How to choose a billing model that protects your margin (not just your comfort zone)

Most guides treat picking a way to charge clients like picking a paint colour: here's the swatch card, choose what you like. That advice is now wrong, and it's costing firms margin every month. Before I joined Teamwork.com, I spent nine years in agencies watching the same thing happen, and it always started with a model nobody had really chosen on purpose. This guide reframes the decision around one idea, then hands you a test to make it. Your way of charging is a bet on how long the work takes, and AI just moved the goalposts.

What a billing model actually is (and why it isn't your pricing strategy)

If the model is a bet on time, it's worth being precise about what you're betting with. A billing model is the structure you use to convert delivered work into an invoice, and your pricing strategy decides the number while the model decides the mechanism that generates it.

For example, "we charge premium rates because we're specialists" is a pricing strategy. "We invoice a fixed monthly fee against an agreed scope" is the model that carries it.

People blur the two, and it matters because you can hold a smart pricing strategy inside a model that leaks money anyway. I've seen firms win a premium rate, then bill it hourly, and hand every efficiency gain straight back to the client. The strategy was sound. The mechanism underneath it gave the money back.

The reason this distinction is worth labouring is that most firms spend real energy on the number and almost none on the structure. They benchmark their rates, study competitors, and agonise over a proposal, then default to whatever billing mechanism they've always used without asking whether it still fits. That default is where margin leaks, invisibly, one invoice at a time.

The seven models in common use are hourly or time-and-materials, fixed fee, milestone, retainer, value-based, outcome or gainshare, and productised subscription. Each one already has a solid primer, so I won't relitigate the basics here.

If you want the wider overview, we've covered agency pricing models, consulting pricing models, and how to match a pricing strategy to your business in depth. What those pieces don't do is tell you how the choice behaves now that delivery time is collapsing. That's the gap this guide fills.

Why the model you pick now decides your margin

Here's the part the swatch-card advice misses. Every model quietly assigns the time-risk to someone. Hourly puts it on the client: the longer the work takes, the more they pay. Fixed fee, value-based, and outcome models put it on you: if the work runs long, you eat it, and if it runs short, you keep the upside.

That single fact used to be a footnote. It's now the whole story, because the amount of time work takes is moving faster than at any point in my career.

Retainers show this most clearly. A retainer that started at a healthy margin drifts, because it bills a flat fee while the work quietly expands. On a $20,000-a-month retainer, it only takes a few unlogged favours a week to turn a 30% margin into a rounding error by quarter's end. The model didn't fail; the absence of a mechanism to see the drift did.

Consider a simple example. Say a project takes 100 hours at a $150 blended rate. Bill it hourly and you make $15,000, no more, no less.

Quote it fixed at $15,000 and then deliver it in 70 hours because your team got sharper, and you've just earned an effective $214 an hour. Same work, same client, wildly different economics, decided entirely by the model.

$$\text{Effective rate} = \frac{\text{\textdollar 15,000}}{70\text{ hours}} \approx \text{\textdollar 214/hour}$$

When budgets tighten and timelines shrink, the model stops being an admin choice and becomes a margin lever. Pick the one that rewards efficiency and you compound every improvement your team makes. Pick the one that punishes it and you're running to stand still.

The seven billing models, at a glance

Before you can choose well, you need the options side by side, framed by the one thing that matters: who carries the time-risk. Here's the short version, a seven-model time-risk map built as a decision aid rather than a glossary.

A few of these deserve a sentence more than the table allows. Hourly is the honest default when nobody yet knows the shape of the work, but it's also the model AI hits hardest, which we'll get to. Fixed fee rewards a team that delivers efficiently, provided you quoted it against real numbers rather than hope.

Milestone billing is underused. Tying payment to defined checkpoints protects cash flow on long engagements and gives the client a natural place to confirm scope before the next phase starts.

Retainers I'll keep short. We've covered whether a retainer agreement fits your firm and how retainer versus project work affects profitability elsewhere. The one-line version: retainers are the easiest model to quietly lose money on.

