How'd you fix SAP's revenue issues in 2026?

The scenario every enterprise RevOps team recognizes

Picture a €3.2M RISE with SAP contract signed on the last day of the quarter. The rep books it, the number lands in the cloud backlog, the CRO reports it on the earnings call, and the champagne gets poured. Eleven months later, that customer is running exactly one module in production, three business units are still on ECC, the Joule entitlement in the contract has never been switched on, and the renewal conversation is being handled by a success manager who inherited the account from someone who left. The revenue was recognized on a ratable schedule, so the P&L looks fine. But the *expansion* that was supposed to follow — the second and third workload, the analytics attach, the AI seats — never happened, because nobody in the organization owned the gap between "signed" and "used."

Multiply that by a few thousand accounts and you get the shape of SAP's problem. The company has one of the largest committed backlogs in enterprise software, growing faster than reported revenue. That is normally a wonderful signal: it means demand is real and durable. But backlog is a promise, and promises convert to compounding revenue only when the customer actually operates the thing they bought. When conversion from pilot to production sits somewhere in the 40% range across a portfolio that big, the difference between a good year and a great year is not a single deal — it is a systemic delay measured in quarters across the whole installed base.

The scenario has a hard deadline attached to it, which is what makes 2026 different from 2024 or 2025. The compatibility-pack and mainstream-maintenance runway for classic ECC customers is closing. Every account still sitting on the old stack is simultaneously a migration opportunity and a maintenance-revenue risk. If migration velocity does not increase, some of those customers do not migrate faster — they migrate *elsewhere*, or they go to third-party support, or they simply stall and stop paying premium maintenance. The clock converts a chronic issue into an acute one.

The RevOps read on this is unsentimental: the go-to-market motion was designed for a license business and retrofitted onto a subscription business. In a license world, the signature *is* the revenue event, so every incentive, every forecast field, every QBR slide points at the signature. In a subscription world, the signature is the *start* of the revenue event, and everything downstream — go-live date, seat activation, workload count, module attach — is where the money actually compounds. SAP is not unique here. Every large vendor that crossed from perpetual to cloud has fought the same war. Adobe fought it. Autodesk fought it. Microsoft fought it and largely won it by rebuilding the field around consumption metrics rather than contract value. The pattern is well-worn enough that the playbook is not mysterious; it is just organizationally painful.

How the activation mechanism actually works

Trace a single euro through the system and the failure points become obvious. A euro of committed backlog starts as a signature. It becomes deferred revenue on the balance sheet. It becomes recognized revenue on a ratable schedule tied to the subscription term. It becomes *durable* revenue only if the customer renews, and it becomes *growth* revenue only if the customer expands. Between signature and expansion sit four handoffs, and each one leaks.

The first leak is scoping drift. The deal was sold on a business case built with the economic buyer, but the implementation is scoped by a systems integrator working with IT. The success criteria in the sales deck and the acceptance criteria in the SOW are different documents written by different people for different audiences. When they diverge, go-live slips, because "done" was never defined the same way twice.

The second leak is sponsor decay. Enterprise transformations run twelve to eighteen months. Executive sponsors change roles in roughly that same window. If the deal depended on one champion's political capital and that champion moves, the project loses its air cover precisely when it needs a budget defense. Deals do not usually die loudly; they get re-phased, then re-phased again.

The third leak is the enablement orphan. A platform is provisioned, credentials are issued, and then nothing happens because no one has told the actual end users — the AP clerk, the planner, the field service dispatcher — what changes on Monday. AI features are the worst offenders here, because their value is invisible until someone builds the habit of using them. An assistant that nobody opens generates zero business outcome and therefore zero renewal argument.

The fourth leak is measurement. If your CRM tracks contract value and close date but not go-live date, activated seats, or workloads in production, then you literally cannot see the first three leaks. You are flying a subscription business on license-era instruments.

What the diagram makes visible is that three of the four leaks route back into an *earlier* state rather than into churn. That is the encouraging part. This is not revenue that has been lost to a competitor; it is revenue sitting in a holding pattern. Holding-pattern revenue is the cheapest revenue in the world to recover, because the sale is already made, the legal work is done, and the customer has already decided you are the answer. You are not fighting a competitor. You are fighting your own cycle time.

The mechanism that fixes it is equally simple to describe and hard to execute: attach a measurable, owned, dated event to each handoff, and pay someone for hitting it. Not a QBR slide — a field in the CRM with a name attached, a date attached, and a compensation consequence attached. Everything else in a turnaround of this kind is commentary on that one sentence.

Real numbers, ranges, and benchmarks worth anchoring on

Practitioners should be careful about anchoring on any single vendor's internal figures, but the industry benchmarks that frame this problem are well-established and worth stating precisely, because they tell you whether a given number is a crisis or a Tuesday.

