Which agency services will survive AI (and which you need to reinvent now)
Two equally useless stories circulate about AI and agencies: the apocalyptic one ("agencies are finished") and the denialist one ("it's a fad, business as usual"). Neither helps you decide what to do with your service catalogue next quarter.
The data backs neither. 76% of Spanish SMEs already use AI daily (2nd Hiscox Report, December 2025): your clients aren't going to pay for what they now do themselves in a chat. But Gartner forecasts a 25% drop in traditional search volume before the end of 2026 (press release, February 2024), and that doesn't destroy the need for agencies: it changes where help is needed. The right question isn't "will my agency survive?" but "which of my services survive as they are, which die, and which need reinventing?". Let's go service by service.
The ones genuinely at risk
Generic content by volume. The "8 posts a month" retainer with no first-party data or judgement was already questionable; now it's indefensible, because the client can generate that same text for free in the chat they already use every day. If what you deliver is indistinguishable from what ChatGPT spits out with a decent prompt, your price tends towards zero. And note the irony: that content doesn't even work for appearing in the AIs, because they cite concrete data and specific answers — exactly what filler doesn't have.
Mechanical reporting. Dumping platform metrics into a monthly report and sending it is work that automation does better, faster and without errors. If your report is descriptive ("traffic dropped 12%") rather than interpretive ("it dropped for this reason, we did this, here's what's coming"), you're charging for a task, not for judgement. Tasks get automated; judgement doesn't.
Pure campaign execution. Ad platforms have spent years absorbing what used to be the craft through automation: bids, targeting, creative combinations. Value remains in campaign strategy and in creative with judgement; less and less remains in pulling levers the platform already pulls by itself.
The ones that survive (and gain value)
Strategy. The cheaper execution gets, the more valuable it is to decide what to execute and what not to. Choosing the positioning, prioritising markets, saying "don't do that" with a rationale: none of that comes out of a prompt, because it demands context about the business, the sector and the history that generic AI doesn't have.
The client's first-party data. ChatGPT knows generalities about any sector; it doesn't know your client's real prices, their lead times, their case studies with figures, the questions they get over the phone. Whoever extracts, structures and publishes that data produces the one kind of content AI can't copy — and, in passing, the kind the AIs cite. The content service doesn't die: it reinvents itself around this, as we set out in the monthly content plan for a GEO client.
The relationship and the trust. When a client's traffic drops 30% and they don't know why, they don't open a chat: they call someone they trust who knows their business. Interpretation with context, being the face in the difficult meetings, and owning the results are the hardest things to replace in everything an agency sells. This isn't sentimentality: it's the material basis of the retainer.
The ones to reinvent now (not abandon)
SEO. The foundation of the craft still holds — useful content, technical health, authority — but the shop window has doubled: on top of Google, it's now also decided in the AIs' answers. That work is called GEO (Generative Engine Optimisation: working on a brand's presence in the answers from ChatGPT, Gemini, Perplexity and the rest), and the honest part is that you already know how to do 60-70% of it. What's genuinely new versus what's the same thing rebranded is separated out in GEO vs SEO. An SEO service that doesn't add this layer will have to explain traffic drops without offering an answer, which is the worst possible commercial position to be in — how to defend the retainer in the meantime, here.
Content. From volume to citable pieces: honest comparisons, FAQs, verifiable data. Fewer pieces, denser, with human editorial review as an explicit (and billable) part of the service.
Reporting. From "what happened" to "what it means and what we're doing". And with new metrics: your clients no longer settle for knowing their Google position; they'll want to know whether the AI mentions them when someone asks about their category. Measuring that by hand across 4 different AIs doesn't scale; that's what we use Surfeo for, which monitors each client's presence in ChatGPT, Gemini, Perplexity and Claude every week and leaves it ready for your report.
The picture of a service catalogue for 2026-2027
Less weight on commodity execution, more weight on strategy, data and new measurement. And at least one new service where demand is growing faster than supply: AI visibility is the best candidate right now — the question "do I show up in ChatGPT?" is already landing in inboxes and almost nobody in Spain answers it with data. The comparison with other new services, with margin numbers, is in new services for your agency in 2026; and the path to turning it into recurring revenue is in how to productise AI visibility.
Frequently asked questions
Should I stop offering content?
No: you should stop offering generic content. Content with the client's first-party data, editorial judgement and an aim of being cited by the AIs is worth more than before, precisely because the generic kind has become free. It's the same service with a different definition of quality and a different sales argument.
How long do I have to reinvent these services?
Frequent ChatGPT use in Spain went from 4% to 28% in two years (Funcas, 3rd AI Survey 2026). At that pace, 2027 budgets will be decided by clients who already live in the AIs. It isn't "you disappear tomorrow", but whoever arrives with the service built when the question goes mainstream will take the clients of whoever is still "looking into it".
Is AI going to replace small agencies?
It'll replace tasks, not relationships. The small agency actually has an advantage: its value was already in closeness and client knowledge, not in armies of execution. Its risk is different: having its book concentrated in the commodity services from the first list. The fix is about service mix, not size.
Where do I start if I can only change one thing this quarter?
With AI visibility measurement for your current clients. It's the piece that turns the threat into a commercial conversation: instead of waiting for the client to ask why traffic is falling, you arrive with data on where they appear (and where they don't) when people ask the AIs. From there, the upsell makes itself.
Want to see where your own agency stands before selling this to anyone? Take the free AI visibility test with your brand or a client's: it's the fastest way to find out whether the problem this article describes already affects you.