Value-based pricing, as HBR's primer lays out, gets the most attention and the least honest implementation. Charging for value only works when you can name the value, agree it with the client, and prove you delivered it. Outcome and gainshare models take that further by tying your fee to a result, which pays when the result is measurable and mostly in your control. Productised subscription packages a repeatable service into a flat recurring price, which clients increasingly like because it feels predictable.

A quick worked example shows why the model, not the rate, moves the needle. Take a brand refresh you'd normally bill at 60 hours and $9,000. If it reliably lifts a client's conversion in a way you can measure, a value-based fee of $18,000 is defensible, because you're pricing the result rather than the hours.

Bill the same project hourly and you cap yourself at $9,000 no matter how much value you created. The catch is proof: value-based only holds if you agreed the metric up front and can show you moved it.

Hybrid models deserve a mention, because they're where a lot of firms are quietly landing. Pairing a retainer for access with a performance element on top lets you keep predictable base revenue while sharing in the upside when you deliver. It's more to manage, but it hedges the time-risk in both directions, which is why it suits firms mid-transition off the hour rather than those making a clean switch.

Where usage-based billing fits in services (and where it doesn't)

Those seven models all price time or scope. There's a family that services borrowed from software worth knowing too: usage-based billing, where you charge for what the client actually consumes rather than the time you spent. As more delivery runs through platforms and AI, that model is creeping into services.

Usage-based billing, also called consumption-based, metered, or pay-as-you-go, comes in a handful of standard shapes.

Set against the alternatives, the trade-off is clarity versus predictability. A flat subscription or per-seat fee is easy to forecast but drifts away from the value delivered. Usage-based billing tracks value closely, but it makes revenue harder to predict, which is why most services firms reserve it for a defined, countable output.

Where it fits services well is productised or AI-delivered work with a clear unit: a report run, an audit, a generated asset, or a resolved support ticket. Where it struggles is bespoke strategy, where a unit of usage is meaningless and the value lives in judgement. I'd be cautious pricing genuine advisory this way.

The operational catch is the one this guide keeps returning to. Usage-based billing only works if you can meter what's consumed, rate it accurately, and invoice it cleanly, which is a data problem before it's a pricing one. That's the bridge to choosing well: whatever model you lean toward, you need to see the numbers first.

The Billing Model Fit Test: matching the model to the work

You've seen the full menu now, usage-based included. The harder question is which model fits the engagement in front of you, and most guides end at "it depends."

That's true and useless. So here's a repeatable way to decide, which I've come to think of as the Billing Model Fit Test. Score an engagement across four levers before you pick how to charge for it.

1. Scope clarity. Do you genuinely know what "done" looks like, or are you still discovering it? Low clarity points to hourly or milestone; high clarity opens up fixed fee.

2. Outcome measurability. Can you attach a number to the value the client gets, and will they agree it? High measurability is what makes value-based or outcome pricing defensible rather than aspirational.

3. Delivery-cost visibility. Do you know, live, what this work actually costs you to deliver? Without it, every fixed and value-based quote is a guess dressed as a decision.

4. AI exposure. How much of this work is the kind AI is already compressing? High exposure is a warning light on any hourly line item.

Here is the same test as a quick scorecard, the Fit Test scoring grid, so you can run it on any engagement in under a minute.

Read the levers together and the model falls out. Low scope clarity and low measurability: stay hourly or use milestones while you learn. High scope clarity with strong cost visibility: fixed fee, and keep the efficiency gains.

High measurability and high trust: value-based or outcome. High AI exposure on well-understood work: move off the hour before your client asks you to.

Here's the test applied. A recurring content engagement has clear scope, moderate measurability, and high AI exposure, because drafting is exactly what AI accelerates. Billed hourly, that account shrinks as your team gets faster.

