What is the bear case for ServiceNow 2027?

The two futures being priced, side by side

Every argument about ServiceNow in 2027 is really an argument between two stories, and the stock price is a weighted average of them. Understanding the bear case requires holding both simultaneously, because the bear case is not "ServiceNow is a bad company" — it is "ServiceNow is a good company being paid for as a great one."

The bull story goes like this. ServiceNow is the third great enterprise platform, alongside SAP and Salesforce, and it is the one best positioned for the agentic era because it already owns the workflow layer where work actually gets executed. AI agents need somewhere to *do* things — open a ticket, provision a laptop, approve a purchase order, route a case, update an entitlement. That execution surface is ServiceNow's home turf. Now Assist becomes a second platform layered on the first, AI credits create a consumption line item on top of a seat-based subscription line item, and the company holds 20%+ subscription growth deep into the decade while free cash flow margin compounds. In this story, the premium forward-sales multiple is not a premium at all — it is a discount to what a durable 20% grower with 30%+ FCF margins and a second growth engine should command.

The bear story accepts most of the same facts and disputes the durability. It argues that the workflow execution layer is valuable but contestable; that the customers most likely to expand ServiceNow beyond IT are exactly the customers Microsoft can reach for free; that AI monetization arrives as SKU adoption rather than compounding consumption; and that the go-to-market machine which powered the last decade is being quietly disassembled by talent competition from AI-native companies with equity stories ServiceNow can no longer match. None of those individually breaks the company. Together, they turn a 20% grower into a 16% grower, and a 16% grower does not hold a hyper-growth multiple.

The critical structural point for anyone in a RevOps or strategy seat: these two stories diverge almost entirely on *net revenue retention*, not on new logos. ServiceNow's growth algorithm has always been expansion-led — land in IT service management, expand into IT operations, then employee workflows, then customer workflows, then whatever the platform team builds on the App Engine. If net dollar retention holds around 120%, the company barely needs new logos to hit its numbers. If it drifts toward 108%, the entire load shifts to new-logo acquisition, which is the most expensive, slowest, and most competitively contested motion the company runs. The bear case lives or dies on that single number, and every risk below is ultimately a mechanism that pushes it down.

There is also a third framing worth naming, because it is where most of the actual probability mass sits: the *muddle-through* case. Growth decelerates gradually rather than sharply, AI attach lands in the middle of the range, Microsoft takes the low end of the mid-market but not the enterprise, and the multiple compresses partially rather than fully. Muddle-through is still a materially negative outcome for a stock priced on continuation, which is what makes the bear case interesting: it does not require the pessimistic scenario to be right. It only requires the optimistic one to be wrong.

How to decide which story you are underwriting

The way to resolve the two stories is not argument — it is instrumentation. Each thesis makes different predictions about observable quarterly data, and those predictions diverge in a specific order. Below is the decision structure, sequenced by which signal resolves first.

The first fork is retention, because it resolves earliest and contaminates everything downstream. Watch it in two places: the disclosed renewal rate, which is a blunt instrument because it measures logo retention rather than dollar expansion, and current remaining performance obligation growth, which is the closest public proxy for forward-booked demand. cRPO is the number that matters most because it is bookings-adjacent and harder to manage optically than revenue, which is smoothed by ratable recognition. A company can print in-line revenue for two or three quarters after demand has already turned, because revenue is recognizing bookings from prior periods. cRPO turns first.

The second fork is AI monetization *quality*, which is a subtler read than AI monetization *presence*. A vendor can report strong AI SKU adoption while the underlying usage stays shallow — customers buy the tier during a renewal because it was bundled into the negotiation, then deploy it against two or three narrow use cases and never expand. That produces a good attach headline and a bad consumption curve. The tell is whether management talks about *deployments* and *customers* or about *usage*, *credits consumed*, and *expansion within existing AI customers*. Language drift from the latter back to the former is the single most reliable qualitative signal that AI is landing as a feature rather than a platform.

The third fork is competitive win rate in the segment that matters. ServiceNow's enterprise core — large regulated organizations with complex, audited IT estates — is genuinely defensible. Nobody rips out a decade-old ServiceNow instance with hundreds of custom workflows because Power Automate is cheaper. But the *expansion* thesis was never mostly about that core. It was about departmental workflow spreading into HR, legal, procurement, facilities, and field service, and about mid-market organizations adopting the platform as their first serious workflow system. Both of those are precisely where a bundled, adequate, already-paid-for alternative wins on procurement math rather than product merit.

