How do you build an expense management go-to-market motion in 2027?

Segment and ICP first: who actually signs an expense platform

The single biggest wasted motion in this category is treating "expense management" as one market. It is at least three markets with different buyers, different price points, and different sales cycles, and a team that runs one playbook across all three will lose the enterprise deals on compliance and lose the SMB deals on cost-to-serve.

Segment one: SMB, roughly 10 to 100 employees. The buyer is usually a founder, an office manager, or an outsourced accountant. There is no procurement function, no InfoSec review, and no RFP. The cycle runs 30 to 60 days from first touch to first card swipe, and often much shorter when a bookkeeper drives it. Software revenue at this tier is near zero — the modern entrants give the software away and monetize card spend. Your acquisition here has to be self-serve: website sign-up, instant virtual card issuance, and an accounting integration that connects in under ten minutes. Any motion that requires a human demo at this tier burns more CAC than the account will ever return.

Segment two: mid-market, roughly 100 to 1,000 employees. The buyer is the controller, with the CFO as economic approver. This is where a real sales cycle appears — typically three to four months — because there is now a month-end close process to protect, a general ledger with meaningful coding conventions, and an existing card program (often a bank-issued corporate card with no software layer) to displace. Deal sizes in the low-to-mid five figures of annual software revenue, plus whatever card spend flows through. This is the sweet spot for most challengers: big enough to pay, small enough to skip a formal RFP.

Segment three: enterprise, 1,000+ employees. Four to five months minimum, formal RFP, InfoSec questionnaire, and a procurement team whose job is to grind your price down. The committee widens to four seats and often five. Software ACV reaches six figures, but so does implementation effort. Do not enter this segment before you have SOC 2 Type II, and do not enter it before you have three referenceable customers of comparable size in the same vertical.

The four-seat committee. Across mid-market and enterprise, four roles reliably show up, and each needs a distinct artifact:

• CFO — owns the spend-control and audit-trail narrative. Wants to know: how much unmanaged spend exists today, and what does the policy engine actually stop? Give this seat a one-page economic memo, not a demo.
• Controller — the daily product owner and the real decision-maker. Runs month-end close, general ledger coding, and audit prep. Wants to know: how many days does close shrink, and how much manual recoding survives? Give this seat a hands-on sandbox against their own data.
• CIO or head of finance systems — owns integration with NetSuite, QuickBooks, Sage Intacct, Workday Financials, Microsoft Dynamics 365 Finance, Oracle Cloud ERP, and Xero. Wants to know: is this a certified integration or a nightly CSV? A missing native ERP connector is the single most common source of a technical veto.
• Head of procurement or sourcing — increasingly co-signs, because card and expense data feeds spend classification and vendor consolidation. Wants to know: can I see all spend by vendor across cards, invoices, and reimbursements in one view?

Vertical overlays that change the ICP materially. Professional services firms need client-billable expense allocation and project coding, which makes the sale about revenue recovery rather than cost control. Field-services and construction firms need mileage, per diem, and job costing, and often have low smartphone adoption among the actual spenders. Nonprofits and grant-funded organizations need fund-level restriction tracking that most consumer-grade tools cannot express. Multinationals need multi-entity, multi-currency, and VAT reclaim, which is a genuinely hard product problem and a legitimate reason to disqualify early rather than lose at month four.

Disqualify aggressively. The accounts that look attractive and are not: companies mid-ERP-migration (they will freeze all finance projects), companies whose card program is contractually bundled into a banking relationship with an early-termination penalty, and companies with under roughly 25 monthly expense reports, where the incumbent process is a shared spreadsheet and the pain is genuinely not worth a purchase decision.

The motion that fits: sandbox-led, controller-validated, CFO-closed

Once the segment is set, the motion follows from it. The mid-market and enterprise motion has five stages, and the compression lever sits in stage three.

Stage one — trigger identification. Expense platforms are almost never bought on a cold call. They are bought on a trigger: an incumbent contract renewal window, a funding round that suddenly puts real cash on the balance sheet, an acquisition that leaves two incompatible expense processes, controller turnover (a new controller reorganizes tooling within the first two quarters far more often than a tenured one), an audit finding on expense documentation, or a new CFO hire. Your outbound targeting should be trigger-driven, not title-driven. Sourcing signals: job postings for controllers and finance systems analysts, funding announcements, ERP implementation partner announcements, and renewal timing inferred from the customer's original announcement of the incumbent.

Stage two — vendor scan. Finance buyers do a defensible-decision scan: analyst coverage, peer review sites, a peer network thread, and two or three vendor websites. You need presence in all four surfaces before you need a bigger sales team. The peer network channel is heavily underweighted by most GTM teams — controllers ask other controllers, and a single strong reference in a well-trafficked finance community converts better than a quarter of paid search.

Stage three — the spend-data sandbox. This is the differentiating motion. Instead of a scripted demo, ingest six to twelve months of the prospect's historical expense and general ledger data into an isolated environment, then show four numbers: what percentage of transactions your engine auto-codes to the correct GL account, how many duplicate submissions it catches that the incumbent process missed, how many policy-violation flags it raises, and the projected card-spend economics. Run this as a fixed 30-day engagement with a named controller-side owner and a scheduled readout. The reason this works is that it converts an opinion argument ("our AI is better") into an arithmetic argument on the buyer's own data. Operationally, protect it: require a signed data-handling addendum, use anonymized or tokenized employee identifiers, and cap the number of concurrent sandboxes your solutions team runs — three per solutions engineer is a realistic ceiling if each one gets a real tuning pass.