Repackaged as a productised retainer with a defined monthly output, the same work holds its price while your cost to deliver falls. That's the Fit Test doing its job: it doesn't tell you your favourite model, it tells you the one that survives contact with how the work has changed.

Run it on a different shape of work and you get a different answer. A one-off systems migration with a tightly agreed spec has high scope clarity, low ongoing measurability, and moderate AI exposure. That points to fixed fee or milestones, not a retainer, and certainly not open-ended hourly.

The levers don't all carry equal weight, either. When delivery-cost visibility is low, fix that first, because a confident answer on the other three levers is worthless if you're guessing at what the work actually costs you.

Why AI breaks the billable hour (and what quietly replaces it)

AI exposure was one of the four levers, and it's the one quietly rewriting the others. Picture the task you'd have quoted at 20 hours two years ago. Your team now does it well in four, because the tooling got good.

Bill that hourly and you've just cut your own invoice by 80% as a reward for getting better at your job. That's not a hypothetical anymore; it's the maths playing out across professional services right now.

I billed plenty of work by the hour in my agency years. Back then it was fine, because time and value moved roughly together, and getting faster wasn't yet the same as getting paid less.

The instinct is to assume AI makes the work cheaper. It doesn't, and that misread is dangerous. AI is a supercharger for the low-value parts of delivery, the admin, the first drafts, the reporting, while the expensive part, the judgement, still needs a person. What changes isn't the value of the work. It's that the hour, as a unit of billing, no longer tracks the value at all.

So what replaces it? The same models that were always available, now with real urgency behind them. Fixed fee suits defined work, value-based and outcome pricing suit measurable results, and hybrid setups pair a retainer for access with a performance element on top.

The through-line is that you charge for the outcome, not the hours. That matters because Teamwork.com's research found 33% of leaders now say clients think they can do the job themselves with AI. If your invoice is a receipt for hours, that argument is hard to win. If it's evidence of impact, it's a different conversation.

The pressure isn't landing evenly, either. Consulting, IT services, and accounting firms are feeling it first, because their product is expert time, which is exactly what clients now question. Firms with a clear, embedded, hard-to-unbundle offer are holding their prices better, because value that's tangible is harder to argue down.

The lesson isn't to bolt off the hour overnight. It's to move the work that's genuinely measurable, controllable, and easy for a client to buy, and leave the genuinely uncertain work where hourly still earns its place.

Whatever model you move to, the client conversation decides whether it lands. A new model framed as a price change invites a fight. The same model framed around what the client now gets, faster delivery, clearer outcomes, and senior attention on the work that matters, reads as an upgrade. The firms handling this well are having the conversation early, before budget season, rather than defending an invoice after the fact.

There's also a cost you may not be pricing yet: the AI itself. Compute, licences, and the tokens your delivery now consumes are a genuine cost of goods, and the firms getting this right are building them into the model rather than absorbing them silently.

This is part of a bigger repositioning. Teamwork.com's research found 58% of leaders say their role has changed but don't yet know into what. It's the same problem at the firm level: you're being asked to charge for what your business is now, not what it used to be. Across the industry, firms are already wrestling with how to reinvent the way they charge clients as the billable hour comes under pressure.

Five billing-model mistakes that quietly kill margin

The move to a smarter model goes wrong in predictable ways. These are the five that cost firms the most, and none of them announce themselves; they show up three months later in a margin report.

1. Treating the model as permanent. The model that fit a client two years ago may be actively wrong now. Firms review their pricing and never revisit the mechanism underneath it. The fix is a standing review: put the model itself on the renewal agenda, not just the rate.

2. Quoting fixed fees without delivery-cost data. A fixed price is only as good as the numbers behind it. Quote from memory and you're gambling; quote from real cost history and you're deciding. Real cost history turns a fixed quote from a hopeful guess into a margin decision you can stand behind.

3. Letting retainers overservice into break-even. This one I know from the inside. Scope creep rarely starts with an unreasonable client ask. It starts with a small favour nobody logs, and by month three there's no paper trail showing why a profitable retainer went flat.

