Top Marketing Agencies in the World: A Review

What changed in the holding groups

Any list of the world’s top agencies written before 2026 is describing a market that no longer exists. The holding-company layer went through its largest restructuring in decades, and several of the network names that appear in older rankings have been retired outright.

Omnicom absorbed Interpublic. The acquisition closed in November 2025 and created the largest holding company by revenue at the time. The advertising business was reorganised into a division called Omnicom Advertising, and the surviving global creative networks were cut to three: BBDO, TBWA and McCann.

Three well-known networks were retired. DDB, FCB and MullenLowe no longer exist as standalone brands. DDB was folded into TBWA from 1 December 2025. Where the people and client relationships from FCB and MullenLowe ended up varied by market rather than following one rule — in the UK, for example, Adam & Eve/DDB London and TBWA\London were combined into Adam & Eve\TBWA under a single management team. Trade coverage at the time disagreed on which surviving network inherited FCB globally, which is itself a signal of how market-by-market the process was.

WPP collapsed its media brands into one company and then restructured again. GroupM was renamed WPP Media in May 2025, and Mindshare, Wavemaker and EssenceMediacom continue as client-facing brands inside it rather than as separate businesses with their own profit and loss. In February 2026 WPP simplified the whole group into four operating units: WPP Media, WPP Production, WPP Enterprise Solutions and WPP Creative.

Dentsu consolidated earlier and more aggressively than the others. It unified its creative offering into a single Dentsu Creative network in 2022, absorbing McGarryBowen, Isobar and 360i in the process. In the US it now runs on three service lines: Dentsu Media, Dentsu Creative and Dentsu CXM, the last anchored by Merkle.

The consultancies are a separate category that belongs on the same list. Accenture’s marketing arm started as Accenture Interactive in 2009, was rebranded Accenture Song in 2022 after bringing more than forty acquired agencies under one name, and has since been folded into Accenture’s Reinvention Services alongside strategy, consulting and technology. David Droga stepped down as its chief executive in September 2025 and stayed on as vice chair.

The practical consequence for a buyer: if a shortlist you were given names DDB, FCB or MullenLowe, it was written against the old map. Ask who holds the relationship now, because in several markets the team changed along with the badge.

Why the revenue rankings you find disagree

Search for the largest marketing companies and you get confident numbered lists that contradict each other. That is not carelessness on the part of one publisher — the underlying figures genuinely are not comparable, and it is worth understanding why before you use any ranking to make a decision.

• Reported versus pro forma. After an acquisition, a group can be described by what it actually reported for the year or by what the combined business would have earned had the deal closed earlier. Those two numbers can differ by billions, and lists rarely say which one they used.
• Net revenue versus gross billings. Media agencies handle client money that passes through them. Counting it as revenue produces a far larger number than counting only the fee retained.
• Mixed currencies. The major groups report in dollars, euros and pounds. Lists that convert at different dates, or do not convert at all, produce different orderings.
• Different definitions of a marketing company. Some rankings place a consultancy’s entire reinvention or digital unit in the list. That unit sells technology implementation and systems integration alongside marketing, so it is not like-for-like with a creative network.

The result is that one aggregator will put a consultancy first with a figure several times larger than anyone else’s, another will put a holding group first using a pro forma number, and a third will reproduce figures that look like market share rather than revenue. All three present their list as authoritative.

If the size of your agency matters to you, go to the primary documents: the annual report or regulatory filing of the specific group, which states the basis of the figure. If size does not materially change your decision — and for most B2B buyers it does not — the ranking is entertainment, and the useful comparison is the one in the next sections.

How agencies charge, and what each model does to the work

Commercial models shape behaviour more than any statement of values on an agency’s about page. Four are common, and each has a predictable failure mode.

Monthly retainer. A fixed fee for an agreed scope and a named team. It suits ongoing programmes where the work compounds — search, content, lifecycle. Its failure mode is drift: the scope quietly erodes as the team gets pulled onto other accounts, and nobody notices until a quarterly review. The defence is a written scope with named deliverables rather than a number of hours.

Project fee. A fixed price for a defined outcome: a site, a campaign, a migration. Clean for both sides when the outcome is genuinely definable. Its failure mode is the change request — everything not written down becomes billable, and the relationship turns adversarial halfway through. The defence is spending real time on the specification before signing.

