In February I got on a call with the founder of a small consultancy who had tried almost everything. He’d mailed promotional packets, sponsored conferences, run AI readiness surveys as a lead magnet, paid a service to pitch him onto podcasts, and posted on LinkedIn with mixed results. Most of his business still came from referrals, cold emails and meetups. When I asked what was in the way, he named two things: inbound leads, and his ability to pick something and stay focused on it.
I liked him right away, because the effort was all there and so were the ideas. The trouble was the order. He was running go-to-market experiments for an agency that hadn’t decided who it was for yet, so every channel he tried was carrying a message too broad to land anywhere, and each failure looked like proof that the channel didn’t work.
That’s the most common way I see agencies stall, and it’s why I don’t answer “how do I grow my agency” with a list of tactics.
To grow a marketing agency, make four decisions in order: who you serve, how you want them to see you, what you sell them, and where you’ll reach them. Most stuck agencies made the fourth decision first.
What’s the right order to grow a marketing agency?
The order is niche, then positioning, then offer, then go-to-market, because each decision is the input to the next one. You can’t say how you’re different until you know who you’re different to, you can’t design an offer until you know which problem your position claims, and you can’t pick channels until you know who you’re trying to find and what you’re putting in front of them.
Most guides to growing or scaling a marketing agency list the same pieces and present them side by side: pick a niche, write SOPs, build a sales engine, hire managers, buy a project management tool. Google’s AI Overview for “how to scale a marketing agency” did exactly this when I checked it in October 2026, pulling from software vendors and course sellers. None of that advice is wrong, and none of it tells you what to do first. That’s the question an owner with forty hours a week is asking.
This order is the same one I teach inside Dynamic Agency OS, and it’s built around where agencies leak. When I scored 53 agency websites against the 11 Differentiation Levers, the gap between agencies that had made the first decision and agencies that hadn’t was bigger than any other split in the data. So the sequence below starts there.
I’ve broken the order myself. I wanted to do leadership coaching, so I started selling it, and only then found out the audience I already had didn’t want leadership coaching. I’d built the offer before I’d decided who it was for, and the market answered the question I’d skipped.
Two more pieces sit underneath the four: delivery that doesn’t swallow the time growth needs, and a short list of numbers that tells you whether any of it is working. I cover both after the four decisions, since they only matter once there’s growth to protect.
Should a marketing agency niche down before anything else?
Yes, and the data on this is lopsided: in my sample, niched agencies cleared the bar on 1.22 differentiation levers each, and generalists cleared it on 0.13, a ninefold gap. That comes from the public-footprint half of the 53-agency scoring set I describe in the agency positioning guide: 46 agencies scored from their websites alone, split evenly between niched and generalist, with no pre-screening for how good their positioning was.
A niche makes the other ten levers answerable. A generalist can’t own a specific problem, can’t write a point of view anyone disagrees with, and can’t build proof that a buyer recognizes as theirs, because every one of those depends on knowing who the buyer is. Outside research points the same way: Promethean Research found that agencies which narrowed their services grew 13% on average.
I define a niche in three parts, and owners usually stop after the first one: a vertical (the industry), a horizontal (the job function inside it, say marketing directors rather than CEOs), and a specific problem that person has. An owner I spoke with on the same day as the founder above was selling into nine sub-industries at once, from manufacturing and CPG to construction, logistics and oil and gas, with a team of three. He described the result in one line: once deals close, it’s back to square one. Nine verticals means nine sets of buyer language and nine reasons a prospect can decide you’re really for someone else.
The fear is always the revenue, and it’s fair. You can keep the clients you have. Niching means work outside the niche becomes a choice you make deliberately each time, and the default yes goes away. I have owners score leads against the niche: a perfect-fit lead is the niche, but a strong lead from another industry still beats a junk lead you’d only take for the money. Roll it out over a year and the dip mostly never comes. I wrote up the decision itself, including how to do it while client work still pays the bills, in should your agency niche down, and how to pick one in finding your agency niche.
How do you position a marketing agency once you’ve picked a niche?
Positioning is deciding how you want that niche to see you and giving them a reason to believe it, and it has to come before the offer and the marketing because every headline, ad and price you write afterward expresses it, whether you chose it on purpose or not. Tactics don’t save bad positioning, and I’ve watched a lot of owners spend a year proving that.
The usual first draft of a position is some version of “we do better work.” It’s never a usable claim, because nobody can verify it before the sale and every competitor says it too. Agency positioning holds up when it’s built from things a buyer can check. The 11 Differentiation Levers exist to surface those: market focus, point of view, problem ownership, methodology, talent, outcomes, delivery model, economic model, operational constraints, risk reversal and relationship capital.