4. Pricing value you can't measure. Value-based billing without evidence is just a higher number the client hasn't agreed to. If you can't show the impact, you can't hold the price. Agree the metric before the work starts, capture it as you go, and the price defends itself at invoice time.

5. Switching models without explaining why. A new model lands badly when it looks like a price rise in disguise. Walk the client through what changed and what they get, and the same change reads as a partnership move.

Pro tip: Make mistake three impossible to miss. Budget alerts in Teamwork.com's cost and profitability tools flag a retainer heading past its hours before the month closes, not after.

The firms that dodge all five have one thing in common. They use real time and utilisation data to adjust how they structure billing and contracts, rather than guessing at renewal. That's the difference the next section is about, because every one of these mistakes traces back to missing data.

The delivery data you need before you switch models

You can't defend a model you can't measure, and you can't switch one safely either. Before you change how you charge, you need four numbers close to hand, live rather than at the quarterly review.

• Delivery cost per deliverable or project: what the work actually costs you, fully loaded, not what you hoped.

• Billable versus non-billable time: where the hours really go, so you know what's recoverable.

• Margin by project and by client: which work and which relationships actually pay.

• Planned versus actual: how your estimate held up, so the next quote is sharper than the last.

Get these in one place and the risk drains out of changing models, because you're no longer guessing at the effect. SugarCRM is the clearest proof I've seen: by unifying projects, time tracking, and billing, they reached near-perfect invoicing accuracy, less than $20,000 credited on over $10 million in annual invoicing. You don't get that kind of accuracy from a spreadsheet stitched to a timer. If you want a fast way to sanity-check one of these numbers, our billable utilisation rate calculator is a decent place to start before you commit to a model change.

Planned versus actual is the number people skip, and it's the one that compounds. Every project you compare against its estimate makes the next quote sharper, which is what turns a fixed-fee model from a gamble into a repeatable margin. Over a year, that feedback loop is the difference between a firm that prices with confidence and one that keeps flinching at renewal.

How Teamwork.com helps you price, bill, and defend any model

This is where I'll talk about the platform directly, because the data problem above is exactly what we built for. Generic project management tools can track tasks but can't see money, so they leave you flying blind on margin. Traditional PSA systems can see the money but teams resist using them, so the data going in is unreliable and the numbers coming out are worse. Teamwork.com is the agentic PSA that connects projects, resources, and financials in one place teams actually want to work in, which is what makes the numbers real enough to price against.

• Budgets for every model. The problem with most tools is they assume one way of charging. When you run fixed-fee, time-and-materials, and retainer work side by side, you need a budget that flexes to match. Budgeting and profitability supports all three budget types with margin targets you define, so the model on the invoice matches the model in the system.

• Quote with the margin already visible. Quoting from a blank page is how underpriced fixed fees happen. Quoting and costing work shows cost and profit live as you build the quote. Custom billable and cost rates by role, project, or client mean lever three of the Fit Test is answered before you send anything.

• Turn tracked time into defensible rates. You can't price value if you don't know where the hours went. Built-in time tracking captures billable and non-billable time against real rates, which is the raw material for every other model on this page.

• See margin by project and client, live. Finding out you lost money after the invoice is a post-mortem, not a decision. Profitability reporting shows margin by project and by client as work happens, so you can spot a model that's leaking before it becomes a pattern.

• Let AI forecast the outcome, not just log the hours. The whole point of moving off the hour is charging for outcomes, which means you need to see them coming. The AI Profitability Forecaster reads your live delivery data to project cost, margin, and where a project is heading. That gives the person owning the client time to act while it still matters. Our AI Teammates handle the status-chase and reporting drudgery so that judgement, the part clients actually pay for, gets your attention.

If any of that sounds familiar, if you've ever discovered a dead account only when the numbers finally caught up, you're not bad at your job. You were working without the instruments. That's the gap worth closing before you touch the model.

Billing model FAQs