CAC payback. Healthy SaaS businesses selling to mid-market typically target 12–18 months of gross-margin-adjusted CAC payback. Enterprise businesses with long implementations routinely run 18–30 months and are not automatically unhealthy, because their net revenue retention and contract durations are far higher. The number that matters is not payback in isolation — it is payback measured against contract length and NRR. A 24-month payback on a three-year contract with 115% NRR is a fine business. The same payback on a one-year contract with 98% NRR is a slow-motion cash fire. When you diagnose a revenue issue, always pull those three numbers together.

Net revenue retention. Best-in-class enterprise software lands in the 115–130% range. Solid-but-unspectacular sits at 105–112%. Below 100% means the installed base is shrinking and every euro of new business is being spent to stand still. For a company the size of SAP, a single point of NRR across the cloud base is worth hundreds of millions annually — which is why activation work, boring as it looks on a slide, has better ROI per dollar than almost any net-new campaign you could fund.

Sales cycle length. Enterprise ERP transformations commonly run 9–18 months from first qualified meeting to signature, with a further 9–18 months to full production across a multi-entity landscape. Mid-market packaged offerings compress that materially — often 3–6 months to signature, 3–6 months to go-live — but at 30–50% lower average contract value. That trade is the strategic tension inside any two-tier portfolio, and it is where comp plans usually break, because a rep with both offerings in their bag will always chase whichever one pays better per hour of effort, not whichever one the company strategically needs.

Pilot-to-production conversion. Across enterprise AI deployments generally — not just at any one vendor — the widely reported pattern is that a large majority of pilots never reach durable production. Ranges cited in industry research cluster in the 60–80% failure band, with the dominant cited causes being unclear ownership, missing data foundations, and no defined success metric rather than model quality. If your organization is converting 40% of pilots, you are *above* the commonly reported norm and still leaving enormous money on the table. Both things are true at once, and executives tend to hear only whichever half suits their argument.

Ramp time. New enterprise AEs typically take 6–9 months to full productivity in complex ERP sales, longer when the portfolio spans two distinct motions. If voluntary attrition in a field organization runs above roughly 20% annually, you are permanently carrying a large cohort of unproductive ramping reps, and the quota-coverage math quietly breaks. Coverage ratios of 3–4x pipeline-to-quota assume productive reps; the same ratio with a third of the team ramping is closer to 2x in practice.

Implementation cost ratio. For large ERP programs, services-to-license spend commonly runs 1:1 to 3:1 depending on customization depth. This is the single most under-discussed number in migration stalls. The customer is not weighing your subscription price; they are weighing your subscription price *plus* an integrator bill that may be triple it. Any move that compresses implementation scope — fit-to-standard, pre-built industry content, reference architectures — attacks the real objection rather than the stated one.

Support-window economics. When a maintenance deadline approaches, the customer's alternatives are: migrate, pay extended-support premiums, move to third-party support at a steep discount, or replatform to a competitor. Third-party support typically prices at roughly half of vendor maintenance, which means every account that chooses that path takes a large, durable bite out of high-margin revenue. The window in which you can influence that decision is 12–24 months before the deadline, not the quarter it lands.

Trade-offs, alternatives, and what you give up either way

There is no version of this fix that is free. Every lever costs something, and the honest RevOps work is naming the cost out loud before the board does.

Comp redesign vs. field stability. Rebuilding a comp plan around velocity, go-live, and attach is the highest-leverage move available, and it is also the one most likely to trigger attrition among exactly the reps who thrived under the old rules. The mitigation is a transition-quarter guarantee and a genuinely higher ceiling — reps forgive a changed formula, they do not forgive a pay cut disguised as strategy. Practical guardrail: hold on-target earnings flat, shift the mix toward variable, and publish the math showing a rep at 120% attainment earns more under the new plan than under the old one at the same attainment. If you cannot produce that slide, your plan is a cut and the field will know within a week.

Land-and-expand vs. new logo. Shifting incentive weight toward expansion improves NRR and payback, but it starves the top of the funnel eighteen months out. Companies that over-rotate here get a beautiful two years followed by a cliff. The usual balance in mature enterprise portfolios is 60–70% of field capacity on the installed base and 30–40% on net-new, with dedicated hunters rather than asking the same person to do both.

Two-tier portfolio vs. focus. Running a large-enterprise motion and a mid-market packaged motion in parallel is strategically correct — it defends the base while opening a cheaper acquisition channel — but it is operationally expensive. The two motions need different reps, different quotas, different partner economics, and different marketing. Trying to run both through one field organization produces exactly the confusion described above: deals stall in the middle market because nobody's plan pays them to be there.

