Two agencies I know do nearly identical work. One runs at roughly 55 percent gross margin, the other closer to 25. The difference is not talent, city, or client quality. One prices in monthly retainers with defined scope. The other quotes fixed projects and eats every change request.
Pricing model is not a billing detail. It is the decision that determines who absorbs uncertainty, and uncertainty is the actual product in professional services.
The real question: who carries the risk
Every engagement contains three unknowns. How long the work will take. Whether the client will supply what is needed on time. Whether the outcome will be good.
Hourly billing puts all three on the client. Fixed project pricing puts the first two on you. Performance pricing puts all three on you. Retainers split them.
That is the whole framework. Everything else is implementation.
Hourly, and why I mostly do not
Hourly is honest and it is a trap.
It caps your income at hours available, which means growth requires headcount, which compresses margin. It punishes you for getting faster: the better your process, the less you earn for the same outcome. And it makes every client conversation about your cost rather than their result.
The narrow case where it works: genuinely open ended advisory or diagnostic work where nobody can scope the problem yet. I will do a paid discovery on an hourly or day rate basis, deliberately, and then move to something else.
If you are billing hourly for delivery work you have done a hundred times, you are subsidizing your own expertise.
Fixed project pricing
Clean for the client, risky for you, excellent when the work is genuinely repeatable.
It works when you have done the thing enough times to know the hours within about 20 percent, when the deliverable is definable, and when the dependency on client input is limited or contractually gated.
It fails on scope. Always on scope. The project you quoted for 12,000 becomes a 19,000 project through eleven small reasonable requests, none of which felt worth fighting about.
The controls that make it survivable:
Explicit revision limits, stated in the proposal, with an hourly rate for additional rounds. Two rounds is standard and enforceable.
Written client dependencies with dates. If content is due on the 10th and arrives on the 24th, the timeline moves and that is documented before it happens rather than argued about after.
A change order process that is genuinely lightweight. If requesting a change requires a formal amendment, you will skip it and absorb the cost. A one paragraph email with a price and an approval reply is enough friction to make people think and little enough to actually use.
Deposit of 40 to 50 percent up front. Non negotiable for me, because it solves the cash conversion problem discussed in agency profitability metrics and it filters clients who are not serious.
Retainers
My default for ongoing work, and the model I would recommend to most service businesses, with one large caveat.
Advantages are real: predictable revenue, predictable capacity planning, a relationship that gets more efficient over time as context accumulates, and vastly better cash conversion when billed in advance.
The caveat is scope drift. Retainers rot slowly. Month one is exactly what was sold. By month nine the client has added four recurring requests, each individually trivial, and your margin on that account has fallen 15 points without a single conversation.
Three defenses:
Define the retainer as a scope, not as hours or as availability. "Four articles, two landing pages, monthly reporting" is enforceable. "20 hours a month" invites hour counting and "ongoing support" invites everything.
Review the scope every six months as a scheduled event, not as a confrontation. Bring the data on what was actually delivered versus what was sold. Half the time the client agrees to an increase without argument because they can see it.
Build in an annual increase. Three to five percent, stated in the original agreement. It normalizes the idea that price moves and prevents the awkward jump after four flat years. The psychological framing here matters a lot, and I covered the specifics in retainer pricing psychology.
Performance pricing
The most seductive and most misunderstood model.
It sounds like perfect alignment and mostly is not, because you rarely control the variables you are being paid on. You can drive qualified traffic and lose on a sales team that does not follow up. You can improve conversion rate and lose because the client raised prices. You are taking equity-like risk for fee-like compensation.
The conditions under which I will consider it: I control the full funnel or close to it. The measurement is unambiguous and instrumented before we start. There is a base fee that covers my costs regardless. And there is a cap or a defined term, because unbounded upside sounds great until you are delivering 400,000 of value for a 40,000 fee and the relationship curdles.
Model it before you agree. Run the realistic case, not the optimistic one, through a ROAS calculator and ask whether the fee at plausible performance is worth the hours. Twice this exercise has told me the answer was no on deals I badly wanted.
Also insist on clean attribution. If you are being paid on results, you need to be able to prove them, which in the current tracking environment is a real project of its own. The relevant groundwork is in marketing attribution after cookies.
The hybrid I use most often
Base retainer that covers cost and a reasonable margin, plus a performance component on a metric I genuinely influence, capped.
It signals confidence, which wins deals against agencies quoting flat fees. It protects downside. It caps the client's exposure so procurement can approve it. And it puts the conversation on outcomes without making you a co-founder without equity.
The number that matters in structuring it: the base should cover roughly your full delivery cost plus a thin margin, so a zero performance quarter is survivable rather than fatal.
Choosing, in three questions
Can I scope it accurately from experience? If yes, fixed project or retainer. If no, paid discovery first, then decide.
Is the work ongoing or finite? Ongoing goes to retainer. Finite goes to project. Do not run finite work as a retainer, because you will be paid for maintenance you are not doing and the client will eventually notice.
Do I control the outcome variable? Only add performance components where the answer is genuinely yes. Enthusiasm is not control.
Raise the price on your next proposal
Here is the specific action. On your next proposal, quote 20 percent higher than you were going to, and add one concrete scope limit you have never written down before.
Two things will happen. Some prospects will accept without comment, which will tell you that you have been underpricing, and that number is now your floor. Others will push back, which gives you a real conversation about scope that you would otherwise have had six months into the engagement for free.
I have never met an agency owner whose problem was charging too much. I have met dozens whose problem was a pricing model that quietly transferred every risk in the engagement onto their own team, and who found out only when the margin report finally got built.