Nobody else showed up to the Dynamic Agency Q&A one Thursday in August, so the one owner who did got an hour with me to herself. She runs an agency that works exclusively with accounting firms, and about twenty-five minutes in, she put into one sentence what I’d been hearing from owners in pieces all year.
“It used to be like, oh, we understand accounting. So you don’t have to explain to your marketer accounting. Now, with AI, I feel like it’s hard to make that case still.” And a few seconds later: “Our old shtick doesn’t work as well.”
She’s right. She’s also in better shape than most owners dealing with the same thing, because she noticed it and said so, and most people find out when the pitch stops landing and they can’t work out why.
AI made the agency differentiators worthless that an agency can only claim, like industry knowledge and shared relationships, while the ones built into how the agency runs still win deals because a buyer can check them before signing.
Quick Take
- Industry knowledge still helps on the call. It stopped being a reason to hire you, because every competitor in your niche can now sound fluent in your buyer’s business with a little prep and a chatbot.
- The popular advice to specialize is half right. A niche you only put in the headline is as easy to copy as any other claim, and the version that holds up changes how the work gets done.
- Run the portability test on your market focus: if you could hand your exact system to a client in a different industry tomorrow, your niche is a label.
- The differentiators that held up are structural, like a named problem, a point of view somebody disagrees with, a delivery model shaped around the buyer’s week, and pricing built on the buyer’s own math.
In this article
- Did AI make industry expertise worthless?
- Which differentiators lost their value
- Are specialist agencies safer from AI?
- The portability test
- What still works
- How to check your own
Did AI make industry expertise worthless for agencies?
Mostly yes when it comes to winning the deal, though it still helps keep a client comfortable once you have them.
Knowing a client’s industry used to spare the buyer a painful stretch of explaining their own business to a new vendor, and that was worth paying for. Most of that stretch is gone now. When I asked the accounting-firm owner how prospects react to her background today, she said a lot of founders tell her they like that she’s from their world, and she’s had calls where a founder drops an accounting reference into the conversation. She’s also been asked straight out whether it matters. Her answer was “not today anymore,” and she described it as “great for rapport building. But after that. Like doing the hard work, like, no, it doesn’t matter.”
The industry data points the same way. Forrester’s State of AI Inside US Marketing Agencies, 2026 found that nine in 10 agencies use generative AI, and 70% of them use it for research and competitive intelligence. That’s the homework industry experience used to let you skip, and every competitor in your niche has the same shortcut on their laptop.
Which agency differentiators lost their value to AI?
The ones a buyer can hear but can’t check. We went through her positioning lever by lever on that call, and the pattern showed up fast.
Industry knowledge is the one covered above. It still opens the conversation.
Relationship capital went next. She’s been on the popular podcasts in the accounting space, she goes to the conferences, and the software vendors are happy to send her referral links. Her read on it: “there’s like 20 of us,” and “all my competitors have that too.” In a small industry, everyone serious ends up in the same rooms, so the relationships prove you belong there without separating you from anybody else who belongs there. The question I ask about this lever is whether you can activate the relationship for the buyer’s benefit at any given time, and most agencies can’t.
The standard process was the third. When we got to how her process differs from other agencies, the honest answer was that it mostly doesn’t: foundation, then strategy, then assets, then marketing. (That’s my process too, even with agencies, so I’m not throwing stones.) A process everyone runs was never much of a differentiator, and AI makes it cheaper for every shop to describe that process well.
Capability claims round it out. “AI-powered” means nothing to a buyer who assumes everyone has it. I made that argument in Parity Lag: a tool advantage lasts only as long as it takes competitors to buy the same tool, and that’s now a matter of months.
Are specialist agencies safer from AI than generalists?
Probably safer than a generalist selling commodity production, and a lot less safe than this year’s advice suggests.
The consensus isn’t hard to find. Violetta Bonenkamp’s June 2026 piece on why marketing agencies are getting crushed by AI puts it in one line: “Niche knowledge still earns trust. Generic production does not.” ALM Corp’s 2026 agency growth guide says flatly that “specialized agencies outperform generalists.”
I agree with the direction, and I’d push back on what owners do with it. Most people hear “specialize” and rewrite the homepage headline. That produces a specialist label, and a label is a claim, and claims are exactly what AI devalued. The owner on my call had already done what the advice says. She picked one industry and stuck with it, and she was still sitting there telling me her edge had gone soft.
A niche still does useful work. It makes every message cheaper to aim, and picking one that fits you is still the right first move for most small agencies. It just can’t carry the differentiation by itself anymore.
What is the portability test for a niche agency?
The portability test asks whether your system would work unchanged for a client in a completely different industry. I put it to her this way on the call: “If you could take the same system that you’re using right now for accounting firms and just use it on anybody else, like the same exact system, then it doesn’t really matter that you’re working with accounting firms.”
