If you want to know where your law firm, accounting practice, or consulting group is headed, stop looking at your own vertical. Look at marketing agencies.
Agencies started dealing with AI’s effect on their business well before the rest of professional services. They’ve already made the decisions that every expertise-based firm will eventually face. That makes them a useful scouting report.
What happened to agencies
S4 Capital, one of the largest agency holding companies in the world, reported a 12.7% revenue decline in the first half of 2025. Its CEO, Sir Martin Sorrell, attributed the cause directly to AI compressing the value of their core deliverables.
Copy, design concepts, social media content, campaign briefs: agency deliverables were among the first things AI got good at. They’re pattern-based, language-heavy outputs, the kind of work large language models handle at speed and near-zero marginal cost. Typeface’s 2025 survey, reported by eMarketer, found 60% of senior marketing leaders already spending less on agencies because of it.
When a client can produce a first draft of the thing they used to hire you for, the pricing conversation changes. A social media calendar drafted in minutes makes the three-week timeline, and the invoice behind it, hard to defend.
Search Engine Land reported in March 2026 that agencies are now being squeezed from both sides. They automated delivery. Their clients automated the same work. The efficiency advantage that took months to build disappeared in weeks.
That’s the part of the story most people know.
Agencies hit this wall 18 months before law and accounting will. They paid for what they learned in lost revenue.
What the growing agencies did
Not every agency saw decline. Some moved early, before the pressure forced their hand.
They stopped selling deliverables and started selling judgment. The agencies growing right now sell strategic advisory, brand architecture, and market intelligence. That work requires knowing the client’s business and making calls that depend on experience rather than pattern matching. The deliverable became a byproduct of the advisory relationship. On the operational side, AI content governance subscriptions repackage the brand-voice review and output audits agencies used to absorb for free.
Pricing was the next thing to change. Per-deliverable and per-hour models were the most exposed, and agencies moved to retainers or outcome-based pricing. Revenue became recurring. The client conversation shifted from “how many assets do you need” to “what are we trying to achieve.” The same outcome pricing shift is now happening in every professional services vertical.
AI freed up hours, and the growing agencies spent them moving upmarket. They took on work they couldn’t have done before: deeper analysis, real-time market monitoring, strategic recommendations backed by data the client didn’t have.
Why the agency numbers look milder than they are
Our own census looks like good news for agencies until you read which lines landed where. When we ran the same exposure test across law, accounting, and agencies, agency service lines came out the least replaced of the three: 7 of 16, against 13 of 21 legal practices and 12 of 17 accounting service lines. On paper the vertical that got hit first looks like the one holding up best.
The seven a client can now handle alone are the deliverables. Creative and graphic design, video, social media management, content writing, email, influencer work, audience research: the things a brand used to brief out and now produces itself. Agencies look less exposed because the exposure already happened, and what the census counts is the business left standing after the deliverable went.
Of the nine agency lines that didn’t come out replaced, eight are only repriced: a client still needs someone to run the tool and check what it produced, at fewer billed hours than before. Only one is genuinely protected, brand strategy and identity, and it survived on the judgment involved rather than on anything shielding it from competition. Law and accounting each protected four lines outright, for reasons the cross-vertical read goes into. Nothing ever required a client to buy a deliverable from an agency: no license, no filing, no signature a regulator insists on. That’s why agencies reached this stage first, and why the rest of professional services is walking toward it with 18 months of cover the agencies never had.
The same pattern, different timelines
The sequence that played out in agencies is already visible in other verticals.
In legal, more than 500 of Harvey AI’s 1,000-plus customers are in-house legal teams. Garfield.Law delivers legal services entirely through AI, charging per document instead of per hour. The pattern is the same: AI gets good at a deliverable, clients bring it in-house, new entrants show up with different economics.
In accounting, the AICPA’s Dynamic Audit Solution ingests data in real time, and the annual audit cycle is becoming continuous. The Big Four are putting their AI investment to work serving mid-market clients profitably, clients that used to be yours.
Staffing firms face the same shift: Bloomberg reported that clients are automating candidate sourcing and screening, the service the industry is built on.
The dynamics are the same across verticals. The timeline is the only difference. And that timeline is useful.
What transfers
If your revenue depends on producing a thing (a contract, an audit, a report) and AI can produce a comparable version faster and cheaper, the pricing conversation will change. Agencies learned this when clients started generating their own content. By then, the survivors had already moved their value to what surrounds the deliverable: the strategy, the interpretation, the relationship.
Efficiency alone isn’t a strategy. Agencies that adopted AI to produce deliverables faster saw margins improve briefly. Then competitors did the same. Then clients noticed and asked why the bill hadn’t changed. Efficiency within an existing model has a ceiling. What matters is what you do with the time and margin it creates.
Moving early left agencies with choices. Waiting left them taking whatever price the client decided was fair. That’s probably true in every vertical.
Clarity, not urgency
You have something agencies did not: you can watch someone else go through it first. The same dynamics are playing out more slowly in law, accounting, consulting, and staffing.
In agencies, the growth came from changing what they sell, how they price it, and how they use AI. Running the old work faster improved margin for a while, and then clients took it back. The commercial decisions mattered more than the technology decisions.
Sorrell named the cause only after S4 had already lost the revenue. You can learn it from the agencies’ numbers now, or from your own in 18 months.