For a RevOps practitioner reading this as an operating question rather than an investing one, the same fork structure applies to your own stack decision. If your organization's ServiceNow footprint is deep in IT and shallow everywhere else, you are the customer whose expansion the bear case assumes will not happen — and you should be honest with yourself about why. Usually the answer is that the incremental license cost for an HR workflow is hard to defend when a Power Platform app the internal team can build in a sprint gets to 80% of the outcome. That is the bear case, expressed as a purchase order.

The Microsoft bundling problem, in procurement terms

The Microsoft threat is routinely described in product terms and it is almost always a procurement story instead. This distinction matters because product comparisons flatter ServiceNow and procurement comparisons do not.

On product merit, ServiceNow generally wins. Its data model, its workflow engine, its CMDB, its governance and audit posture, its out-of-the-box ITIL process coverage — these represent an enormous accumulated advantage that a general-purpose low-code platform does not replicate. Ask any experienced ITSM administrator to compare a mature ServiceNow implementation against a departmental Power Platform build and the comparison is not close.

But that is not the comparison being made in the room where the decision happens. The actual comparison is: we already pay Microsoft for the productivity suite, the identity layer, the security tooling, and increasingly the AI assistant. The workflow builder is included or nearly so. The internal team already knows it. It shares single sign-on and data governance with everything else we own. It does not require a new vendor security review, a new master services agreement, a new procurement cycle, or a new line item the CFO will question. It gets to *good enough* in weeks.

That framing wins a specific category of deal: the departmental expansion, the greenfield mid-market deployment, the "we need to digitize this one process" project that would historically have been ServiceNow's land motion. It loses the deep enterprise replacement, which it will not attempt. So the effect is not a share collapse — it is a *ceiling*. ServiceNow keeps what it has and grows it, but the addressable frontier narrows, and narrowing frontier plus intact core equals decelerating growth with strong retention. That combination is exactly what produces multiple compression without any dramatic bad news.

Three mechanisms make the bundling pressure worse over time rather than better:

Identity and governance gravity. The more sensitive the workflow, the more the compliance and security organization prefers the platform already integrated with the corporate identity provider, data loss prevention tooling, and information protection labels. HR service delivery, legal operations, and anything touching employee records or regulated data are precisely ServiceNow's highest-value expansion targets — and precisely where "keep it inside the existing security perimeter" is the most persuasive argument in the building.

Systems integrator economics. The consulting ecosystem follows deal volume, and the partner network attached to the Microsoft stack is dramatically larger than any single ISV's. For a mid-market implementation where the SI's margin is thin, the partner steers toward the platform with the deeper bench, the cheaper resources, and the shorter ramp. Channel preference is a slow-acting force but it compounds, and it is very hard to reverse once a regional SI practice has standardized on an alternative.

The agent-builder collision. Both companies now sell the same story — describe a process in natural language, get an agent that executes it against your systems. When two vendors tell the same story and one of them is already in the customer's bill, the differentiated one has to prove a materially better outcome, not a marginally better one. Proving *materially better* is a long sales cycle and an expensive proof-of-concept, and it is exactly the kind of friction that shows up later as lengthening sales cycles and declining win rates in the mid-market cohort.

The counter-argument, which deserves fair weight: bundled tools have historically been good enough for simple use cases and inadequate for complex ones, and enterprise workflow trends toward complexity, not away from it. Organizations that start on a bundled low-code platform frequently hit a governance wall — sprawl, no service catalog, no CMDB, no auditable change process — and then buy the platform that solves it. That has been the pattern for two decades. The bear case's claim is not that this pattern is wrong; it is that the wall now arrives later, because the bundled tooling is meaningfully better than it used to be, and *later* is all the bear case needs.

Where Salesforce actually threatens, and where it does not

The customer-service AI battle is worth isolating because it involves a genuine structural asymmetry rather than a pricing one.

Service AI, done well, requires two things: the ability to *understand* the customer's situation and the ability to *act* on it. Understanding requires the customer record — who is this person, what did they buy, what have they contacted us about before, what is their entitlement, what happened in the last three interactions across every channel. Acting requires the workflow engine — issue the credit, dispatch the technician, escalate to tier two, trigger the replacement order, update the case with an auditable trail.

ServiceNow owns the second half natively and has to integrate to get the first. Salesforce owns the first half natively and has to build or acquire the second. Both are solvable, but the asymmetry favors whoever's native half is harder to replicate — and a customer data platform assembled from every interaction across marketing, commerce, and support is a heavier lift than a workflow orchestration layer. That is the real argument for why service AI may consolidate toward the system of record for the customer rather than the system of record for the work.