If the answer is yes, your niche is a label. If the answer is no, you’ve found the part of your market focus a chatbot can’t hand to a competitor, because it lives in how the work runs.
Her agency passed on a few points once we went looking for them. Most web designers require a client to show up with brand photos and copy, and her team doesn’t, because accounting firms don’t have those sitting around. Her website process runs two weeks, with three meetings and the assignments delivered a week ahead. And a point of view came up that ties both of those together: most accounting firms don’t need a digital masterpiece.
Watch how those feed each other. Because she knows the industry, she knows what these buyers don’t care about. Because they don’t care about a month of design exploration, she can finish in two weeks. And because she finishes in two weeks, the point of view has proof behind it. A competitor who wants to copy that has to believe the same thing about the buyer and rebuild their delivery around it. That stacking is what I mean when I say one lever is never enough.
Which agency differentiators still work now that AI does the work?
The ones that cost you something to hold. Four came up on that call, and they survive for the same reason: matching any of them forces a competitor to change how their business runs.
A named problem. Give a problem a name, and buyers tend to assume you’re the one with the solution (some people call this the Rumpelstiltskin effect). I talk about referral reliance a lot, enough that other coaches started using the phrase, and I’ve been on calls with $3 million agencies where the owner told me that phrase is why they booked. One of my clients runs a design agency and calls his problem design drift, the way every new page on a site looks a little less like the original design once marketers get their hands on it. A name like that works because you’ve been the one saying it, and that history is hard for a competitor to fake. More on this in how to explain what makes you different.
A point of view somebody disagrees with. This is the lever where AI helps you. Ask it for the conventional wisdom around your buyer’s problem, then read the list and circle what you disagree with strongly enough to bet the business on. Be careful, though, because conventional wisdom is conventional because it usually works, so you need a case you can defend with data. If your buyers aren’t comparing agencies yet, aim the point of view at their assumptions (“most accounting firms believe…”). If they’re already weighing three of you, aim it at the competitors.
A delivery model shaped around the buyer’s week. Her two-week build is one. Mine is that I used to meet clients weekly for an hour, and I switched to every other week without adding clients, so the open time went into doing marketing work between sessions. We make more progress meeting less, and I say so on sales calls. It’s a choice with a cost, and buyers can see the cost.
Pricing built on the buyer’s own math. Her clients often run businesses that look a lot like hers, so a firm charging its own clients $600 a month balks when she quotes $1,200. The answer is arithmetic the buyer already understands: at $600 a client, she needs to bring them two new clients to pay for herself, and every client after that is money in their pocket. That’s the kind of thing I mean by anchoring the price to the problem.
And if you built your agency on knowing an industry, you didn’t make a mistake. That knowledge is exactly what lets you spot which structural choices your buyers would pay for. It just has to show up somewhere a buyer can see it now, and the headline stopped counting.
How do I check whether my agency’s differentiator still holds up?
Write down every reason you give prospects to hire you, and next to each one, note whether a buyer could verify it before signing.
Then open a chatbot and ask it what a marketing agency should know about your buyer’s industry. Anything on your list that shows up in that answer is something any competitor can say now, so move it off the sales call and into the rapport-building part of the conversation where it still helps.
Run the portability test on whatever’s left. The items that fail it are labels. The items that pass are where your positioning lives, and they’re what your owned word should grow out of. If nothing passes, you’ve found next quarter’s work, and it usually starts with deciding what you’re willing to give up to be the obvious choice for someone.
If you’re an agency owner whose old pitch stopped landing this year, you noticed sooner than most, and working out what replaces it is what we do inside the Dynamic Agency Community.
FAQ
Is it still worth niching down my agency now that AI exists?
Yes, for aiming your marketing. A niche makes every message cheaper to point at the right buyer. Just don’t expect the niche alone to set you apart once every competitor in it can sound as fluent as you, so pair it with at least one structural choice that only makes sense for that market.
Do clients still care that my agency knows their industry?
They like it, and it builds rapport quickly. It’s rarely why they sign anymore. One agency owner who works exclusively with accounting firms told me she’s been asked directly whether her accounting background matters, and her honest answer was “not today anymore.”
Can speed still differentiate an agency?
It can if you prove it with a number. Never call yourself the fastest, because the first prospect who asks who’s second will expose that you don’t know. Measure time to first result instead: finish a client’s site, start a timer, record how long until their first lead, then average it across clients and publish the figure.
Should I stop telling prospects my agency uses AI?
Tell them if they ask, and keep it out of the pitch. Buyers assume you use it, and leading with it tends to make them wonder why they’re paying full price. That’s the trap I wrote about in how an AI pitch accidentally devalues your work.