How to Think About B2B Pricing in the Age of GenAI
Huge numbers of companies will have to rethink the way they price
Back in 2005 (!), I diligenced a company that offered a somewhat specialized service that was done by humans. (I’m purposely being a bit vague for reasons that will be clear in a second.) New technology was automating much of what these people were doing. An average job used to take 10 hours a few years ago. Now, it was down to 8 hours. In the next couple of years, we thought the technology could reduce it to 4 hours. In the past, they’d been able to find more jobs as the number of hours came down and had also increased the price per hour. We were worried that they couldn’t keep pulling those levers as the hours collapsed. We encouraged our client not to do the deal, they eventually backed out, and the company ended up struggling (although I believe they have pivoted a bit and are doing well today).
I’ve been thinking about that case a lot in the past few months.
GenAI is going to impact a range of business models. So, it’s time to take a hard look at your pricing model while there’s still time to change things. Traditional modes of pricing by seat or billable hour may be at risk as GenAI introduces efficiencies that break these models. And this will be different from the ongoing march of automation over the past decades, GenAI represents a discontinuity with other technologies: suddenly tasks that seemed very hard to automate with accuracy (e.g., translation) can be done at near human levels.
Companies will insist that they can change their pricing system to match this change, but it won’t be easy. The longer they wait, the more value they may leave on the table.
Why Pricing by Seat Can Be Dangerous
One common way to price software products is by the number of seats, or users, who have access to the product. This makes sense when the value of the product is proportional to the number of people who use it. Also, as companies come to value the software, the hope is that more people will use it, which will increase revenue per customer. This is a long-time model for SaaS companies, but with GenAI coming, it suddenly feels risky.
GenAI could disrupt this model in two ways today:
The company could add GenAI features that make the users more productive. The result would be fewer people to do the same amount of work and, therefore, fewer seats are needed. Examples include software for call centers, for law and accounting firms, for creative agencies, etc. The good news is that the software is creating real value for customers in this scenario. The job that once took 8 hours now takes 4. An economic surplus has been created, but the big question is how it will be split between the software company and the customer. That will be a negotiation. If you are pricing per seat, you potentially start that negotiation from a weak position.
The second (worse) situation is if you are downstream of a company doing number 1. For example, if you sell software to law firms to track billable hours, and new software means that companies can do the same amount of work with fewer paralegals and associates, the per seat model could mean less revenue per customer. The difference with the first scenario is that you didn’t create the benefit, which means it will be even tougher to migrate the customers to another pricing plan once these changes have played out.
And it could get much worse! Imagine if companies are just using agents to talk to your software, and there are no humans at all to take up seats.
The solutions are, of course, either raise per seat prices by a lot more than usual or switch to a different pricing plan. Platforms do have valuable company data that users need to access, and with GenAI, the data is becoming even more valuable. If customers are getting value from the data, the company can show the productivity gains are real, and the company has the discipline to pass through price increases, then price increases alone may be the solution.
For many companies, those large per-seat price increases may not be possible. In that case, companies will need to change the meter: charge a flat fee, charge based on the size of the customer, charge based on usage, charge based on individual modules and features, or even a hybrid model that combines one of more of these with the existing per seat system. As I’ll discuss below, though, it’s easier said than done.
Why Services Businesses May Also Need to Change
A common way to price services is by the billable hour, or the amount of time spent on a project or task. If it takes less time to do the task, then you’ll have fewer hours to bill.
GenAI may increase the productivity and efficiency of service providers, allowing them to deliver the same or better results in less time. We see this in consulting where new tools are already making our teams more efficient. In premium consulting, we see this as an opportunity to go deeper on the answer and ultimately provide better recommendations. This is no different to other technologies over the years that have made slide production and analysis easier. A three-week diligence today has many more insights than one 20 years ago when I first started at Bain.
But, that won’t be true for everyone. For some companies, there may not be a way to go much deeper, it may simply be that much easier to deliver the work.
As with software, there are plenty of ways to rethink pricing to preserve revenue (and possibly improve margin substantially as cost to serve drops): charge a fixed rate based on the value of the project, the impact or results of the work (i.e., a percentage of money saved), lower prices to drive up volume (without reducing margin percentage), shift more towards a software delivery model with an annual license, etc.
While these strategies may work, just like with software, it is hard to change pricing levers.
Why Switching Pricing Meters Is Hard
Picking a new model is not easy. It requires a lot of research, experimentation, communication, and negotiation. It also involves risks, such as losing customers, aiding the competition, or triggering legal issues. Some of the challenges that software and services companies might face when switching pricing levers are:
Finding the right pricing lever that aligns with the value created by GenAI and the willingness to pay of the customers
Testing and validating the new pricing model with existing and potential customers and collecting feedback and data to optimize and refine it
Communicating and explaining the new pricing model to the customers, and addressing their questions, concerns, or objections. Some customers will be winners and some will be losers under the new model, and companies will quickly figure out which ones they are
Negotiating and contracting the new pricing model with the customers and resolving any disputes or conflicts that might arise (this is the hard one). Discipline is super important. If you only migrate the companies that pay less under the new system and not the ones that would have to pay more, you will have a big problem.
Companies without disciplined migration plans inevitably end up with an unmanageable web of legacy pricing schemes that add complexity without mitigating the risk
Managing the transition from the old to the new pricing model, and dealing with any technical, operational, or financial issues that might occur. In fact, even building the infrastructure to measure the new pricing model can be hard. If you move to usage-based pricing, how do you build the ability to track and bill consumption of the product? This can be tough for legacy software
There are lessons to be learned from the last big transition when the software industry moved from on-prem to SaaS. Designing an effective system of incentives and credits can help accelerate transitions, but this can be especially tricky if you have multi-year contracts with customers. That means that it could take many years to flow these changes through the full book of business.
This all sounds a bit gloomy, so I want to end on a positive note.
But GenAI also might let you raise prices!
If you add GenAI features to your product or service, and they work well, you are creating additional value for your customers. As mentioned above, that should translate into higher revenue per customer. That can be achieved in two ways:
Charge extra for GenAI features, selling them as separate modules or add-ons
Use the value from GenAI to justify above average price increases in the future
For an (admittedly B2C) example of the first one, see Amazon’s recent announcement to put Claude into Alexa and charge $5-10 per month for the upgraded version. I will be interested to see what the uptake is for that service.
Of course, any pricing strategy is helped if you have data showing that customers are using and getting value from the new services.
So, to sum it all up, what does this mean? If you are currently pricing per seat or per hour, I recommend that you think about whether you can or should migrate to a new model. Today, when GenAI impacts are still quite modest, there is still time to recontract at terms that have total revenue looking like today. If you wait too long, the negotiations will get tougher. Good luck!
P.S.: Big thanks to Bain pricing experts Josh Sandberg and Mark Burton who gave me several excellent suggestions that made this piece much better.
Great article! That has always been true about seat based pricing, but I agree that it could be accelerated by efficiency gains through AI. For what it's worth, I think headcount/layoff trends have probably had a greater impact on licenses in the past few years. Either way the risk is there, and seats are often an imperfect proxy for value when quoting a price.
My 2 cents is that flat-fee pricing will be the best model for B2B pricing moving forward, since buyers prefer the predictability. And 100% "use the value from GenAI to justify above average price increases in the future."
A really interesting and informative article. Thanks