From billable hours to products: set a fixed fee from your time tracking data

A firm that bills by the hour can also sell part of its services as products, at a fixed fee. You do not have to guess that fee: the basis is already in your time tracking data. Map what each type of service really costs, and you can quote a price you are confident about.

In short

• When AI does the research and the first draft, billing by the hour earns you less for exactly the same result. A fixed fee per product rewards you for working faster.
• Not every service is a product. Work with a fixed shape suits a fixed fee; true bespoke work often stays hourly.
• You can derive the cost per service from data you already have: your own time records.
• If the activity labels in your time records are inconsistent, you can relabel them after the fact with AI, with a person checking the result.
• Per employee you can see who spends how much time on which type of work. That helps with planning, delegation and pricing.

Why is the hourly rate under pressure?

Lawyers, accountants, tax advisers, family offices and other advisers have traditionally sold time. That works as long as time and value rise together. AI pulls the two apart. Research and the first version of a document go faster, while the value for the client stays the same: the advice is right, the memo is done, the return is correct.

A worked example shows the difference. Take a single advisory memo.

Illustrative example. Hourly rate, hours and fixed fee are assumptions; € 220 is a round figure, not a market rate. The cost of the AI itself is not included.

So the question for the coming years is: do you sell hours, or do you sell the result? Many firms are asking it. In the Thomson Reuters Future of Professionals research (2025), 80% of law firm respondents said they expect AI to fundamentally change how they work, especially how they price, staff and deliver their work.

Which services suit a fixed fee?

The likely outcome is a mix: fixed fees for work with a fixed shape, hours for bespoke work. Use this checklist for each service:

1. Does the service recur often? Annual accounts, a standard contract, an objection letter or a standard advisory report do; a unique dispute usually does not.
2. Can the end result be defined? If you can say in one sentence what the client receives, you can price it.
3. Is the input predictable? The more alike the client's documents are, the smaller the spread in hours.
4. Can you define additional work? A fixed fee works when you state up front what is not included.

A service that scores well on all four is a candidate product. If in doubt, start with the services you deliver most often. That is also where you have the most time data.

How do you get a cost price from your time tracking data?

Your time records are more than an attachment to an invoice. Over a year or two they show how long each type of work really takes, who did it and how much that varies per engagement. This is how you turn that data into a price:

1. Export the hours. Per line: date, employee, matter, activity, description and number of hours.
2. Define your services. Make a short list of the products you want to offer, in the client's language.
3. Link every line to a service. Add up the hours per matter and you see what one delivery of that service cost.
4. Look at the spread, not just the average. A few outliers pull the average up. Look at the middle value and at what happened in the difficult cases.
5. Use your cost per hour. Not the billing rate, but what an hour of that employee really costs you, plus overhead.
6. Set the price. Cost, plus margin, plus a risk premium. Then test what the client is willing to pay.
7. Review afterwards. Track per product whether it stays within the price, and adjust.

The result is a price you can quote with confidence. It also saves time: no quote per engagement, no discussion about the time sheet.

What if the activity labels in your time records are inconsistent?

In practice, activity labels are rarely consistent. One person writes "meeting", another "call", a third books everything as "general". The free-text description often says more than the label itself.

That is no reason to wait until the records are clean. You can relabel the hours after the fact with AI. A language model reads the description on each line and picks a label from your own list of services and activities. Where the model is unsure, a person decides. A sample check shows whether the labels are right before anything is calculated. The original records stay untouched: the label is added as an extra column.

The same principle helps going forward. Agree on one list of labels and keep it in one place. That keeps your data usable, for AI as well. Read more in One fact, one place.

What do you learn per employee?

With the same labelled hours you see, per employee, how much time goes to which type of work. Is a senior doing a lot of work a junior could do? Does specialist knowledge sit with one person? Who has capacity, and who is overloaded?

Use these insights to divide work better, train people where it matters and support your prices. They are not meant to judge individual employees. Discuss with your team beforehand what you will look at and why. If your organisation has a works council, it has a role here too (see "For the specialist").

