Imagine, for the length of this paragraph, a person who holds a job that three years ago needed a hundred colleagues. She works in the sales operations department of a large software company. Her days go to moving information around: which customer is due a renewal, which quote is stuck, which account manager has not been told. Most of it travels by email. On a Wednesday evening in late October she is not on the earnings call, and nobody on the call knows she exists, because she is an illustration. But the executives speaking that evening describe her department in the past tense.

The call was ServiceNow’s, held on 28 October. The company, which sells workflow software to large organisations, reported a third quarter that beat expectations. Revenue rose more than 20% to $4.106 billion against the $4.095 billion analysts expected, and adjusted operating margin came in at 32.5% against a forecast of 31.2%, according to Calcalist’s Ctech. Then an analyst from Bernstein asked about falling headcount, and the answer became the story.

What the executives said

Peter Weed of Bernstein noted that productivity gains seemed to be showing up in the headcount numbers, mostly in the front office, product, and sales and marketing. Gina Mastantuono, the president and chief financial officer, replied that ServiceNow had “crossed a billion dollars in efficiency gains” through its own internal use of AI, most of it from running its own products on its own platform. Ctech renders the figure as more than $1 billion in annualised efficiency gains. “That efficiency gains means that we’re able to not hire as many people,” she said, according to the transcript.

She was careful about where. Quota-carrying sales staff are up by double digits, she said, while sales operations is “down significantly”. Research and development shows the same pattern: more AI engineers, fewer support and overhead roles.

Chairman and chief executive Bill McDermott added the sentence that gave the coverage its headline. “Our ultimate headcount has actually slightly declined, which is a gorgeous situation,” he said. He put the outcome as “less people doing extraordinary things, enjoying their work much more”, in Ctech’s quotation. He also supplied the picture behind the sales-operations numbers: three years ago, he said, that function was “like 100 people” moving information around on email, and it is now “completely autonomous”.

Ctech observes that ServiceNow is not the first company to link AI to a smaller payroll, but that few large software companies say so this directly, on the record, as a selling point. The reason is plain in its own account: ServiceNow sells the same technology to other companies, so what it says about its own staff is part of the pitch.

A line that had been drawn earlier

The call did not introduce the idea. In late July Ctech reported that ServiceNow had begun a round of layoffs affecting hundreds of employees, including some in Israel, without giving a number. The company told the outlet that it was “driving efficiencies across the business, actively investing in and hiring for AI-focused skills, and managing headcount with discipline to end 2026 where we started, as we shared earlier this year.”

Read against that statement, the October remarks are a progress report. The company said in July that it would finish the year with about the headcount it started with; in October its chief executive described the figure as slightly down and the workforce as steady, while the mix moved toward salespeople who carry quotas and engineers who build AI. McDermott’s own gloss was that “while the headcount steady, the innovation and the improvement of the human capital condition has only increased substantially”, something that “sometimes doesn’t show up in the numbers”.

The market looked at something else

Investors did not reward the candour. On Thursday ServiceNow shares fell roughly 25%, Ctech reports, having been down about 1% for the year before the results. Ctech puts the cause in the forecast. The company guided to fourth-quarter growth in current remaining performance obligations, the measure of contracted revenue due within a year, of 18.5% in constant currency. Wall Street had expected 19.8%. The third-quarter figure itself, up 21% to $13.65 billion, was ahead of consensus.

Mastantuono said the gap was mostly accounting and timing, not demand. Early renewals made up about half of the third-quarter beat, leaving less for the fourth. Strong public-sector demand shifted the mix toward software installed on customers’ own systems, which costs the metric about a quarter of a percentage point. And the company is now booking some bundled AI delivery work as professional services instead of subscription revenue, a shift of about 25 basis points in the third quarter that she expects to reach about 50 in the fourth. “cRPO is a very nuanced accounting metric,” she said, “and it doesn’t necessarily capture all of the momentum in the business.”

The sharper question came from Morgan Stanley, whose analyst noted that ServiceNow had raised its AI contract-value target twice, to $1.7 billion for the year from $1 billion in January and $1.5 billion in May, while the broader backlog looked stable at best. When, the analyst asked, would AI momentum turn into acceleration? McDermott answered that the business “has never been in better shape than it is right now”, and that customers would see “the true acceleration” in 2027. The company also said agentic AI consumption in production was up more than 13 times from a year earlier.

Set side by side, the two halves of the evening make an odd pair. Inside the company, AI is credited with more than $1 billion of efficiency; in the company’s reported numbers, the AI business is a $1.7 billion target that has not yet moved the measure investors watch. The efficiency figure is ServiceNow’s own estimate, stated on a call, and the sources do not say how it is calculated. Neither source connects the share price to the headcount remarks, and the 25% drop is explained, by Ctech, through the guidance.

What the call leaves open

What the call does establish is narrow. A large software vendor said, to analysts and on the record, that AI lets it avoid hiring operational roles while it adds quota-carrying sellers and AI engineers. It did not say how many jobs that is, and the July layoffs came without a figure either. Which roles go, which stay and how quickly are described by function, not by count.

The vendor’s customers will draw their own lesson. A buyer weighing ServiceNow’s agents is told that the seller uses them and has fewer people in sales operations as a result. A buyer’s finance team may also notice what the market noticed: the proof is a cost line, and the revenue line that should follow is promised for 2027.

Back on the invented Wednesday, the woman from sales operations does not hear any of this. She finds out the way people in her position usually do, from a calendar invitation to a meeting about how her team is “structured”. Whether it is a good day or a bad one depends on a question the call did not answer: whether a department that shrinks because software does its work is a gorgeous situation for the people who used to be in it.

References

  • Calcalist (Ctech), Sinai, A. (2026). “A gorgeous situation”: ServiceNow’s CEO celebrates a shrinking workforce. calcalistech.com
  • Calcalist (Ctech) (2026). ServiceNow continues layoffs as it restructures around AI. calcalistech.com
  • Platform Aeronaut (2026). NOW Q3 2026 Earnings Call Transcript (ServiceNow third-quarter 2026 earnings conference call, 28 October). transcripts.platformaeronaut.com