Why Salesforce Paid $3.6 Billion for a Company That Had Stopped Growing

Most SaaS founders I talk to carry the same quiet worry. The product works, customers are paying, the team is solid, and yet new ARR has flattened. Every quarter the plan to restart growth looks a lot like the last quarter's plan, with a little more spend and a few more features. If that sounds familiar, the story of how one support software company climbed out of that exact spot is worth your time.

I sold my company iContact for $169 million, and since then I've made a habit of studying SaaS and AI exits to understand what the buyer was really paying for. This June, Salesforce agreed to pay $3.6 billion for the company formerly known as Intercom, which now goes by the name Fin. It is a turnaround story, and it carries three lessons for any SaaS or AI CEO whose growth has plateaued.

Where the company started: close to zero net new ARR

Go back to late 2022. According to its CEO, Eoghan McCabe, Intercom was close to zero net new ARR. It was a mature support software company, and its growth was gone. Net new ARR is what is left after you add new customers and expansion and subtract churn, so a number near zero means the business was selling just enough to replace what it lost. Plenty of founders live in that zone for years, because the revenue is real and the customers are loyal, which makes it easy to postpone a hard decision.

• The base keeps paying. A mature product has revenue, brand recognition, and a customer list, so nothing looks broken from the outside.
• New sales only replace churn. Near-zero net new ARR means the sales engine is running hard to stay in place.
• The easy fixes are used up. When a category is mature, another round of feature work or another discount rarely changes the growth curve.

That is the starting point worth keeping in mind. The company that Salesforce agreed to buy did not have a growth problem that a better marketing campaign would solve. It needed a different product at the center of the business, and what it did next shows how that kind of shift can happen.

Lesson one: move fast on the new product

McCabe says that six weeks after ChatGPT 3.5 arrived, the team had a working AI agent called Fin. Six weeks is a short time for a company of that size to go from a new technology to something customers could use. Most mature software companies would have formed a committee, run a survey, and published a roadmap for the following year. This team built the thing, and it put the new product in front of customers while the technology was still fresh.

Speed matters here for a practical reason. When a new technology shows up, the first working product teaches you more than any amount of planning does, because real customers tell you what the agent gets wrong and what it handles well. A team that ships in six weeks gets months of learning that a slower competitor never sees. For a company with no growth to protect, the cost of moving fast was low, and the cost of waiting was high.

• Build a working version first. A rough agent that handles real tickets teaches you more than a polished plan.
• Use the plateau as permission. If growth is already near zero, you have less to lose from a fast experiment than from another quiet quarter.
• Let customers grade it. Their reactions tell you where the product is strong and where it needs work.

None of this requires a billion-dollar company. Any founder can pick the single workflow in their product where an AI agent could do the job a person does today, and get a version running in weeks rather than quarters.

Lesson two: price on the result, not the seat

The second decision was about pricing. Fin charged 99 cents per resolved ticket. If Fin does not solve the problem, the customer does not pay. That is a very different deal from a seat license, where the customer pays every month whether or not the product delivers anything in particular. Per-resolution pricing puts the risk on the vendor, and it makes the buying decision easy, because the customer only pays when the product has earned it.

Salesforce says Fin now resolves 76% of support volume on average without a human. By March, McCabe says, Fin was near $100 million in ARR. Those two numbers explain why the pricing worked. A product that resolves most tickets on its own can afford to charge only for resolutions, and every resolution is a visible, countable event that the customer can compare against what a human agent would have cost.

• Define the unit of value. For Fin it is one resolved ticket. Pick the event your customer would happily pay for every time it happens.
• Share the risk. No resolution means no charge, which removes the biggest objection a buyer has to trying new AI software.
• Let revenue track results. When the product gets better at the job, revenue grows with it, with no renegotiation needed.

Seat-based pricing made sense when software was a tool that people used. An AI agent does the work itself, so charging per seat starts to look odd, and charging for the outcome starts to look obvious.

Lesson three: bet the whole company on it

The third move is the one most founders avoid. In May, the company renamed itself Fin. A month later, Salesforce signed. A rename sounds cosmetic, but it signals to customers, employees, and buyers that the company is no longer a support software company that also has an AI agent. It is the AI agent company. That clarity changes how the team prioritizes, how sales explains the product, and how an acquirer values the business.

Think about the alternative. A mature company with a new AI feature will usually add it to the existing product line and mention it on the pricing page. Resources stay divided between the old product and the new one, and neither gets the full attention it needs. Committing the whole company to one product ends that tug of war.

• One product gets the best people. Splitting talent across a dozen initiatives produces a dozen average results.
• One story for the market. Customers and buyers understand a company that is clearly about one thing.
• One metric to track. When everything points at the same product, progress is easy to measure.

The three moves also depend on each other. Moving fast gave the team a product worth betting on. Pricing on results gave that product a business model that customers could say yes to without a long approval process. Betting the whole company on it gave the first two moves the time, people, and attention to compound. Take any one of them away and the story looks much weaker, which is why copying only the pricing or only the rename will not produce the same result.

What Salesforce was really buying

So why would Salesforce want it? Fin serves more than 30,000 businesses across chat, email, WhatsApp, SMS, and phone. Agentforce, Salesforce's own agent product, reached $1.2 billion in ARR earlier this year, up 205% from the year before. Fin gives Salesforce a support agent it can sell to companies of every size, with a customer base and a result-based pricing model already in place.

The buyer paid for a proven product, a set of customers who already trust it, and a company that had reorganized itself around the one thing that was working. It did not pay for the mature support software that was close to zero net new ARR in 2022. That is the quiet lesson inside the headline number: the value came from the new product and the decision to commit to it, and the old business was the platform the turnaround stood on.

What to do this week

Here is my takeaway from this story. If growth stalls, do not spread a little AI across every product. Pick one product, price it on the result your customer wants, and bet the company on it. That is the pattern behind the $3.6 billion, and it is available to founders whose companies are far smaller than this one.

Start with a short exercise this week. Write down the one result your customers would happily pay for every time it happens. It might be a resolved ticket, a booked meeting, a qualified lead, or a finished invoice. Once you have that sentence, ask whether your current pricing and your current roadmap point at it. If they do not, you have found the first thing to change.

If you run a SaaS company at $1 million or more in ARR and want to work through decisions like this with other CEOs, our growth mastermind meets weekly at sasrise.com, and our AI founders and builders community lives at lounge.ai.