creative and media agencies?

Splitting gives you the best specialist in each and costs you coordination time, so someone on your side has to own the brief across both. One integrated agency is simpler to manage, and the quality is set by its weakest discipline. Decide based on how much client side capacity you actually have.

How long before I can judge an agency?

Paid media gives readable signal within four to six weeks. Brand and organic work needs a quarter or more. Agree the judging milestones at signature so nobody is renegotiating the goalposts later.

What belongs in the brief?

The commercial objective, the audience, the budget, the non negotiables, the measure of success, who approves, and the deadline. One page is enough. Agencies produce better work from a short sharp brief than from a long vague one.

A question your shortlist has probably not been asked

You can ask every agency on your scoring sheet for case studies. Ask something different too: when a marketer asks an assistant for a leading agency in your category, who is named?

The word leading in a ranking is compiled from somebody's criteria. An assistant's answer is compiled from what the web says, which is another, independent view of reputation. Compare it with your sheet and ask any agency that appears on one list but not the other to explain why. AI Knows Us can record that answer for the question you choose.