Contents7
A web agency that bills by the day sells time. As long as the time needed to build a website, a campaign or an integration stayed stable, the model held. AI weakens it from the middle: every hour the team saves is one hour less on the invoice, and clients know it.
This article takes the economic angle. For the tools and uses of AI inside an agency, our article AI and digital marketing: what changes for agencies covers the subject.
In short
According to Bain, which sells strategy and AI consulting and analysed large listed technology service providers, clients award new contracts assuming at least 30% lower delivery cost, up to 50%.
Forrester, which sells research and advisory services, ranks software development, integration, creative work and localisation among the markets most disrupted by AI.
The largest firms’ answer is known: less time and materials, more recurring services and fixed or outcome-based prices.
Agencies are not disappearing: companies that adopt AI keep relying on providers, and what they buy is changing.
What is happening to prices
The clearest finding comes from Bain’s 2026 technology report; Bain sells strategy and AI consulting. Its chapter on technology services is not based on a survey but on an analysis of large listed providers: it describes the top of the market, not fifteen-person agencies. The trend it describes always trickles down to smaller players, though.
lower delivery cost, at least, assumed by clients when awarding new contracts (up to 50%)
efficiency requested at contract renewal, at most
of hours AI could save, from consulting and IT to infrastructure outsourcing
The mechanism is simple. A client reads everywhere that AI saves time; they conclude their provider saves time too, and ask for their share. Whether that gain is real on their project changes nothing in the negotiation. Bain draws an explicit recommendation from it: move away from time and materials towards outcome-based pricing.
Forrester, which sells research and advisory services, reaches the same place by another route. Its AI disruption model ranks more than 200 technology and service markets: the most exposed are labour-intensive knowledge-work markets, including transformation services, integration, software development, creative work, localisation and training. That is almost the list of a web agency’s trades. The model gives no market figures: it is a reading grid, not a measurement.
Why time and materials is the most exposed model
When billing for time spent, the agency and the client have opposite interests on productivity. If the team delivers in three days what used to take five, the agency loses two days of revenue. It therefore has no reason to adopt the tool that would make it faster, or it adopts it and watches its margin melt. The client, meanwhile, suspects the billed days no longer reflect the real work.
Time and materials
Every productivity gain reduces revenue
The client pays for effort, not results
Negotiation is about the day rate
The risk of overrun stays with the client
Fixed price or recurring service
Every productivity gain improves margin
The client pays for a deliverable or a service level
Negotiation is about scope
The risk of overrun moves to the agency
A fixed price is not risk-free: an agency that scopes badly loses money on every project. But it is the only model where AI works for the agency rather than against it.
What the large consulting firms are doing
The large consulting groups, which sell AI services themselves, show the direction. Accenture, which sells consulting, integration and managed AI services, published its results for the fiscal year ended 31 August 2026: for the first time, its managed services overtake consulting, with 43.67 billion dollars in bookings against 40.86. Recurring service, billed by the year, is becoming the core of the business.
BCG, which sells AI consulting, goes further on price. According to its CEO, quoted on BCG’s website, three quarters of its largest AI engagements carry variable fees tied to a client objective. He immediately adds that across the whole business it is significantly less than a third: outcome-based pricing is growing, it is not replacing everything.
Key takeaway
Outcome-based pricing is still a minority, even at the largest firms
The direction is clear, the shift is slow. An agency does not have to change everything at once: a fixed price on a well-defined deliverable is already a first step.
Agencies are not disappearing, their orders are changing
The story announcing the end of agencies forgets a stubborn fact: companies that adopt AI keep using providers. According to Eurostat, in France 22.5% of small companies that use AI go through an external provider, as do 30.5% of medium-sized and 49.1% of large ones (Eurostat, isoc_eb_ai, 2025 survey). The bigger the company, the more it relies on outside help.
Among the large companies BCG surveyed for a study on technology providers, two thirds expect their providers to build and run their priority use cases: not just to deliver, but to operate. The sample is small (just over 115 company executives), but the demand it describes resembles what we hear.