Fit-to-standard vs. customization. Pushing customers to standard processes shortens implementations dramatically and improves upgradeability forever after. It also loses deals to competitors willing to say yes to every requirement. The defensible position is to make the trade explicit in the sales cycle with a written cost comparison rather than discovering it during a change order.

Partner leverage vs. margin control. Systems integrators expand delivery capacity enormously, but they also own the customer relationship during the exact window where activation is won or lost, and their commercial incentive is billable hours, not your go-live date. Structuring partner incentives around a dated go-live milestone rather than a staffed timeline aligns them; leaving it unstructured guarantees drift.

The diagram encodes the single most important prioritization rule in this whole exercise: when backlog is large and conversion is slow, spending on demand generation is the *worst* available use of a euro, because you are adding to a queue that is already backed up. You fix throughput before you increase arrival rate. Any RevOps leader who has ever run a support queue or a manufacturing line knows this instinctively; sales organizations forget it constantly, because pipeline generation is the culturally prestigious activity and go-live operations are not.

Common pitfalls and how to avoid them

Pitfall: treating bookings and revenue as the same forecast. They are two different systems with two different failure modes, and blending them hides both. Fix: maintain two forecasts. One predicts signatures. The other predicts *realized* revenue based on go-live dates, and it is owned by services and success, not sales. When the two diverge, the divergence itself is the leading indicator you have been missing.

Pitfall: buying tooling before defining the metric. Conversation intelligence, competitive intelligence, and data enrichment platforms all deliver real value, but only against a metric someone already owns. Deployed into an organization with no agreed definition of activation, they produce dashboards nobody acts on and a renewal conversation in eleven months about why the tool did not work. Fix: define the metric, instrument it manually in a spreadsheet for one quarter, prove someone changes behavior because of it, *then* buy the platform that automates it.

Pitfall: measuring AI adoption by entitlement rather than usage. Counting how many customers have access to an AI capability is a vanity metric. Counting weekly active users per entitled seat, and the specific workflows they complete, is the real number. The gap between those two figures is usually enormous, and closing it is the cheapest expansion revenue available, because the customer already pays for it and just needs to be shown the door.

Pitfall: enabling the buyer and forgetting the user. Sales enablement is aimed at economic buyers; activation depends on operational end users who never attended a single sales meeting. Fix: a mandatory 30-day post-signature enablement sequence aimed at named end-user roles, tracked as a milestone, not a courtesy email.

Pitfall: letting the deadline do the selling. A maintenance cliff creates urgency, but urgency without a business case produces the minimum viable migration — a technical lift-and-shift that captures none of the expansion upside and leaves the customer resentful. Fix: pair every deadline-driven conversation with a value case for what the customer gets *beyond* compliance. The deadline gets the meeting; it should never be the argument.

Pitfall: assuming churn when the real problem is stall. Teams reflexively treat flat revenue as a competitive loss and respond with discounting. In an activation-constrained business, most of the flatness is delay, and discounting solves nothing while destroying margin permanently. Diagnose before you discount: pull the cohort, look at go-live dates versus contract dates, and see whether the money left or simply has not arrived yet.

Pitfall: no single owner for the handoff. Sales owns until signature, services owns delivery, success owns renewal, and the gap between them belongs to nobody. Fix: name one accountable owner for the signature-to-go-live window with authority over both sides, and report their metric at the same cadence and prominence as bookings. If activation is not on the weekly executive dashboard next to pipeline, it is not actually a priority regardless of what the strategy deck says.

What this looks like in adjacent situations

The pattern generalizes well beyond one vendor, which is a useful sanity check on whether the diagnosis is real. Any business with a long gap between contract and consumption has this exact structure: enterprise infrastructure, medical devices with clinical implementation cycles, industrial equipment with commissioning periods, even large-scale staffing agreements where the requisition is signed months before the first placement. In each case the pathology is identical — the commercial team is compensated on the promise, the delivery team is compensated on the effort, and nobody is compensated on the moment value actually starts flowing.

Two adjacent motions are worth borrowing from. Cloud infrastructure vendors solved this by moving to consumption billing, which makes the incentive alignment automatic: if the customer does not use it, the vendor does not get paid, so the vendor becomes fanatical about adoption. Most enterprise application vendors cannot move wholesale to consumption pricing without wrecking their revenue predictability, but they can borrow the *behavior* by making internal comp partly consumption-shaped even while external pricing stays committed. That is the single cleanest structural fix available and it requires no customer-facing change at all.

The second borrowing is from product-led companies, which instrument activation obsessively because they have no sales team to paper over a bad first week. Their core discipline — define one specific action that predicts retention, then engineer everything toward getting new customers to that action fast — transfers directly to enterprise. The action is bigger (a workload in production rather than a first project created), the timeline is longer, but the logic is the same, and the enterprise version is arguably easier because you have humans who can call the customer.

