Between September 30 and October 7, HubSpot and SAP each said the measure of their AI products is business results, and HubSpot reorganized its product teams around that claim.
The pitch moved from capability to result
In an essay dated September 30, HubSpot CEO Yamini Rangan set out what she called the Outcomes Era. Her description of the old model: “Software delivered capabilities. Humans delivered the outcomes.” Her description of the new one: “Software can now help deliver outcomes, not just give people the capabilities to achieve them.” The outcomes she lists are the ones revenue teams already track, which are building demand, winning deals and delighting customers. Rangan wrote that the focus is shifting “from what AI can do to what it can actually deliver for scaling businesses,” and the October 6 message says the shift is changing HubSpot’s product, pricing and how it serves customers.
A week later the language reached the org chart. In a message to employees on October 6, published in HubSpot’s newsroom, Rangan said HubSpot would reduce its team by about 7%, nearly 660 people, and organize product teams around customer outcomes instead of Hubs, with each team owning the full customer journey.
Advertisement
SAP made a related argument from a different starting point. At SAP Connect on October 7, CEO Christian Klein reported that SAP has created more than 20 Joule Assistants and over 200 Joule Agents since May, and then said: “But the real story isn’t the number of agents. It’s the business outcomes they deliver.” The keynote recap shows SAP also letting customers choose the level of autonomy for each process, with human approval kept where risk is high.
Context is the mechanism the vendors name
If the product delivers results, something has to explain why the results vary. HubSpot’s answer is context, which it defines as “the dynamic knowledge about your business, your customers, and your team.” Rangan gives a sales example. Any AI can help write a sales email, she wrote, but an AI with context knows that a prospect visited the pricing page three days ago, that the deal has been stalled for two weeks and that the last demo did not cover integrations. The essay says HubSpot looked at data from its customer base of more than 300,000 companies and found that the same AI models produced very different results depending on the context they received.
With good context, HubSpot reports, MQLs generated improved 264%, deals closed won improved 197% and customer meetings booked rose 200%. With bad context, MQLs declined 28%, deals closed won declined 27% and meetings booked declined 49%, which the essay describes as worse than using no AI at all. It lists three ingredients of good context: business context, customer context and team context, the last covering workflows, handoffs, roles, approval rules and escalation triggers.
SAP’s version appears in Klein’s keynote. He said SAP’s more than 400 million users can talk to SAP in their own language through Joule Work, which he described as a new interface for AI and is rolling out to customers this month. Near the close he said, “AI also requires significant change management to ensure it delivers the desired outcomes for your business.” He named tools for measuring process value before and after a change and for supporting user adoption. Salesforce has pushed the same direction on the service side, as we noted when customers described agents turning cost centers into revenue engines.
What the shared framing leaves open
Two vendors using one frame does not make their results comparable. The percentages in HubSpot’s essay are relative changes, and the essay gives no absolute counts. An earlier piece of ours found that revenue platforms all claim unification while publishing little that would let a buyer check it. Outcome claims face the same test.
This is our reading, and the vendors have not said it: an outcome claim is only as checkable as its baseline, its sample and its definition of success. Those three items are what a buyer can ask for.
What it means for the sales leader
The shift changes three conversations a revenue leader has with a vendor.
Contracts. A product sold on results invites contract language about results. Seat counts and feature tiers describe what you bought. A result needs a definition, a measurement window and a named data source, written into the order form before renewal rather than argued over after it.
Ownership. HubSpot describes results that depend on data the customer holds, and Klein says AI needs change management on the customer side. HubSpot’s three ingredients include workflows, approval rules and escalation triggers, which only the customer’s own team can supply. Someone on the buyer side has to own them, and in most revenue organizations that someone sits in RevOps.
Measurement. If a vendor reports that outcomes improved, the useful follow-up is the comparison group. Ask what the improvement is measured against, over what period and for which customer segment.
How to evaluate an outcome pitch
Four steps work for any vendor making this argument, including the two named here.
- Ask for the baseline behind every percentage, and whether it is a before-and-after comparison or a comparison between customer groups.
- Ask how the vendor defines good context, and whether you can measure your own account against that definition today.
- Write the result you expect, the window and the data source into the contract, and name a person on your side who owns the inputs.
- Pick one process and measure it for a quarter before expanding. SAP’s own framing of choosing autonomy per process supports starting narrow.
