If you work in the Salesforce ecosystem, you have likely heard of Agentforce – Salesforce’s answer to the rising demand for efficiency-boosting and cost-saving AI solutions. In case you missed the Dreamforce announcement, Agentforce is the new incarnation of Einstein Copilot, but to simply call it a re-labeling would be a gross oversimplification.
Agentforce brings a crucial piece that was previously missing in the Einstein AI offering: agency. In practice, Agentforce allows you to build autonomous AI agents that can not only converse and respond but also to act upon inputs they receive. There is no doubt that the potential is huge, but this is not the only change Agentforce brings with it.
Unit Costs: The Art of Volume and Scale
There’s another power in play that promises to change how companies view Salesforce as an investment. Thus far, Salesforce has relied on a licensing model where customers pay for user licenses from the moment of provisioning – i.e. when the platform is made available for use. In this model, you incur costs even before configuring and adopting the tool in practice.
The business value, on the other hand, often comes with a significant delay. Companies must prepare solid business cases and forecasts in advance to justify such investments.
With Agentforce, the pricing model is different. Salesforce has revealed that Agentforce’s pricing will be conversation-based rather than following a traditional licensing model, or even a consumption-based model used by Data Cloud and Marketing Cloud. Salesforce has announced an initial $2 fee per agent conversation – with possible volume discounts where applicable.
This means that every time an employee, partner or end-customer chats with an agent, a fee will be added to your next invoice. While this may seem like a lot considering the potential volume of these discussions, the idea is to reduce or even eliminate human intervention in the same process. You essentially replace an employee’s manual work with the agent doing the same autonomously.
Agentforce taps into unit costs instead of indirect costs. Conversations have much closer relation with a company’s value creation processes (e.g. selling and providing customer support) than some background process or visualization – typical benefits that CRM solutions offer.
With Agentforce, cost is incurred in direct relation to activities that bring future revenue to the company. This means that costs scale as the conversation volume grows, which in turn should have a positive effect on the bottom line. The assumption here is, of course, that the autonomous AI agents have to be placed in processes where their added revenue exceeds the $2 conversation cost.
Three Key Business Impacts of Agentforce Pricing
While conversation-based pricing may not be as thought-provoking as “agentification” in general, it does matter. Agentforce’s pricing model is different from the standard licensing model, which means that there are real and positive business impacts, such as:
- Lower barrier to entry
- Clarity of business case
- Linear cost forecasting
One of the key impacts is that costs are generated alongside usage – not beforehand like with licensing.
Let’s say that you want to launch a proof of concept for Agentforce. With the conversation-based pricing, you don’t have to invest heavily in licensing to do a small-scale test, but can instead launch a single agent and test it in a controlled environment before committing more resources. This should translate to less financial risk and lower entry barriers for Salesforce customers.
With the conversation-based pricing, the business case is clearer. As a point of comparison, a standard CRM adoption, where business benefits are indirect and KPIs, are often difficult to set. With Agentforce, the situation is different.
You have direct lines to revenue generation; whether it is sales development representative work, win-back support, or even upselling/cross-selling to customers. Agent conversations have very tangible business outcomes and can be mapped directly to revenue generation or cost savings.
Cost forecasting is arguably easier with this conversation-based model. At this point, you may be rolling your eyes and asking “why?”. Sure, with license fees, you pay a fixed fee in advance for the whole year. However, understanding what licenses to add and when is a huge unknown, and must be estimated roughly. With autonomous agents, the cost progression should align linearly with the revenue generated.
Forecasting with agents is more akin to sales or churn forecasting rather than calculating ROI, which is indirect and full of assumptions.
Conversations vs. Consumption Credits
Reading this post, you may be wondering if this thing about $2 conversations is anything new, and you’d be partially correct.
Marketing Cloud Engagement users should be familiar with consumption-based pricing by now, and Data Cloud has used a credit-based system since 2023. Consumption-based pricing does bear similarities with Agentforce’s pricing model. For one, both scale with usage volume as your business grows. However, consumption relies on pre-bought credits, whereas conversations are marked for invoicing as they happen.
Another key difference is the user group. Data Cloud and Marketing Cloud credits are somewhat easier to monitor and control as they are used by internal users and processes. Agents, on the other hand, have conversations with end-users, both internal and external.
You cannot really control the volume of these conversations in advance. You instead want to position the agents in business processes where every conversation saves money from otherwise manual tasks or adds a possibility to sell more or increase efficiency. If this is done well, you should not have to worry about conversation volume increasing. After all, the benefits should outweigh the unit cost.
Finally, consumption-based solutions like Data Cloud and Marketing Cloud can be difficult to monitor and forecast. Credits are typically counted on several categories, e.g. data services and data storage.
Salesforce has done a great service to users by introducing a digital wallet. However, the digital wallet only tells you how many credits you’ve already consumed – it doesn’t give you a burn rate or runway for remaining credits (yet!).
Conversations, on the other hand, have no upper or lower limit. You don’t have to worry about exceeding your allotted credits, but you have less control over the volume. But then, that’s the name of the game when you’re setting up autonomous agents. They act autonomously and accrue costs autonomously too.
Open Questions
As Agentforce is a completely new product (and now generally available), there are several questions in the air. For one, an inside source within Salesforce has hinted that in addition to the $2 per conversation, there may be a special license type included for agents.
In this model, each agent would consume a license, similar to how it is with integration user licenses. Another question is whether there will be features to monitor and forecast conversation volume. Otherwise, customers would have to set up monitoring by themselves.
Final Thoughts
Salesforce’s Agentforce brings a fresh approach with conversation-based pricing, so costs now grow with usage and revenue-generating actions.
This shift makes ROI easier to see and simplifies cost forecasting, simplifying the process for companies to get started and scale smoothly as they deploy more agents.
The post Salesforce’s Bold New Pricing Strategy: What You Need to Know appeared first on Salesforce Ben.