AI is here to eat your seats

We’ve been giving a lot of thought to how artificial intelligence will change seat-based pricing models. Companies that build software for customer service, recruiting, human resources, lead generation, and outbound marketing tend to charge by the number of humans using the platform. Yet if those same companies start building automated AI features, it means AI features that might reduce or replace headcount. In some cases, implementing AI could mean building self-cannibalization. 

How does that change revenue strategy and roadmaps for seat-based enterprise software companies? And how can founders position their companies so they're on the safe side of the innovation line?

Here are some questions and potential solutions to this conundrum.

‍

What are we talking about?

Seat-based pricing is when a company charges by the number of access points to its software—in the form of licenses, accounts, or logins. This doesn't necessarily equate to the number of employees who can use the software. However, the number of seats often equals the number of employees using it. Which means the loss of a user means the loss of revenue.

Seat-based pricing stands in contrast to consumption-based pricing, where customers pay by the software they consume. Vendors with competing products often differentiate with one model or the other. The right decision for the customer depends on how they will use the product.

‍

What happens when AI gets involved?

The rise of generative AI has meant software companies can automate an array of jobs encompassed by their products. Asana just started offering an "AI teammate" that flags issues such as missing information in documents, can advise on ideas, and learns about how an organization works. Web design companies like Webspot now offer AI-generated sites. Adobe has added AI to Photoshop, Acrobat, and other products as they deal with more AI-native approaches from companies like Runway and Canva (both of which are Felicis investments). Plenty of companies are simply painting "AI" on the same old software, but most leading software companies are taking this particular movement in tech very seriously. There's a reason Nvidia is worth trillions of dollars. 

So far, customer service is one of the first industries to really feel the effects of AI on its workforce. While executives are happy with those overhead cost savings, those are also "seats" that a software provider somewhere counted on in their quarterly forecasts. 

‍

What options do founders have?

Raise prices? 

One might see the rise of AI as a boon for seat-based software models. If a company stops building "low-end" features (those which AI solves for their customers) and puts R&D behind functions only humans can manage, they will "move up" the value chain to become premium software—and can charge premium prices. 

But not so fast. Going "premium" means spending more on engineering and development. And a "better" product doesn't necessarily translate to higher margins. At the same time remember that AI is functioning as the Clay Christensen low-end disruptor. If, say, a recruiting software company that has abandoned building low-end features shifts upstream from lead generation to negotiation and relationship management, they leave their flank open for an upstart to use led gen as a toehold to eat their business from below. And if a company like this only builds AI that automates the low-end features, it likely means that many competitors will be able to do the same, creating a price war to deliver similar value.

‍

Add features and charge for them?

One option companies have is to build and ship new and impressive features and build them into their cost structure. Take Figma and Notion, which are taking different, but related routes. For Notion, you get a certain amount of complimentary AI responses before you have to pay to add AI to your workspace. Smartly, you cannot provision AI to individual users, it has to be the whole workspace.

Figma became a multi-product company with the introduction of FigJam, and they charged per seat to use that product. When they recently announced their AI-powered Figma Slides presentation product, it was another per-seat charge in addition to the existing Figma seat. However, for the grander AI vision at Figma, CEO Dylan Field says they are eating the cost of AI for this year while they figure out how customers use the features and which prove most popular. 

‍

Change metrics?

Seat-based companies will need to distill the true value they're delivering for their customers and then define it in a measurable aspect of performance. This can mean charging by the amount of data stored or transferred, calls to an API, time on the platform, number of features implemented, issues resolved, and essentially any other creative metric a company can think of to "meter" the use of their product.

For customer service software, this equates to resolutions. Did the customer get their question answered to their satisfaction? Already, we're seeing incumbents shift towards this model. Zendesk just published a note explaining how they will stop charging per monthly user (per seat) and instead charge per "automated resolution"––issues resolved by a Zendesk AI agent.

