A customer signs for a new finance system. Before anyone can rely on it, somebody has to connect the payroll data, decide who can approve a payment and check that the reports agree with the old system. The subscription is one commercial unit. Making it work is another.
For a software business, that distinction reaches beyond the implementation invoice. Deployment can help a customer stay and expand. It can also absorb staff time, require subcontractors and leave the supplier with a services bill larger than the revenue attached to it.
The useful question is who performs that work, who pays for it and what the accounts reveal about the connection to recurring software income. A small professional-services business does not prove a product is easy to install. A loss in that business does not, on its own, prove the wider customer relationship is uneconomic.
Fact: three services lines, three reporting boundaries
Salesforce (NYSE: CRM) reported $525 million of professional services and other revenue for the quarter ended 31 July 2026, against $628 million of corresponding costs. Revenue less those costs gives a $103 million gross loss. The category includes advisory work, process mapping, project management, implementation and training; it is broader than installation alone. Management attributed the revenue decline to weaker demand for larger, multi-year transformation engagements. Those definitions and explanations appear in its July 2026 quarterly filing.
ServiceNow (NYSE: NOW) reported $118 million of professional services and other revenue and $157 million of related costs for the quarter ended 30 September 2026. That is a $39 million GAAP gross loss, alongside $3.988 billion of subscription revenue. Its 28 October results therefore show subscription growth and a loss in the services category at the same time. Neither observation cancels the other.
Workday (Nasdaq: WDAY) reported $178 million of professional-services revenue and $216 million of related costs for the quarter ended 31 July 2026, a calculated $38 million gross loss. These are the reported expense categories, including share-based compensation, rather than adjusted cash costs. The figures come from its August quarterly results.
ServiceNow · Q3 2026 · GAAP
The services delivery bill
Revenue
Professional services and other
Cost
Same category and quarter
The periods differ, and the services categories are not identical. These figures establish that delivery work can have a different economic profile from a subscription. They are not a league table of implementation efficiency, nor estimates of the total cost borne by each company’s customers.
Interpretation: follow the work beyond the vendor
Consider a hypothetical customer that buys software directly from its developer but pays an independent implementation partner to configure it. The software supplier records its own contractual revenue. The partner’s invoice belongs to a separate business. The customer still faces both bills.
Moving work to a partner can consequently reduce a vendor’s services revenue without removing the underlying work. It may also free the vendor’s staff for specialist projects. Alternatively, falling services revenue can reflect fewer projects or weaker demand. The direction of the line cannot distinguish these explanations by itself.
Workday’s fiscal 2026 annual report provides a concrete example of the boundary: both Workday and its services partners help customers deploy its products. Its explanation of the annual services-revenue decline includes project size and mix as it expands its use of partners. That annual explanation should not be silently carried forward as the cause of every subsequent quarterly movement.
There is a second boundary inside the customer relationship. A supplier might accept a thin services margin because implementation supports a longer subscription relationship. That is an economic hypothesis to test over time. It requires evidence that deployment leads to durable paid use, and that the later contribution covers the resources committed along the way.
Gross loss is also different from cash lost during the quarter. Revenue recognition, payment timing and non-cash compensation affect the comparison. Equally, a positive services gross margin would still leave sales, development and central expenses to fund. The services calculation answers a limited question.
Who could benefit?
Interpretation: Salesforce, ServiceNow and Workday could benefit if effective implementation turns contracted software into workflows that customers continue to use and pay for. The potential gain sits in the wider relationship over subsequent renewal periods, rather than necessarily in the initial services engagement.
The links are documented, but the payoff remains conditional. Salesforce says its services organisation helps adoption and larger subscription contracts. Workday describes its deployment partners. ServiceNow’s 2025 annual report identifies implementation difficulties as a risk to renewals and additional sales. Together these disclosures support examining deployment as part of commercial execution; they do not quantify a return on each services dollar.
Materiality depends on the effect on the much larger subscription relationship. A cheaper installation that produces poor adoption could save the wrong cost. Useful evidence would connect delivery resources, successful deployment and comparable renewal outcomes, rather than simply celebrate a smaller services team.
Who’s at risk?
Interpretation: The same three suppliers face exposure if difficult projects require more work than the engagement recovers, or if customers struggle to put purchased software into use. The current gross losses make the cost boundary relevant, but do not identify which projects caused the shortfall.
For Salesforce, the stated weakness in large transformation demand also cautions against reading every services decline as a successful partner shift. For Workday, partner delivery makes the availability and quality of external implementation capacity relevant. For ServiceNow, the disclosed implementation risk creates a route from project delays to weaker expansion or renewal outcomes.
These are operational mechanisms over project and contract cycles, not forecasts for the shares. Aggregate accounts cannot tell us whether one customer’s delayed deployment has already been offset by successful work elsewhere.
Speculation: automation could change the bill
Tools that automate data mapping or routine configuration could reduce delivery hours. They could also create additional checking and integration work. Neither outcome follows automatically from an attractive demonstration.
The evidence that would change the assessment is specific: comparable project scope, fewer total delivery resources, reliable outcomes and a clear account of whether savings remain with the vendor, partner or customer. Until then, the services line is a place to investigate the deployment burden, rather than a shortcut to declaring the software business successful or broken.
Sources checked 30 October 2026. This article explains business economics and is for information, not investment advice.
Photo by Matheus Bertelli on Pexels. People working together on laptops.