One number from the scoring set changed how I coach this. The levers an agency can express in copy alone (point of view, problem ownership, market focus) averaged 4.32 out of 10. The levers that require an actual structural commitment (operational constraints, risk reversal, economic model) averaged 2.52. Same agencies, scored the same way. The levers you can write score 1.7 times the levers you have to do, and 37 of the 53 agencies had no lever at all that cleared the bar of 7.
So when an owner tells me they repositioned last quarter and nothing changed, my first question is what they changed besides the words. If the answer is the homepage and the LinkedIn banner, the position hasn’t moved yet. The full version, including how to score your own agency, is in the agency positioning guide.
How do you make your agency hard for a competitor to copy?
Pick at least one lever that costs you something, because a claim that costs nothing can be pasted onto a competitor’s site by Friday. The test I use is simple: could the agency across town publish your sentence tomorrow and have it be true? If yes, it’s a description of the category, and buyers will choose inside the category on price.
The differentiators that survive this test usually sound small and turn out to change everything. Month-to-month contracts can do it in a category that demands annual terms. So can a hard cap, like never taking more than ten clients at a time, because a buyer reads that cap as a promise your senior people won’t be stretched thin the month they sign. So can a guarantee that costs you money if you’re wrong. Relationship Capital, the lever built on who you know and who vouches for you, is the most open ground in the set: not one of the 53 agencies cleared the bar on it.
Watch for claims that used to differentiate and now read as table stakes. Every lead gen agency says you don’t pay until you get a lead, and a promise everyone makes can’t set anyone apart. I took this apart in more detail in how to differentiate your agency, and the lever-by-lever version lives on the 11 Differentiation Levers page.
What should a marketing agency sell, and how should it price it?
Sell the fix for the problem your positioning claims, and price it from what that problem costs the buyer, because positioning sets pricing far more than your hourly cost does. An offer designed before the niche and the position has to be general enough to fit anyone, and a general offer is the easiest thing in the world to compare on price.
I’ll use myself as the example. My prices are low on purpose: my positioning is about being accessible to smaller agencies, and charging premium rates would tell a sub-$1M owner I’m not for them. The price is an expression of the position. When an owner says their market won’t pay more, I tell them it might be true and to test it before believing it. Every market has a ceiling, but there’s usually a higher tier inside it: some small businesses pay $20,000 for the same website others won’t pay $5,000 for.
On packaging, I see three ways to run agency services. You can productize (an assembly line where everyone gets the same thing), standardize (process and automate the parts that repeat, stay custom on the variables that make each client different), or go fully bespoke. Few agency types benefit from full productization, and most belong in the standardized middle, which I covered in standardized, not productized and in should I build a productized service or stay custom.
For the pricing model itself (project, retainer, hybrid, performance, value-based), match it to how the client’s problem behaves. I made that case in agency pricing models, and which one fits your offer. If you want to check whether the offer you already have is pulling its weight, start with what makes an agency offer worth buying. And if you’re weighing a new service line because the market is moving, the same order applies: whether to sell AI search work walks through that call for one service a lot of owners are considering this year.
How do marketing agencies get clients once the foundation is set?
Go where your niche already spends its attention, run demand and brand marketing at the same time, and get one channel working before you add a second. No single channel works for every agency, and anyone who tells you there’s one perfect channel is selling it.
Most agencies are leaning on one channel already, they just didn’t choose it. In RSW/US’s 2025 survey, 75% of marketing services firms named referrals their best lead source and 93% said their growth engine wasn’t strong enough to hit their goals. Those two numbers describe the same agency. Referrals are great, and every one is evidence you do good work, but you can’t turn them up when you need them.
The order matters here more than anywhere. Jumping straight to paid ads with no positioning behind them means paying to bring someone to a website that tells them “this isn’t for you,” or worse, losing a buyer who was a fit because you gave them no reason to stay. With the first three decisions made, a channel test finally means something, because the message is aimed and the offer is clear.
A few rules I hold clients to. Every channel does up to three jobs: discovery (where people learn a solution exists), relationship (where they come to trust you) and conversion (where they can buy). LinkedIn is rarely a discovery channel. That’s why a LinkedIn-only plan can build relationships with people who never find you. Smaller agencies should run roughly 80% demand work that produces conversations this quarter and 20% brand work that compounds, shifting the mix as content starts lowering the cost of a lead. And once one channel works, add a second, because an agency running on one channel is one ad-account ban away from zero.
The Brand-Demand Spectrum is how I decide which channels fit a given stage. For the specific plays, read how to generate leads for a digital marketing agency, agency strategic partnerships for the partner route, and how to build a no-fluff go-to-market strategy for putting it into a plan.
How do you keep delivery from eating the growth?