Where that argument overreaches: it assumes customer service is one market. It is not. There is a large, structurally different segment — service that involves physical assets, field dispatch, complex entitlements, regulated SLAs, and deep integration with asset and configuration data — where the workflow half dominates the value and the CRM half is table stakes. Telecommunications, medical devices, industrial equipment, financial services operations, healthcare provider support. In those segments the workflow-native vendor has a real and defensible position, and losing a flagship consumer-facing service deployment says very little about them.

So the honest bear formulation is narrower than "Salesforce wins customer service." It is: *the customer service TAM that ServiceNow underwrote in its expansion story includes a large tranche of high-volume, CRM-adjacent service work that is likely to consolidate onto the customer-record platform, and if you remove that tranche the CSM growth contribution caps below what the bull model assumes.* That is a smaller claim and a far more defensible one, and it is worth roughly what a TAM haircut is worth — meaningful but not thesis-breaking on its own.

A related downstream effect that rarely gets modeled: the cross-sell leverage. Part of ServiceNow's expansion economics assumed that landing customer workflows would pull through additional platform adoption — field service, order management, custom apps built on the same platform. If the CSM anchor is weaker, the pull-through is weaker too, and that shows up as lower expansion revenue in adjacent product lines that were never directly contested. Competitive losses propagate; they rarely stay contained to the product that lost.

The numbers that would have to be true

Abstract arguments are cheap. Here is the arithmetic the bear case actually requires, expressed as ranges rather than false precision.

Growth. ServiceNow's model has been a roughly 20%+ subscription grower with guidance the company has historically set conservatively and beaten modestly. The bear case needs that to land in the mid-to-high teens and stay there — not a single soft quarter, which the market forgives, but a sustained new baseline. The reason the specific threshold matters: the market treats 20% as a psychological regime boundary. A durable 20% grower is compared to the small set of large-cap software companies still compounding at that rate. A 16% grower is compared to the much larger set of mature enterprise software franchises, and that comp set trades at a materially lower multiple on identical margins.

Retention. Net dollar retention is the transmission mechanism. Historically ServiceNow has operated well above 120%. The bear case needs it in the 105-110% range. The arithmetic is unforgiving: at 120% retention, the installed base alone delivers most of the growth target and new logos are upside. At 107%, the installed base delivers roughly a third of a high-teens growth target, and the remaining two-thirds must come from new customers acquired in the exact segment where bundled competition is strongest. The company would have to dramatically increase new-logo productivity at the same moment competitive pressure is reducing it. That is the bear case's central mechanical claim, and it is internally coherent in a way most bear cases are not.

AI monetization. The bull model implies AI becomes a material, separately meaningful revenue contribution — a second engine, not a feature toggle. The bear case needs AI attach to plateau and per-customer consumption to stay shallow. The diagnostic is the gap between *tier adoption* and *credit consumption*. If customers are on the premium tier but consuming a small fraction of their entitled AI capacity, the renewal conversation two years later is a downgrade conversation, because procurement can see the utilization data as clearly as the vendor can. Shelfware is always discovered at renewal, and it is discovered with a number attached.

Pricing. Aggressive uplift on the AI tier is a one-time revenue benefit and a recurring relationship cost. It works if utilization justifies it and backfires if it does not. The second renewal cycle after a large uplift is where the bill comes due — customers who accepted the increase once, tracked usage for two years, and arrive at the next negotiation with a utilization report and a competing quote. If a meaningful tail of the base steps down a tier, ARR contracts inside accounts that are still nominally retained, which is exactly how NDR erodes without churn showing up in logo retention.

Sales productivity. Net new ARR per quota-carrying rep is the quiet variable. Talent competition from AI-native companies with compelling equity stories pulls experienced enterprise sellers out of mature platforms; each departure costs the ramp time of a replacement plus the relationship equity that does not transfer. A double-digit percentage decline in productivity per rep means the company must carry more headcount for the same bookings, which pressures operating margin at precisely the moment growth is decelerating. Margin and growth deteriorating together is what breaks a Rule of 40 story — either one alone is survivable.

Multiple. Put those together and the re-rating math is mechanical rather than mysterious. A high-teens grower with excellent margins and a contested AI story does not hold a hyper-growth forward-sales multiple; it converges toward the mature-platform band. The gap between those two bands is where the entire bear-case return lives. Note what is *not* required: no revenue decline, no margin collapse, no scandal, no product failure. Just deceleration plus a contested second act.

