TL;DR
The playbook that worked in 2019 now gets you fired by month four. What actually sells, what to charge, and the delivery model that survives 2026.
→ See how this applies to your business (free 30-min call)The standard advice for starting a digital marketing agency was written for a market that no longer exists. Pick a niche, learn Facebook Ads, charge $2,000 a month, deliver a report. That model is now being undercut by tools a business owner can operate themselves, and by a hundred other agencies offering the same deliverable for $800.
What changed isn't the demand. Local businesses still desperately need help. What changed is which part of the job is scarce. Running ads is no longer scarce. Making the leads those ads produce turn into booked revenue still is.
Here's how to build for the market that exists now.
The Positioning Problem You Have to Solve First
If a prospect can't tell you apart from the last three agencies that pitched them, you'll compete on price, and price competition in this category ends badly.
The differentiators that no longer work: "we're data-driven," "we're a true partner," "we care about ROI," "we're a full-service agency." Every single agency says these. They're not positioning, they're wallpaper.
What actually differentiates in 2026:
If your pitch could be copy-pasted onto a competitor's site without changing a word, you don't have positioning. You have a website.
Pick a Niche You Can Actually Learn
The advice to niche is universal and usually badly executed. The point isn't marketing focus, it's compounding operational knowledge.
When you've served fourteen roofing companies, you know which qualifying question predicts a closed job, what a storm season does to lead volume, what homeowners actually ask, and which offers work in a hail market. Client fifteen takes a fifth the effort of client one and gets better results.
Choose on three criteria:
Average customer value above roughly $1,500. Below that, the math never supports a real retainer.
Enough businesses in your reachable market that you can serve 20+ non-competing clients.
You can get access. A friend, a family member, or one prior client in that vertical is worth more than any market research.
Good candidates: roofing, HVAC, med spa, dental implants, law, solar, remodeling, pest control, restoration. All have real ticket sizes and chronic follow-up problems.
What to Actually Sell
Here's the strategic call that determines your next three years.
Selling campaign management — you run ads, report monthly, get judged on cost per lead. Easy to sell, easy to replace, price pressure forever. This is where most new agencies land and where most of them stall.
Selling revenue infrastructure — you build and operate the system that converts leads into booked jobs: instant response, qualification, follow-up, booking, attribution. Harder to sell, dramatically harder to replace.
The second is what I'd build now, for a simple reason: the first thing a struggling business does under pressure is cancel the ad management. The last thing they cancel is the system that answers their phone.
Practically, most agencies do both — but lead with the infrastructure and treat ad management as an attachment, not the other way around.
The Numbers That Sell It
The most effective sales tool in this business isn't a portfolio. It's a five-number diagnostic run on the prospect's own data:
Then the arithmetic. Instant response plus a proper follow-up sequence typically moves booking rates into the 18–25% range. At 20%, that's 48 booked jobs instead of 22 — 26 additional jobs, or roughly $88,000 in monthly revenue from leads they already paid for.
You didn't pitch. You showed them a leak in their business and quoted the cost of plugging it. Against that number, a $3,000 retainer is a rounding decision.
Pricing
Three components, and skipping any of them causes a predictable problem:
Setup fee: $2,500–$8,000. Real work happens in onboarding. Charging zero trains the client that the build is worthless and starves your first-month cash flow.
Monthly retainer: $1,500–$4,000. For operating the system — monitoring, transcript review, tuning, reporting. Must be genuinely delivered.
Ad spend, always on the client's account. Never comingle. Their credit card, their ad account, their pixel data. This protects you legally and them practically, and agencies that resist this arrangement are telling you something.
Your delivery cost per client is realistically $300–$900 a month once you include platform, usage, and human oversight. Price below $1,500 and you're subsidizing them.
Take equity or pure performance deals only when you have enough accounts to absorb the variance. Early on, it makes your income dependent on the client's sales team, which you don't control.
Getting the First Five Clients
Nobody's first clients come from content or ads. They come from proximity and proof.
Do one for free or near-free, in your chosen vertical, with a written outcome target. Not charity — this is your case study and your process shakedown. Insist on data access as the price.
Get the number. Booking rate before, booking rate after. Written, specific, verifiable.
Go direct to fifty similar businesses with that number in the first sentence of the outreach. Not "we help businesses grow." "We took a roofing company from 22 booked jobs a month to 48 without increasing ad spend — here's how."
Ask every client for two introductions at month three, when the results are visible and they're happiest.
Only then invest in content and inbound. It works, it just works slowly, and it works far better once you have real numbers to write about.
The Delivery Model That Survives
Agencies stall at six to eight clients for one reason: everything is bespoke and the founder is the only operator.
Fix it structurally:
What Not to Do
Where the Bar Is Now
The agencies winning right now aren't running better ads than everyone else. They're operating the layer between the click and the booked job — AI callers reaching every inbound lead within 90 seconds, qualification against criteria the owner defines, follow-up that runs to touch twelve, and pipelines where every dollar of spend traces to a booked job.
That's the stack we run, and it sits behind $102M+ in tracked client revenue at a peak ROAS of 9.2×. Not because the ads are magic — because the response layer stops the leak that every competitor is ignoring.
If you're a business owner comparing agencies, ask each one for cost per booked job on a live account and watch what happens. If you'd rather have the system built and operated for you, [book a free strategy call](/book).
Free Weekly Briefing
One AI Marketing Tactic.
Every Tuesday. Free.
What's actually working across our client accounts right now — ROAS moves, follow-up sequences, creative angles. The stuff that isn't in any blog post yet.
No spam. Unsubscribe anytime. 1,200+ business owners already in.