Standardize the repeatable parts of delivery and protect a fixed block of time for marketing, because client work always wins the week if you leave it to compete. Client work is immediate: you finish it, the client’s happy, the box is checked. Marketing waits on a market to respond and gives you far less control, and in a head-to-head with a client deadline it loses every time.
I call the loop that follows Delivery Drag: delivery keeps consuming the time you meant to spend on marketing, and the pipeline thins for months before you feel it. Most owners caught in it believe they’re one good month away from getting ahead of the work. What is Delivery Drag covers the mechanism and how to spot it.
The fixes are operational and mostly unglamorous. Standardize the parts of the work that repeat so your senior people spend their hours on the variables. Separate the person who manages the client relationship from the person who manages the work. I wrote about that split in project management vs account management. And bake content into the operations you already run: record client and sales calls, because the questions people ask on them are next month’s posts, and every sales question is an objection you can answer before the next call.
One delegation move I like for owners who are stretched: treat your own agency like a client. Assign it an account manager, put a junior on it, and become the client yourself, still setting the budget and approving what goes out. It’s low risk, it trains the team, and marketing stops depending on your free evenings.
Which numbers should you track while you grow?
Track a small set: the share of new revenue that came from referrals, your demand-to-brand mix, how many of the right people saw your positioning against how many calls it booked, and revenue per person. A dashboard with twenty numbers gets checked for a month and then ignored, and the owners I work with only need enough data to know which lever to pull next.
Judge marketing by exposure: how many of the right people have seen your position, and how often. Set tests in those terms, a specific number of people in front of a new message measured against calls booked, and you’ll know far sooner than you would by waiting to see how the quarter feels. If you know your site traffic and your conversion rate, you already have two levers.
Sequence applies here too. I believe in profit-first businesses, and I still tell owners under $1M to drive revenue first and treat roughly $1M as the first milestone, because a 10-point margin gain on a small base moves very little money. I made the full case in revenue earns you the right to optimize profit. For the specific metrics, see a simple KPI dashboard for $1M marketing shops, revenue per FTE, and the industry baselines in marketing agency benchmarks for 2026.
When should you bring in outside help?
Bring in outside help when you can’t make one of the four decisions yourself, and hire a marketer only after something is already working and the job has become volume. Those are two different kinds of help, and owners mix them up all the time.
A marketing hire made before the positioning is settled can’t get results, because they’re building on a foundation the founder hasn’t built, and the miss usually costs a salary and a year. Positioning is the founder’s job to lead. Once you’ve seen a channel work and need someone to run it at volume, a hire makes sense, because now you’re handing them a working system to improve.
Help with the decisions themselves is a different question: a positioning consultant, a coach, a peer group, a course. If you’re deciding between a positioning firm and a branding firm, start with agency positioning vs branding, because the two fix different problems. The coaching side is covered in the guide to agency growth coaching. Before you pick any model, run it through the four tests in which agency growth frameworks hold up, mine included.
The Monday version
Take a sheet of paper and write the four decisions down the left side: who you serve, how you want them to see you, what you sell, where you reach them. Next to each one, write the answer you’d give a stranger today, in one sentence. The first line you can’t fill in, or can only fill in with something a competitor could also say, is where your next 90 days go. Everything below it can wait, including the channel you were about to try.
If you’ve already got the first two lines filled in, hell yeah. You’re further along than most of the agencies I scored. And if you’d like to work through the rest next to other owners doing the same thing, that’s what the Dynamic Agency Community is for.
FAQ
How long does it take to grow a marketing agency?
It depends on exposure more than on the calendar. A position starts working once enough of the right people have seen it, so a new offer in front of a few hundred buyers tells you very little, and the same offer in front of a few thousand tells you a lot. Set the test in those terms and you’ll know which way it’s going long before the calendar would tell you.
Is scaling an agency different from growing one?
Most people use the words interchangeably. When owners say “scale” they usually mean growing revenue without the same growth in hours or headcount. That’s mostly a delivery question, and it only matters once there’s demand to deliver. The four decisions come first either way.
Can I skip niching if referrals are working?
You can, for as long as they keep working. The trouble is you can’t predict when referrals slow down, and an agency that never chose a niche has nothing aimed and ready when they do. Niching while referrals are strong is the cheapest time to do it, because you aren’t desperate for the next deal.
Should I master one marketing channel before adding another?
Yes, get one working first, and then add a second as soon as it does. Running five half-tested channels at once means none of them gets enough attention to tell you anything, and running on one forever means a single platform change can take your pipeline to zero.
Do I need to productize my services to grow?
No. Few agency types benefit from full productization. Most do better by standardizing: running the repeatable parts on a process, staying custom where clients really differ, and changing the process whenever you find a better way.