Finally, the boring internal angle: none of this works without clean account data. Hierarchies that do not reflect how the customer actually buys, duplicate records across acquired subsidiaries, and stale contact data all corrupt the activation metrics before anyone gets to act on them. A data-integrity pass is unglamorous and it is the precondition for everything above. RevOps teams that skip it end up debating whose dashboard is right for two quarters instead of fixing the revenue issues in front of them.

Related questions

Is a large cloud backlog a good sign or a warning?

Both. Growing backlog proves durable demand. But backlog growing materially faster than recognized revenue for several consecutive quarters means conversion is slowing, and the gap is a queue, not a cushion. Track the ratio, not either number alone.

Should sales or customer success own the go-live milestone?

Neither alone. Name a single accountable owner for the signature-to-production window with authority across both functions, and put their metric on the weekly executive dashboard beside bookings. Split ownership is why the gap exists in the first place.

How fast can a comp plan change actually move the number?

Behavior shifts within one quarter; revenue impact lands two to four quarters later given enterprise cycle lengths. Expect a transition-quarter dip in reported activity as reps re-learn the plan. Guarantee earnings through that quarter or you will lose good people.

Does an approaching maintenance deadline help or hurt?

It creates urgency and compresses decision timelines, which helps. It also invites minimum-viable migrations that capture none of the expansion upside, and pushes some accounts toward third-party support at roughly half your maintenance price. Pair the deadline with a real value case.

What single metric would you add to CRM first?

Go-live date, with an owner and a change history. It is the field that converts a bookings forecast into a revenue forecast, and its slippage rate is the earliest reliable warning that backlog is not converting.

FAQ

Is SAP's 2026 problem demand or execution?

Overwhelmingly execution. Backlog growth indicates customers are still choosing the platform and committing multi-year spend. The constraint sits between signature and production — scoping drift, sponsor turnover, unenabled end users, and forecasting instruments built for a license era. Fixing throughput on committed business returns more per euro than generating additional demand into an already-backed-up queue.

Why does pilot-to-production conversion matter more than win rate?

A won deal that never reaches production generates ratable revenue but no expansion, no reference, and a fragile renewal. Conversion is the multiplier on every deal you already won, so improving it compounds across the entire installed base at once. Win rate only improves the deals you have not closed yet, and it costs far more to move.

What does the ECC maintenance deadline actually change?

It converts a chronic migration backlog into a dated decision for every remaining legacy customer. Each account must migrate, buy extended support, move to third-party support at a steep discount, or replatform. The influenceable window opens roughly 12–24 months before the date, which means the work is largely front-loaded and mostly done by the time the deadline arrives.

How do you monetize AI capabilities that customers already have entitlements for?

Stop counting entitlements and start counting weekly active users per entitled seat, broken out by workflow. Build a 30-day post-signature enablement sequence aimed at named end-user roles rather than economic buyers. Adoption creates the usage evidence that justifies the next tier — without it, the renewal conversation has no data behind it and defaults to a price discussion.

Should the mid-market offering and the enterprise offering share a sales team?

Generally no. They have different cycle lengths, different average contract values, and different partner economics. A rep carrying both will optimize for whichever pays better per hour of effort, which strands the middle market. Separate quotas and, where volume justifies it, separate teams. Keep a defined, compensated path for upgrading a mid-market account into the enterprise motion.

What is the realistic timeline for this kind of turnaround?

Instrumentation and metric definition take one quarter. Comp and process changes take a second quarter to design and land. Behavioral change shows up in the third. Revenue impact appears in quarters four through six, because enterprise cycles are long. Anyone promising results inside two quarters is describing a bookings pull-forward, not a structural fix.

Sources

• https://www.sap.com/investors/en/reports.html
• https://www.sap.com/products/erp/rise.html
• https://support.sap.com/en/release-upgrade-maintenance.html
• https://www.gartner.com/en/information-technology
• https://www.mckinsey.com/capabilities/growth-marketing-and-sales/our-insights
• https://hbr.org/topic/subject/sales
• https://www.bain.com/insights/topics/customer-strategy-and-marketing/
• https://www.forrester.com/blogs/category/b2b-sales/
• https://www.bvp.com/atlas/state-of-the-cloud-2024

Related on PULSE

• How'd you fix Oracle's revenue issues in 2026?
• How'd you fix Workday's revenue issues in 2026?
• How'd you fix Salesforce's revenue issues in 2026?
• How'd you fix ServiceNow's revenue issues in 2026?
• How'd you fix Snowflake's revenue issues in 2026?
• How'd you fix HubSpot's revenue issues in 2026?