How Prudai works

Prudai helps firms with this step as an innovation partner. We start with the question of which services you want to offer as products. Then we analyse your time records together. That fits the business case step of our five-step approach, in which we analyse the data sources and the quality of the data.

In practice:

• We relabel the hours after the fact with AI, against a list of services you define yourself. Your own people check the labels before anything is calculated.
• Per type of service you get the spread in hours and a basis for the cost price.
• Per employee you see how the work is divided.
• You set the prices yourself. If you work with AI, we also include the AI costs per product in the calculation: if you price per product, you need to know those costs per product too.

Prudai is certified to ISO/IEC 27001:2022 and NEN 7510-1:2024. Time records say a lot about your clients and your people, so we handle them with care.

For the specialist

Cost from time. The method above resembles time-driven activity-based costing by Kaplan and Anderson (Harvard Business Review, 2004). It uses two parameters: the cost per time unit of your capacity and the time an activity takes. Your time records supply the second parameter, your budget the first.

A fixed label list. If you do not want to start from scratch, look at UTBMS, the Uniform Task-Based Management System codes for legal invoices. They have separate codes for tasks and activities and are maintained by the LEDES Oversight Committee. A firm's own shorter list, in the client's language, is usually more practical, but the split between task and activity is a good starting point.

Dutch lawyers: a fixed fee is not a contingency fee. Article 7.7 of the Dutch Regulation on the Legal Profession (Verordening op de advocatuur) prohibits a lawyer from agreeing that a fee is only charged if a certain result is achieved, or that the fee is a proportional share of the value of the result. Exceptions apply to debt collection (article 7.8) and to personal injury and wrongful death cases (articles 7.9 to 7.12). A fixed fee for a defined product does not depend on the outcome, so it falls outside that prohibition. Do record clearly what is and is not included.

Employee data and the works council. Under article 27(1) of the Dutch Works Councils Act (Wet op de ondernemingsraden), the works council must consent to arrangements on the processing and protection of staff data (point k) and to facilities suited to monitoring the attendance, behaviour or performance of staff (point l). An analysis per employee can fall under this. The GDPR also applies to processing personal data.

AI costs per product. An AI service charges in tokens. Per item that is usually small compared with an hour of work, but it is not zero. In heavy workflows across many documents it adds up. Include it in the cost price.

Frequently asked questions

Isn't a fixed fee risky if an engagement overruns?

The risk is smaller if you base the fee on the spread in your own time data, not on an average. Also state up front what the fee does not cover. Anything outside that is additional work.

How much time data do I need?

The more comparable engagements, the more reliable the price. So start with the services you deliver most often. For rare work, a fixed fee makes less sense anyway.

May a Dutch lawyer agree a fixed fee?

Yes. The Dutch Regulation on the Legal Profession prohibits a fee that depends on the result (article 7.7), with exceptions. A fixed fee for a defined product does not depend on the outcome.

Should the time AI saves go back to the client?

Opinions differ. In the worked example both sides gain: the client pays a quarter less than before and the firm earns more per hour worked. A fixed fee makes that trade-off visible and open to discussion.

Sources

• Verordening op de advocatuur, article 7.7 (wetten.overheid.nl, version in force from 1 July 2026, consulted 6 October 2026)
• Wet op de ondernemingsraden, article 27 (wetten.overheid.nl, consulted 6 October 2026)
• Kaplan and Anderson, Time-Driven Activity-Based Costing (Harvard Business Review, November 2004)
• Thomson Reuters Institute, "Future of Professionals" report analysis: why AI will flip law firm economics (16 October 2025)
• UTBMS, Uniform Task-Based Management System (LEDES Oversight Committee)

Want to know how to set a fixed fee per service from your own time records? See how Prudai helps as an innovation partner, or read more about AI ROI in 2026.

Updated on 6 October 2026

Photo: cpastrick via Pixabay