Salesforce Agrees to Buy Cresta. It Wants Every Place Customers Talk

Salesforce signed a definitive agreement to acquire Cresta on November 2. Cresta is a voice-first customer-experience company whose voice agents and real-time call guidance run at Marriott, Comcast and Alaska Airlines. Salesforce disclosed no terms. The Information reported that the deal is worth $2.7 billion. Cresta will be combined with Service Cloud. Salesforce expects the deal to close in its fiscal Q1 2028, which runs February to April 2027, subject to regulatory approval.

Cresta is Salesforce's third service or engagement deal since mid-June. Fin was signed June 15 and closed in September for about $3.6 billion. Braze followed on October 20 at an enterprise value of $5.66 billion (CMSWire counts five CX and marketing deals this year, including the Listen Labs agreement in September). Salesforce already sells Agentforce Voice, built on the same tools as its other agents, and its release on closing Fin lists voice among the channels Fin's agent handles. With Cresta, Salesforce would own three products that can answer a customer's call. Cresta adds voice agents built for complex calls and real-time guidance for human reps. (We covered its Deputy Agent, which keeps working cases after the call ends, in October.)

Salesforce already prices voice beside Service Cloud, and Cresta gives it more to sell

Salesforce hasn't announced how it will price Cresta. We're inferring from how it sells voice today, and it already has a product line to build on. Salesforce launched Agentforce Contact Center in March, with native voice, and lists it at $125 per user per month as an add-on for Service Cloud customers. Contact Center Plus, at $250, already folds in workforce and quality management, the part of the suite closest to Cresta's coaching. No Jitter's launch review said the product lacked the real-time assistance for human agents that mature suites offer, which is what Cresta sells. Cresta's CEO says the plan is to embed its platform "directly into Salesforce's first-class customer service Platform." At contact centers that run Service Cloud, Salesforce already holds the contract and knows the renewal date.

The cheapest way to earn back $2.7 billion (if that's the price) is to quote voice agents next to the seats and agent fees a customer already pays, at a renewal Salesforce already controls. It has also said it will meter the calls outside agents make into the CRM. If voice lands on a Service Cloud renewal, a standalone voice vendor is competing with a renewal discount, and Cresta's product barely enters the buyer's decision.

The deal hasn't closed, and Salesforce's release commits Cresta to keep supporting the contact-center and telephony platforms its customers rely on today. That promise matters more now that Salesforce sells native voice against those platforms, though CMSWire notes it doesn't sell a telephony stack of its own. CMSWire says buyers on them will want the commitment in writing, with a timeline. COMMfusion analyst Blair Pleasant wrote in a LinkedIn post, quoted by CMSWire, that the deal "isn't necessarily about forcing customers into a single ecosystem." Cresta also pitches itself as running on existing systems. It plugs into the telephony, CRM and knowledge tools its customers already use and writes call summaries back to the CRM.

Fin is the best evidence against the bundling read so far. Nearly two months after the deal closed, its pricing page still charges $0.99 per outcome and says it works with any help desk. That shows a standalone price can survive a Salesforce purchase, at least for now. It doesn't say what Cresta's voice agents will cost inside a Service Cloud renewal, and price is where bundling would show up. Salesforce disclosed Fin's price and Braze's enterprise value, while the $2.7 billion for Cresta is only reported.

Salesforce is buying every place a customer talks to a company

Line up the deals and the plan is easy to see. Contentful runs the pages a customer reads, and Braze sends the company's messages. Fin answers the chat. Cresta would take the call, and Listen Labs asks customers afterward what they thought. CMSWire puts the 2026 total above $15 billion if Cresta is $2.7 billion. Spending over $15 billion is a lot for a company that already owns the customer record, and we think that's the point.

The record is getting cheaper to own. If agents do the work and can read a CRM from anywhere, the value moves to wherever the customer and the agent actually talk, and to whoever sees what happened next. No Jitter argued something similar when Salesforce launched its contact center: housing customer data won't keep earning a premium, so Salesforce wants the conversation too. Charging outside agents to read CRM data is the other half of that plan. Salesforce charges for access to the record and buys the interactions that feed it.

What this means for CX leaders

It's a smart plan with an obvious weak spot. Salesforce hasn't said how three products that can answer a call fit together, or which one a Service Cloud customer should buy, and Fin still sells to companies on other help desks. Buying is the fast part. Until Salesforce turns these into one product, it owns a stack of overlapping vendors, and its customers have more room to negotiate than they will later. If you run Cresta today, get the telephony commitment in writing with a timeline, as CMSWire suggests. Lock voice-agent prices into the contract before the deal closes.

Until Salesforce turns these into one product, it owns a stack of overlapping vendors, and its customers have more room to negotiate than they will later.

Before the deal closes between February and April 2027, get export rights and a first-renewal price cap into your voice-agent contract, and ask your Salesforce account executive to quote voice as its own line.

By Business of GTM