Percentage of media spend. Standard in media buying and the model most often misaligned with the client. It pays the agency more for spending more, which is the opposite of what you are hiring them to optimise. If you use it, cap it or pair it with outcome targets, and read the contract for rebates and principal-based buying arrangements where the agency takes a margin on inventory it resells to you.

Performance or outcome-based. Fees tied to leads, pipeline or revenue. Attractive in principle and difficult in practice: it requires both sides to agree on attribution, and it pushes the agency toward whatever is measurable rather than whatever matters. It works best as a bonus layer on top of a base fee, not as the whole arrangement.

Whichever model you choose, ask what happens to the work product if you leave. Accounts, tags, audiences, feeds, code and documentation should sit in systems you own. An agency that builds inside its own container is selling you a subscription to your own data.

What to ask on the first call

Most selection processes are decided on chemistry and a credentials deck. These questions change the conversation, and the answers are more informative than any case study.

• Who will actually work on this account, and what else are they on? Senior people sell and junior people deliver is the oldest pattern in the industry. Ask for names, roles and the share of their time you are buying.
• Show me a project that went badly and what you changed afterwards. An agency with real history has one. The answer “we have not had that happen” means either very few projects or an unwillingness to be straight with you.
• What will you need from us, and how much of it? Projects stall on the client side more often than the agency side. A partner who has run this before will tell you exactly whose time they need and when.
• How do you report, and what will you report that I will not like? Reporting that only ever contains good news is marketing, not measurement.
• What would make you tell us to stop? A specialist who can describe the conditions under which their own service is the wrong purchase is unusual and worth the shortlist place.
• Who owns everything you build? Ask explicitly about platform accounts, tracking, creative files and documentation.

One more, which sounds blunt and saves months: ask what they would do in the first ninety days and what evidence they would expect to see by the end of it. Vague answers here almost always mean the plan will be assembled after the contract is signed.

Signals that should end the conversation

Guaranteed rankings or guaranteed placement. No agency controls a search engine or a generative assistant. A guarantee of position is either a misunderstanding of how the systems work or a deliberate misrepresentation, and in both cases it tells you what the rest of the relationship will be like.

A proposal that arrives before any questions. A scope quoted without understanding how your sales process works is a template with your name on it.

Metrics without a denominator. “We generated two million impressions” and “we grew traffic 300%” mean nothing without the starting point and the commercial outcome. Ask what happened to pipeline.

Case studies with no numbers or no client. Anonymised cases are sometimes legitimate — contracts restrict disclosure. Cases that are anonymised and free of specifics are a different thing.

Reluctance to name the team. If you cannot find out who does the work before you sign, you will find out afterwards.

Pressure on timing. Discounts that expire this week belong in consumer retail. In a B2B services relationship they signal that filling capacity matters more than fit.

When the answer is not to hire an agency at all

The honest version of this article has to include the cases where the money is better spent elsewhere.

When the proposition is unsettled. Agencies amplify a message; they rarely fix one. If the company describes what it does differently in three meetings, the work will be redone when the positioning settles, and you will have paid twice.

When nobody internally owns the relationship. External teams need decisions, access and review. Without one person whose job includes this, the engagement drifts and both sides blame the other.

When the problem is downstream of marketing. More leads into a sales process that does not follow up produces complaints, not revenue. So does more traffic to a product experience that loses people. Agencies can point at this; they cannot fix it from outside.

When the budget only covers the fee. Media, tooling, production and your own team’s time are separate lines. An engagement funded to the edge of the retainer stalls the first time something needs to be built.

When a contractor or a hire would do. One specialist for a defined gap is often cheaper and faster than an agency relationship. The case for an agency is breadth and continuity — several disciplines, covered when people are away, with someone accountable for the whole.

If several of these apply, the useful first purchase is a short diagnostic engagement rather than a retainer: get an outside read on what is actually wrong, then decide what to buy.

What the specialist tier looks like in B2B

Below the holding groups sits a layer that rarely appears in global rankings and is where most B2B companies actually end up. It is worth knowing how it divides, because the labels overlap and the capabilities do not.