The North American mid-market shows the same ambivalence. RSM, which sells audit, consulting and managed AI services, surveyed companies with 30 million to 10 billion dollars in revenue: those increasing their AI budget plan to cut other lines, including external consulting (40%), while 52% already use external consultants. The client wants fewer billed hours, and still wants help.
Agency work is therefore shifting: less pure production billed by the hour, more design, integration with the client’s tools, operations and review.
What becomes of agencies that move to fixed prices
A reasonable transition fits in four steps.
- Break the offer into deliverables.
An audit, a page, an integration, a campaign: each deliverable has a written scope, a price and a lead time. What cannot be broken down stays on time and materials, for now.
- Measure your own time with AI.
Before setting a fixed price, you need to know how long the deliverable really takes with current tools, review included.
- Add a recurring service.
Maintenance, rank tracking, running a CRM or automations: what is billed monthly stabilises revenue when projects get shorter.
- Keep humans where they matter.
Review, strategy, the client relationship and judgement on quality remain the core of the trade. That is what the client pays for once they no longer pay for hours.
From the field
We left the day-rate grid
On 24 July 2026 we decided to replace our day-rate grid (650, 550 and 450 euros) with packaged fixed-price offers, one page per offer. Our SEO engine, which has scanned our search data every day since 25 August 2026, produces drafts that we review before every publication: productivity gains improve our margin on a published price, they do not shrink an invoice counted in days.
Where to start
If your agency still bills mostly for time spent, start with a single deliverable: the one you produce most often and whose duration you know best. Price it as a fixed fee, test it with three clients, adjust. For the production chain itself (fast websites, controlled publishing, AI in production), see our page on modernising your web agency; the partner agencies programme explains how we work white-label with agencies that want to add automation or CRM to their offer. On that last point, see also why the CRM is becoming a service to sell. To place all this within an overall marketing strategy, our guide to marketing automation for small businesses is a good entry point.
Frequently asked questions
Will AI make web agencies disappear?
Nothing points that way. Companies that use AI keep using providers, and all the more so the larger they are, according to Eurostat. What is retreating is the sale of production hours. What is growing is design, integration and long-term operations.
Should you cut your prices because AI saves time?
Not necessarily. If you bill for time spent, the cut will come through negotiation. With fixed prices, you set the price of a deliverable and the productivity gain improves your margin. According to Bain, large clients already assume at least 30% lower delivery cost: better to anticipate than to suffer it.
How do you move from time and materials to fixed prices without losing money?
Start with a deliverable you know well, measure its real duration with your current tools, review included, and write a precise scope. Anything outside the scope is priced separately. A recurring service, billed monthly, absorbs the gaps.
What is outcome-based pricing?
It is a price tied to an objective reached at the client rather than to time spent. Even at BCG, according to its CEO, it covers significantly less than a third of the business. For an agency, it requires a measurable indicator that both parties accept before the project.
Sources
- AI disruption model (new tab)forrester.com
- Managed services overtake consulting (new tab)newsroom.accenture.com
- BCG’s website (new tab)bcg.com
- Eurostat, isoc_eb_ai (new tab)ec.europa.eu
- Two thirds expect their providers (new tab)bcg.com
- 52% already use external consultants (new tab)rsmus.com
Jim · founder of Brain Plus AI
Also in digital marketing
Understand17 min
AI and digital marketing: what changes for agencies
In three years, AI has automated ad buying, cut the clicks coming from Google and made content production almost free. What this changes for the agency trade, the tools that matter in 2026, and how to build AI into your processes without blowing up your costs.
Understand10 min
CRM for web agencies: why to choose an open one in 2027
A web agency lives on projects, monthly recurring contracts and its clients’ trust. In 2027, a modern open CRM pays off twice: to run its own sales and subscriptions without re-keying anything, and as one more service to sell to its clients, under its own brand.
Choose and price7 min
Marketing agency or in-house team: doing the maths
An in-house marketing team knows the trade and reacts fast; an agency brings specialist skills and absorbs peaks. The honest calculation compares full costs for the same volume of work, and most companies end up combining the two.