Adjacent read-through. The same arithmetic applies across the expansion-led enterprise software cohort — observability, data platforms, sales engagement, developer tooling. Any vendor whose valuation assumes both continued 20%+ growth *and* a credible AI second platform is underwriting the same two variables. When one of them re-rates, the cohort tends to re-rate together, because the market is repricing the assumption class rather than the individual company. That correlation is why bear cases in this category tend to arrive in clusters rather than one at a time.

Implementation: how to actually track this, quarter by quarter

If you are operating rather than speculating — running vendor strategy, RevOps, or an internal platform roadmap — the useful output is not a price target. It is a monitoring cadence that tells you early whether your platform bet is compounding or plateauing, and a sequencing plan for what you do in each case.

The quarterly instrumentation. Four inputs, thirty minutes a quarter. First, forward-demand growth from cRPO — the trend line matters far more than any single print, and two consecutive decelerating quarters is the standard threshold before you change behavior. Second, the adoption-versus-usage gap on the AI tier, which you can measure on your own instance more precisely than any analyst can measure it externally. Third, your own expansion pipeline outside the original IT footprint — if you are not building new workflows on the platform, you are personally evidence for the bear case regardless of what the stock does. Fourth, qualitative language drift in management commentary from usage metrics back to logo counts.

The internal audit that pays for itself. Independent of any investment view, run a utilization audit by module and by seat before every renewal. Count active users against licensed users, and count workflows in production against workflows built. Most large deployments carry meaningful shelfware — modules purchased during a bundled negotiation that never reached production, seats provisioned for a team that reorganized, AI capacity entitled but never consumed. That audit is the single highest-leverage RevOps action available on a large platform contract, because it converts a renewal from a vendor-controlled conversation into a data-controlled one. Start it two quarters before renewal, not two weeks.

The sequencing decision for new workflows. The practical fork for most organizations is not "keep or replace" — nobody sensibly rips out a mature core instance. It is "where does the *next* workflow get built." A defensible policy: workflows that require the CMDB, formal change management, auditable approvals, cross-department orchestration, or regulated SLAs go on the incumbent platform, because that is where the accumulated advantage is real and rebuilding it is a multi-year mistake. Workflows that are a form, a routing rule, and a notification go wherever the marginal cost is lowest, which increasingly means the bundled option. Write that policy down explicitly, because in the absence of a written policy the default is whatever the loudest internal team prefers, and that produces sprawl in both directions.

The renewal play. Bring three things to the negotiation: a utilization report by module, a documented alternative for at least one category of workflow, and a multi-year framing. Vendors price against perceived alternatives; a documented one changes the conversation materially, and a credible multi-year commitment is worth real discount because it improves the vendor's own retention metrics. Do not bluff a full replacement you would never execute — it is transparent and it damages the relationship you will still need. Negotiate the tier and the entitlement, not the existence of the contract.

What would invalidate the bear case, and you should say so in advance. Retention holding above 115% while AI consumption compounds inside existing accounts. Mid-market win rates stable against bundled competition over several quarters. Management shifting *toward* usage disclosure rather than away from it, which companies do when the usage numbers are good. Any of those sustained for a few quarters and the bear case is simply wrong — and the discipline of naming the falsifiers in advance is what separates analysis from a position you are talking your book on.

The honest probability weighting. The full bear case — sharp deceleration, failed AI monetization, structural share loss, and a full multiple re-rate arriving together — is not the modal outcome. The modal outcome is partial: some deceleration, mixed AI results, mid-market pressure that shows up in win rates rather than churn, and partial compression. What makes the bear framing worth doing anyway is that a stock priced for continuation loses real value in the partial case too. You do not need the pessimistic scenario to be right; you only need the optimistic one to be incomplete. That asymmetry is the entire argument, and it is why serious analysts run this exercise on their highest-conviction long positions rather than only on the ones they dislike.

*Scenario analysis for planning purposes — not investment advice.*

Related questions

Does the bear case mean ServiceNow loses its existing customers?

No. Logo churn in deep enterprise deployments is very low and the bear case does not assume otherwise. It assumes expansion slows and a tail of accounts steps down a tier at renewal — ARR contraction inside retained customers, which erodes net revenue retention without showing up as churn.

Is Microsoft actually replacing ServiceNow anywhere?

Rarely as a replacement, frequently as a *pre-emption*. Bundled low-code tooling wins departmental and greenfield mid-market workflows that would previously have been ServiceNow land deals. The effect is a narrowed expansion frontier rather than displacement of installed enterprise instances.