• Demand-generation agencies. Paid media, landing pages and lifecycle, measured on pipeline. Strong when you need volume now and have a product with existing demand. Weaker on positioning and on anything that compounds slowly.
• Search and content specialists. Organic visibility, technical foundations, editorial. Slow to show results and the only channel that keeps working when you stop paying. Increasingly inseparable from visibility inside generative assistants.
• Account-based marketing shops. Built for named-account selling: list construction, orchestration across channels, tight coupling to sales. Expensive per account and pointless if your addressable market is large and undifferentiated.
• Revenue operations and CRM consultancies. They work on the plumbing — data, routing, attribution, forecasting. Unglamorous, and the prerequisite for every other agency’s reporting to mean anything.
• Creative and brand studios. Identity, messaging, production. Usually project-based, often the right first purchase when the proposition is the problem rather than the distribution.
• Product and growth engineering. Onboarding, activation, in-product conversion. Sits between marketing and product, and the discipline most often missing from an otherwise complete roster.

The common mistake is buying one of these and expecting the outcomes of another. A demand-generation agency asked to fix positioning will produce more campaigns for a confused message. A brand studio asked to deliver pipeline will produce a beautiful site and no leads.

The second mistake is buying all of them at once. Sequence matters: measurement before channels, positioning before volume, one channel proven before the second is added.

How to run the selection without wasting a quarter

The default process — a written brief to six agencies, decks from all of them, a decision three months later — is expensive for everyone and selects for presentation skill. A shorter version works better.

Write the problem, not the solution. A brief that specifies deliverables gets you quotes on deliverables. A brief that describes the commercial problem, what you have tried and what constraints exist gets you proposals you can actually compare, and it surfaces the agencies that disagree with your framing — who are often the useful ones.

Shortlist to three. Beyond three, the comparison stops being informative and the cost to the unsuccessful agencies becomes unreasonable, which is noticed and remembered.

Pay for the thinking. A small paid diagnostic from two finalists tells you more than any number of free pitches: how they work, how they communicate, what they notice. It also removes the incentive to over-promise in order to win unpaid work.

Score against what you wrote down. Agree the criteria before the meetings and weight them. Without this, the decision reverts to whoever presented most confidently.

Check the exit before the entry. Notice period, ownership of work product, handover obligations. The time to negotiate these is while they still want to win you.

A realistic timeline for a retainer of meaningful size is four to six weeks from brief to signature, including a paid diagnostic. Longer than that and the problem you were solving has usually changed.

The first ninety days decide the relationship

Most agency relationships that fail were already failing in month one, and the causes are dull rather than dramatic.

• Access takes longer than anyone plans. Platform accounts, analytics, CRM, the site. Start this before the contract is signed, not after the kick-off call, or the first fortnight disappears into permissions.
• The baseline has to be written down. Where things stood on day one, measured the way you will measure them later. Without it, every future report is an argument about the starting point.
• One decision-maker on each side. Agencies stall when feedback arrives from four people with different opinions. Name who consolidates it.
• A standing meeting with an agenda that is not a status update. Status belongs in a document. The meeting is for decisions and blockers.

By the end of the first ninety days you should have more than activity. The minimum useful set: the baseline, a written plan with sequence and reasoning, the measurement working end to end, and at least one thing shipped that you can evaluate. If what you have instead is a series of workshops and a roadmap, the engagement is behind.

What good reporting looks like

Reporting is where you find out whether an agency is measuring or presenting. Four properties separate the two.

• It leads with the commercial number. Pipeline or revenue first, channel metrics underneath as explanation. A report that opens with impressions has chosen its own convenience.
• It states what it cannot measure. Every attribution setup has gaps. Naming them builds more trust than a complete-looking picture that quietly assumes them away.
• It includes the failures. Tests that did not work, pages that did not rank, campaigns that were switched off. Their absence over several months means they are being filtered.
• It ends with decisions. What changes next, what the agency needs from you, and what would make them recommend stopping.

Ask for a sample report from an existing client, anonymised, during the selection process. It is the single most informative artefact an agency can show you, and the request itself tells you something: a partner who reports properly hands it over without hesitation.

The short version

If you take one thing from this: the rankings are unreliable, the holding-group map changed in 2025 and again in 2026, and for a B2B company the size of the agency is rarely the variable that decides the outcome. What decides it is whether the people on your account have done your specific problem before, whether the commercial model points them at your result rather than at their utilisation, and whether you own what they build. Those three are checkable before you sign.

Dmitry Serikov

Founder at Divitio · SEO, GEO & automation