But charging by metric isn't as easy as it sounds. Here's what counts for an automated resolution "in conversation" with a customer. Per the post:

• The end user experiences any of these events over the course of the conversation:some text
  • The bot has recommended at least one article as part of the standard bot fallback response or
  • An AI-generated response or
  • When the end-user has reached the end of the conversation and completed the last step of the answer flow.
• The conversation has not been transferred to a live agent (that is, a Transfer to agent step has not been reached).
• The end-user hasn’t interacted with the conversation in the last 72 hours. End-user interaction includes entering free text, clicking on any conversation button or option, entering information into a form, and the like.

It's easy to see how discrepancies will arise; imagine if your internet carrier only billed for "data that answered your question.” The bill would be madness. All this is to say, even if a company lands on the best metric for charging customers, the implementation may still be painful. For many founders and their revenue teams, it will come back to their data and how they can charge based on it, which is why a company like Orb—which automates billing based on usage data—is seeing fast adoption.

‍

Change the model?

For young companies, the answer may be to change the model altogether. Growth Unhinged has a great post about 11x.ai, an automated sales development representative software company that's gone from $0 to $2M in just six months, according to founder and CEO Hasan Sukkar. It's a great example of dogfooding—Sukkar says their product drives 70-80 leads per week for the company—but 11x is bucking the traditional path to revenue. 

SDR companies tend to charge per lead. 11x is trying to charge only for the work done: tasks such as identifying accounts, research, outreach, and scheduling meetings. Sukkar is betting that customers will want to pay for the incremental work rather than the thousands of dollars they're accustomed to paying for qualified leads. As the author notes, "We’ll need to circle back with Hasan to learn more about how the new pricing model is received in the market."

‍

What about a hybrid?

Seat-based pricing offered a simplicity that worked for both customer and vendor. In fact, it might have let vendors get a little complacent. The way a restaurant that offers a buffet doesn't need to know its customers’ favorite dish, seat-based models can rest on overall usage rather than the specific value they provide. 

Ultimately, founders must think creatively about what metrics truly define the value they're delivering. They will find themselves offering hybrids of the above and perhaps layering in tried-and-true models such as subscriptions, credits, and usage tiers. Because overusing tokens is costly, Runway charges its users for additional credits beyond the allotment provided with their subscription seat.

This has big implications for the product. As AI spreads through software, the best-positioned companies will build platforms that are dynamic enough to meter new features and capabilities. Product roadmaps must be ready to accommodate a shift in business model as the company grows and adjusts.

‍

A new seat at the table

Things are changing fast, but the basics will stay the same. No matter how a software company integrates AI, it will remain imperative that they clearly demonstrate value to the customer. In this instance, being clear means showing how their product stands head and shoulders above the non-AI alternative—or pairs with AI as a co-pilot. 

While seat-based pricing may remain the initial pricing layer for many software products, the mechanics will evolve in some industries sooner rather than later. Because of the marginal cost of using AI models, we’ve entered a world where thoughtful pricing is more important than ever. AI companies need to ensure they deliver enough value to charge significantly more than what it costs to deliver the outputs. This is why we see pricing based on value delivered or tied to a business KPI as solid potential models.

Here are some industries we see needing modernized pricing models that go beyond a simple seat-based structure:

1. Customer service - charging per ticket closed or customer satisfaction
2. Productivity software - charging by tasks preformed, and modules (features)
3. HR software - charging by outcomes (interviews, candidate responses), and modules (features)
4. The entire marketing stack - charging based on campaigns, data collected, and/or traffic
5. Financial software - charging by data consumption, models used, asset growth, and modules (features)
6. Revenue operations - charging based on processed revenue, and modules (features)
7. Dev tools - charging by copilot, code output (or code saved), code scans, and modules (features)

While there are many industries where AI will affect pricing, the key point is founders in these verticals need to be prepared to face a true innovator’s dilemma. They do not want to discourage use of their platform, but they need to ensure their revenue can grow. 

At the beginning of the smartphone era, many misguided efforts were made to "fight" the wave. We know how that's worked out. The companies that transition out of seat-based-only models will be the ones that treat AI as the catalyst for changing their business and landing strongly in the next phase of the market.