How would a RevOps team see this before the market does?

Through internal telemetry: licensed seats versus active users, modules purchased versus modules in production, AI capacity entitled versus consumed. Your own utilization data leads public disclosure by several quarters and is far more precise than any external estimate.

What single metric matters most?

Net dollar retention, with current remaining performance obligation growth as the earliest public proxy. Retention is the transmission mechanism for every other risk — competitive pressure, pricing backlash, and AI underperformance all ultimately express themselves as slower expansion inside the installed base.

Does this apply to other enterprise software platforms?

Yes. Any expansion-led vendor priced for both sustained 20%+ growth and a credible AI second act is underwriting the same two variables. Observability, data platform, and sales engagement vendors face structurally similar re-rating risk when either assumption weakens.

FAQ

Is the ServiceNow bear case a prediction that the company will fail?

No, and framing it that way misses the point. The bear case does not require a bad business — it requires an ordinary one. ServiceNow can keep growing, keep generating strong free cash flow, and keep its enterprise customers while still being worth considerably less, because the current valuation embeds continued hyper-growth plus a successful second platform. Remove either pillar and the arithmetic changes substantially without anything dramatic happening operationally.

Why does net dollar retention matter more than new customer growth?

Because ServiceNow's entire growth algorithm is expansion-led. Land in IT service management, expand into operations, employee workflows, customer workflows, and custom applications. When retention runs well above 120%, the installed base alone carries most of the growth target. When it drifts toward 107%, the burden shifts onto new-logo acquisition — the slowest, most expensive, and most competitively contested motion — at exactly the moment competitive pressure makes that motion harder.

How real is the Microsoft bundling threat, honestly?

Real at the edges, overstated at the core. Nobody replaces a mature instance with hundreds of custom workflows, a populated configuration management database, and years of audit history because a bundled tool is cheaper. But departmental expansion and greenfield mid-market deployments are decided on procurement math — already paid for, already integrated with identity, no new vendor review — and that math frequently favors the bundled option. The threat is a ceiling on the expansion frontier, not an eviction from the core.

What is the difference between AI attach and AI monetization?

Attach measures how many customers bought the tier. Monetization measures whether they use it enough to keep paying for it. A vendor can report strong attach while underlying consumption stays shallow, because tiers often get bundled into renewal negotiations rather than bought on demonstrated value. The gap between entitled capacity and consumed capacity is the leading indicator, and it resolves at the second renewal, when procurement arrives with a utilization report.

Should a RevOps leader change platform strategy based on this?

Not wholesale, but yes on sequencing. Keep the core where accumulated advantage is genuine — anything needing the CMDB, formal change control, auditable approvals, or regulated service levels. Evaluate genuinely new departmental workflows on marginal cost, since a form plus a routing rule plus a notification does not require an enterprise workflow platform. Write the policy down; without one, the default becomes whatever the loudest internal team prefers.

What evidence would prove the bear case wrong?

Retention sustained above 115%, AI consumption compounding inside existing accounts rather than plateauing after initial deployment, stable mid-market win rates against bundled competition across several quarters, and management voluntarily increasing usage disclosure — companies disclose usage metrics when those metrics are favorable. Any of those holding for several consecutive quarters materially weakens the thesis, and naming the falsifiers in advance is what keeps the analysis honest.

Sources

• https://investors.servicenow.com/ — ServiceNow investor relations: quarterly results, guidance, and supplemental metrics
• https://www.sec.gov/edgar/search/ — SEC EDGAR full-text search for 10-K and 10-Q risk factor disclosures
• https://www.gartner.com/en/information-technology — Gartner IT research on service management and workflow automation markets
• https://www.forrester.com/research/ — Forrester research on enterprise workflow platforms and competitive positioning
• https://learn.microsoft.com/en-us/power-platform/ — Microsoft Power Platform documentation and licensing structure
• https://www.salesforce.com/products/data/ — Salesforce Data Cloud product documentation and positioning
• https://www.mckinsey.com/capabilities/mckinsey-digital — McKinsey Digital research on enterprise software adoption and transformation
• https://www.bain.com/insights/topics/technology/ — Bain technology insights on software valuation and growth durability

Related on PULSE

• What is the bear case for Salesforce 2027?
• What is the bear case for Snowflake 2027?
• What is the bear case for Datadog 2027?
• What is the bear case for Outreach 2027?
• What is the bear case for Salesloft 2027?