Salesforce adjusts Data Cloud pricing to entice customers to get their data in shape for Agentforce

One of the barriers to productive use of AI in the enterprise is the difficulty of bringing data together from various sources. Helping customers get their data in order is therefore a priority for vendors who are eager to sell their AI agent offerings. This is the motivition behind Salesforce's shake-up of pricing for Data Cloud, which it unveiled on Friday. I spoke to Craig Shull, Chief Pricing Officer at Salesforce, to find out more.

Unlike the seat-based model that applies to Salesforce's traditional enterprise applications, Data Cloud has always been priced on consumption. This means that any existing Salesforce customer that already subscribes for one of the core Salesforce clouds can simply switch on Data Cloud, and they will start with a set quantity of free credits that get used up once they move data in and start working with it. They can then purchase additional credits to fund their ongoing usage as it expands.

Adoption barriers

But the original pricing model turned out to include some barriers to adoption. First of all, Salesforce was charging credits for ingesting data from its own applications into Data Cloud. Customers felt this was making them pay a second time for using their own data, so that charge has been abolished. Shull says that this puts the emphasis on Data Cloud usage that actually delivers some kind of value to the customer. He explains:

If you just move your data from Salesforce into Data Cloud and don't do anything with it, there's been little value delivered. So why should we charge for that? We're going to make that free. The same thing on zero copy. It's the actual utilization of the data in a business process that adds value to you.

The second change simplifies the different price components. For example, there's no longer any distinction between production and sandbox credits, which means that customers can now allocate credits at the time of use, rather than having to work out in advance how many credits they'll need for development and testing versus production. There's a 20% discount applied when credits are used in the sandbox as a further incentive to build on the platform. Shull says:

We effectively consolidated eight different usage types into three to make it much easier for customers to estimate and track, and in general, just take some of the friction out of the buying cycle.

In pricing lingo, this means that consumption credits are now 'fungible' across all the various different types of usage, with the number of credits spent varying according to each type, ranging from data transforms, analytics and queries to more expensive sub-second processing. Shull explains:

The value could be much different based upon what the business process that you're trying to automate. Talking about the marketing use case where you're using sub-second, clearly that is a really valuable thing — you go to the website and there's a pop-up for some offer, that is significant value, if you can get the conversion to occur with that person.

The other two pricing components are a monthly fee for data storage and various premium add-ons for extras such as segregated data spaces within a single enterprise instance, private data connections, real-time user profiles, and segmentation by advertising platform.

Pricing calculator

For customers that are still having trouble figuring out how much they might spend, Salesforce has also introduced a new pricing calculator to help explore the impact of different configurations and usage patterns. The vendor has also improved the usage analytics provided in its Digital Wallet, where customers can create reports and alerts to track usage, down to the relationship between individual features and the usage they're driving.

The aim here is to help enterprise buyers — who are used to the predictability of seat-based pricing — get more comfortable with the kind of consumption-based pricing models that are becoming prevalent among AI-native platforms. This is an important hurdle to get over, as Shull explains:

The biggest challenge that we as an industry have on consumption-based [pricing] is predictability, and giving the customers the visibility and the transparency as to what their spend could be over a certain period of time. Part of that is telemetry... to allow customers to know specifically what's happening and how to ask.

And then, we've invested a lot of time in our estimators. We have some public-facing estimators, and, as I mentioned, our Digital Wallet, which can give people exactly what they're using, predicted usage, etc. So it's really giving the transparency to customers so they can get comfort in sizing what their use cases will be.

The changes to Data Cloud pricing follow other changes that Salesforce has introduced to help customers adapt to agentic AI. When Agentforce launched at last year's Dreamforce, customers were able to opt to pay per conversation, where there is a flat rate charge when an agent interacts with a customer on an issue over a 24-hour period. This was followed by the introduction of Flex Credits in May this year, where customers can use credits to pay for a wider range of agent actions that might not fit into the conversation model. Schull comments:

Again, we're giving customers flexibility. They want to buy conversation because it ties to value, that's great. If they want to just pay per action and use a fungible Flex Credit to do that, then they can do that as well. For some of these edge cases, the tie between value and consumption is much tighter.

From seats to consumption

Introduced at the same time, a new Flex Agreement gives customers the option of converting seats that they've paid for on a per-seat subscription into Flex Credits. He explains:

Salesforce Flex Agreement allows you to actually move spend from seats to consumption product. So if you have bought 1,000 seats and you only need 900, you can turn in those 100 seats and move that to consumption...

The fundamental premise of the Flex agreement is, if your seats are going down, we're going to allow you to move that to consumption-based for automations and agents and those things that you're using that provides you value in a different way. The key is the flexibility that we're offering customers there.

He sums up:

We offer the flexibility of buying per user and having all your employees use your per-user licenses and interact with the agent, and you don't pay for the agent, you just pay for the user. We offer the option of having conversation-based [pricing] for the self-service [use case], and then we also offer action-based pricing for companies that want to build their own agents or have highly tailored agents that they want to pay on a pure consumption basis. So we're giving the flexibility across the portfolio.

My take

Lack of data readiness is increasingly cited as a barrier for enterprises moving forward with AI and agentic automation, so no wonder Salesforce wants to optimize pricing for Data Cloud. It makes sense to entice customers to start using the platform as a foundation for their Agentforce projects. And with a new set of Agentforce goodies no doubt lined up to unveil at the annual Dreamforce conference in just a few weeks' time, the vendor's sales teams don't want to get bogged down in awkward pricing discussions on the show floor. It's hard enough as it is to wean enterprise buyers off the familiar certainties of seat-based pricing without making new consumption-based models even harder to adopt than they need to be.