Salesforce and Evidence-Based Marketing: When AI Matters Only If Charged per Work Unit
At times, the market seems to possess a singular kind of patience: that of the next quarter. Salesforce experienced this firsthand. The company reported a solid fiscal fourth quarter, with revenues of $11.2 billion, up 12% year-over-year, and showcased an impressive profitability punch: non-GAAP earnings per share of $3.81, far surpassing the $3.05 that analysts were expecting. Yet, the stock fell 1.2% following the announcement.
As a researcher in innovation and consumer behavior, I find the underlying mechanism more intriguing than the superficial paradox. In enterprise software companies, a stock's behavior acts as a forward vote on the quality of the “contracted future,” not on the past performance. This time, the vote was largely driven by guidance: Salesforce projected $45.8 to $46.2 billion in revenue for fiscal 2027, equivalent to 10% to 11% growth, with the first quarter in the range of $11.03 to $11.08 billion.
The hard data coexists with another figure that's more revealing for marketing and strategy: Agentforce, its AI agent platform, reached $800 million in annual recurring revenue (ARR). Along with Data 360, it exceeded $2.9 billion in ARR, +200% year-over-year. The critical story is how Salesforce is attempting to convert that traction into a more demanding purchasing standard: moving from “AI as narrative” to “AI as budget line.”
The Stock Falls Despite a Strong Quarter: The Market Buys Guidance, Not Headlines
That a company can “beat” expectations yet still decline is not a mystery: it's a reminder that the measuring stick is constantly shifting. Salesforce not only presented growth; it presented a 12% acceleration that, according to their briefing summary, was the first in five quarters. It also showcased future visibility with current remaining performance obligations of $35.1 billion, +16%, with Informatica contributing 4% to that growth.
However, the market doesn't punish the quarter; it punishes the entry angle into the next one. The annual guidance, described in reports as “below estimates,” ended up weighing more than the profitability beat. This typically occurs when investors interpret that there is friction in converting a tech wave (AI) into predictable revenues. In simple terms: the quarter confirms that Salesforce knows how to sell and operate; the guidance raises doubts about how quickly it can scale the new engine without diluting margins or running into slower buying cycles.
Here marketing plays an uncomfortable role: when the product is commercial infrastructure, corporate buyers do not “fall in love” with AI. They tolerate it if it reduces risk, accelerates results, and fits into a defensible budget against a CFO. In this context, the guidance becomes a signal about something more: how much of the demand for AI is repeatable, how much is pilot-based, and how much expansion comes from existing customers.
Salesforce revealed a key piece of information: more than 60% of Agentforce and Data 360 bookings in the quarter came from existing customers. This is good news in terms of adoption, but also a warning: growth heavily depends on upselling to the installed base. The strategic internal question (without needing to pose it to the reader) is whether the promise of AI agents holds up equally well when trying to win new accounts, where switching costs and comparative evaluation are tougher.
Agentforce as a Marketing Product: Sell Productivity, Not “Intelligence”
The most interesting aspect is not that Salesforce has an agent offering, but how it attempts to package AI adoption to make it purchasable. In the quarter, Salesforce closed 29,000 Agentforce deals, +50% sequentially, and reported strong growth in the unit where much of that proposition resides: Agentforce 360 Platform, Slack, and others, +37%.
The strategic novelty is the introduction of a metric aimed at disciplining the conversation: Agentic Work Units (AWUs), units for measuring tasks performed by agents. CEO Marc Benioff stated that they have consumed “almost 20 trillion tokens” and turned them into “over 2.4 billion agentic work units” to date. Regardless of the technical discussion, from a marketing standpoint this is a clear repositioning: the unit of value is no longer “tokens,” “models,” or “capacity,” but work done.
This choice aligns with how companies buy. A sales director does not wake up eager to purchase AI; they wake up with a pipeline to fill, lengthening response times, and saturated teams. The most direct way to convert AI into budget is to promise a reduction in operational friction with a metric resembling productivity.
It’s also a defensive decision against a common temptation in the corporate market: “I’ll build it myself.” Benioff made this explicit during the earnings call by comparing “roll my own AI” against “plugging it in” within the existing product. The battle is not just technological; it’s about total cost of ownership, implementation time, and accountability. Salesforce wants the customer to “hire” a quick fix to complexity, not a lab.
In behavioral terms, a pattern emerges: when an innovation shifts from curiosity to serious purchase phase, buyers begin to demand three things: evidence, control, and accountability. AWUs attempt to provide all three.
The Real Risk: Over-attending to Large Clients and Opening the Door to Simple Alternatives
Salesforce also reported that transactions over $1 million grew by 26% in the quarter. This is a strong signal: the company is capturing expansion in large accounts, where the pain for efficiency and automation tends to translate more quickly into money.
However, this success brings a structural risk I’ve seen repeatedly in leading corporations: the obsession with the most profitable customers often drives the creation of layers upon layers of product, controls, integrations, and offerings “for everyone.” The result: increasingly comprehensive solutions that are also more challenging to buy, implement, and govern for the mid-market.
Benioff’s own narrative about how “agents cannot work in isolation” and that they “need to call home” reflects a platform bet: each agent becomes more useful the more tied it is to data, applications, and internal flows. That increases value… and also raises the exit cost. In marketing, this can be sold as continuity and security; in procurement, it reads as dependency.
This is where space opens for others to compete not with “better AI,” but with “less friction”: simpler products, with a narrow use case, shorter deployment times, and a price that’s easier to justify. When the leader chases seven-figure deals, the natural opportunity lies with those who package a concrete advance for the rest.
Salesforce is attempting to mitigate this tension with two moves that serve as signals to the market: (1) financial discipline, and (2) a bridge between adoption and return. On the financial side, it announced a $50 billion stock buyback authorization and a 5.8% increase in the quarterly dividend to $0.44 per share, in addition to having returned more than $14 billion in free cash flow to shareholders by the close of fiscal year 2026. It’s a message: “we can invest in AI and still maintain returns.”
On the return side, AWUs and the emphasis on “work done” aim to prevent AI from being viewed as another layer of cost. The challenge is that the corporate buyer does not pay for metrics; they pay for results that withstand internal auditing. And there, the conversation is no longer about external marketing, but enabling: use cases, governance, and evidence of impact on critical processes.
The New Golden Rule of B2B Marketing: Guide Purchases Toward Operational Metrics
The briefing mentions a linguistic data point that, when read carefully, is a clue to commercial maturity: an analysis of the earnings call observed that the “Delta Score” fell to a historic low, suggesting a shift towards more sober, performance-based communications. For a company pushing a platform change, this is significant. In AI, over-narration comes at a steep price when guidance doesn’t accompany it.
My reading is that Salesforce is attempting to solve an adoption problem that is not technical, but about purchase: to turn AI into a product that the customer can defend internally without relying on faith. In enterprise software, the real “user” is not just who uses the tool; it’s the committee that approves the expenditure, the area that takes on the risk, and the team that implements the change.
This is why the center of gravity for marketing shifts. It is no longer the one who promises “intelligence” that wins; instead, it is the one who converts that promise into an allocation mechanism: how much it costs, what work it replaces or accelerates, how it is measured, and what happens when it scales. By introducing AWUs and publishing strong financial signals, Salesforce is attempting to shift the conversation from “innovation” to “operational capability.”
The market, with its reaction to the guidance, made the standard clear: the AI narrative is validated not by a brilliant quarter, but by revenue repeatability and clarity in the pace of expansion.
The success of Agentforce indicates that the work corporate clients are contracting is not “having AI,” but converting complexity into measurable productivity within a defensible budget.
