What Gross Margin Should an AI Agency Target?

An AI agency should target 70 percent gross margin or better, not the 50 to 60 percent the old services playbook calls healthy. When the model does the production work, your cost of delivery drops, and if you hold prices, that gap becomes margin. Agencies that pass the AI savings to clients through lower prices are giving away the one advantage that makes this era worth operating in. The benchmark moved up. Price like it.

What gross margin means for an agency

Gross margin is revenue minus the direct cost of delivering the work, divided by revenue. For an agency, direct cost is mostly the labor and tools that go into producing what the client pays for. It does not include sales, overhead, or your own time running the business. That is the line that tells you whether the actual work is profitable.

The traditional agency ran on labor. A designer, a strategist, a media buyer, all billing hours against retainers. Fifty to sixty percent gross margin was considered strong, because people are expensive and people did everything. That number is a relic of a labor-bound delivery model.

Why the old benchmark is too low now

When AI does the first draft, the research, the production pass, and half the reporting, the labor per deliverable falls hard. The same retainer that used to consume 30 hours of human work now takes 10 hours of human oversight on top of the model. If you keep charging the same retainer, your gross margin climbs, and it should.

The mistake is treating the AI savings as a reason to cut prices and win on cost. That race ends with everyone at thin margins and no room to invest. The right move is to hold price on outcomes and pocket the efficiency. This is the core of value-based pricing for an AI agency: you sell the result, not the hours, so cheaper delivery means fatter margin instead of a smaller invoice.

Seventy percent is a reasonable floor for an agency that has genuinely moved delivery onto AI. The best-run ones go higher, because the marginal cost of another deliverable keeps falling while the price holds.

How to actually hit 70 percent

Three levers get you there. First, price on value, not effort. If your invoice is tied to hours, every efficiency gain shrinks your revenue. Get onto packaged, outcome-based pricing so the savings stay with you. Start with productized services vs hourly billing.

Second, cut the human hours per account without cutting quality. That means real delivery automation, not a human doing the same work with a chatbot open. The measure of this is revenue per person: if AI is working, revenue per head should be climbing every quarter. That metric is your margin story told from the labor side.

Third, watch your tooling cost. AI shifts cost from labor to software and model spend. That is a good trade, since a subscription is far cheaper than a salary, but it is still a real line. If your tool stack sprawls into 15 overlapping subscriptions, the savings leak out. Consolidate.

What eats agency margin even with AI

AI does not fix a bad book of business. The biggest margin killers survive the model.

Scope creep is the first. A vague retainer lets clients pull unlimited work for a fixed fee, and no amount of automation outruns unbounded demand. Define the boundary and enforce it. Here is how to stop scope creep in agency retainers.

Bad clients are the second. Some accounts consume triple the delivery cost for the same fee, usually through revisions, meetings, and drama. Find them. Which clients actually make you money is a margin question before it is a relationship question, and the answer often is to fire a few.

Underpricing is the third and quietest. Agencies that lowered prices to win in the AI era locked themselves into 50 percent margins by choice. If your delivery got twice as efficient and your margin did not move, you handed the gain to your clients.

The number is a decision, not a discovery

Seventy percent is not a law of physics. It is a pricing decision. You get there by charging for outcomes, automating real delivery, and refusing to compete on price when the model already made you cheaper to run.

I hold high margin across a portfolio by running delivery and pricing through one system with Agency Script, so efficiency gains land in margin instead of leaking through loose scope. The AI lowered your costs. Whether it raises your margin is up to how you price.