Stage four — references. Finance buyers weight peer references more heavily than almost any other function. Build a reference program deliberately: three to five customers per segment and per ERP, each briefed on what to expect, each compensated with something real (early access, an advisory seat, conference travel — not cash). The reference call the buyer actually wants is with a controller who runs the same ERP and closed a comparable migration.

Stage five — procurement, legal, and security. Three to six weeks in mid-market, longer in enterprise. Have SOC 2 Type II, ISO 27001, PCI DSS attestation for any card issuance, documented SOX-relevant controls, and GDPR/data-residency answers packaged before the first enterprise conversation, not assembled during it. The most common self-inflicted delay is a security questionnaire that takes your team two weeks to answer because nobody owns it.

Channel mix that supports this motion. At scale, a workable distribution is roughly 40 percent inbound (review sites, content, investor-portfolio referrals), 25 percent outbound targeted at CFO, controller, and procurement titles, 15 percent partner-led through accounting firms and bookkeeping networks, 15 percent conference and community presence in finance-specific venues, and 5 percent marketplace listings inside the CRM and ERP ecosystems your buyers already run. The accounting-firm channel deserves disproportionate attention in SMB and lower mid-market: a single regional firm that standardizes on your platform brings dozens of clients with near-zero incremental CAC, and the firm's own staff become your implementation labor. Build that channel with a revenue share, a partner portal with a multi-client console, and continuing-education credit content — not with a generic referral link.

Unit economics and the benchmarks that decide whether this works

The economics of this category changed structurally when card issuance became a viable revenue line. You cannot design the GTM without deciding which side of that line you are on.

The two revenue models. The classic model charges per user per month for software, with published list pricing commonly in the range of single-digit to mid-double-digit dollars per user per month depending on module depth, plus per-report or per-transaction fees in some legacy contracts. The interchange model charges little or nothing for software and earns a share of the interchange fee on card spend — interchange on commercial cards typically runs materially higher than consumer debit, which is what makes the model viable. A customer running large monthly card volume can generate more revenue through interchange than they would ever have paid in seat fees.

What this means for pricing strategy. If you are entering as a challenger, per-seat pricing alone is structurally hard: an incumbent offering comparable functionality at zero software cost will win the price conversation before you finish the demo. The realistic options are (a) issue cards and take interchange, accepting the regulatory and capital overhead that comes with it, (b) sell into segments where card issuance is impractical — heavily regulated entities, organizations with entrenched banking relationships, multinationals with complex entity structures — and price on software, or (c) hybrid, where software is priced but bundled cards subsidize it. Option (a) is not free: card issuance drags in PCI DSS scope, KYC and KYB obligations, sponsor-bank or issuer-processor relationships, fraud loss exposure, and a compliance function you must staff. Budget for that before you promise zero-cost software.

Discounting and term structure. Multi-year commitments close meaningfully more often than annual ones in this category, because finance buyers value predictable cost and hate re-running a procurement cycle. Expect to trade a high single-digit to low double-digit percentage discount for a three-year term. Build a volume curve rather than negotiating ad hoc: list price up to a first threshold, then stepped discounts at meaningful user-count bands, then a negotiated tier for the largest deals. In interchange-funded deals, the negotiation lever is usually a rebate — a share of interchange returned to the customer on their own spend — which large customers will increasingly demand and which you should model as a margin reduction from day one rather than a surprise concession.

The ROI argument you actually make. Three quantifiable buckets, and you should be able to compute all three from the sandbox data:

1. Close-cycle labor. Time the finance team spends chasing receipts, recoding transactions, and reconciling card statements. Quantify in hours per month multiplied by loaded cost, and be conservative — controllers discount aggressive estimates instinctively.
2. Leakage and policy. Duplicate submissions, out-of-policy spend that gets approved because nobody has time to check, and unused SaaS subscriptions surfaced by card-spend visibility. This is often the largest bucket and the least anticipated.
3. Working-capital and rebate yield. Card float, early-payment discount capture on the AP side, and any negotiated interchange rebate.

Benchmarks to run the business against. Net revenue retention in the 115 to 135 percent range is the mark of a platform that is attaching adjacent modules; a single-product expense tool with no expansion path tends to land near or slightly above 100 percent, which is survivable but not fundable. CAC payback in the 8 to 16 month range is normal for a mid-market motion with a solutions-engineering-heavy sandbox stage; if you are materially above that, the sandbox is being run on unqualified deals. Gross margin varies dramatically by model — software-only is a conventional SaaS margin, while interchange-funded revenue carries real cost of revenue in card processing, rewards, and fraud reserve. Win rates against a legacy incumbent in a competitive evaluation in the 30 to 45 percent range are healthy; below that, you are usually losing on ERP integration depth rather than on product.

Cost-to-serve is the hidden variable. Implementation for a mid-market customer should run two to eight weeks. If it routinely runs longer, either your ERP connectors are incomplete or you are selling to companies whose chart of accounts is a mess. Both are fixable, but only one is a product problem. Track implementation hours per deal as a first-class metric and gate segment expansion on it.

Common misfires that kill an expense GTM

Demoing instead of sandboxing. The scripted demo shows your data, your clean chart of accounts, and your happy path. Every controller has seen it and discounts it entirely. Deals that never touch the buyer's own data close slower and churn earlier, because the coding accuracy that looked great in the demo collapses against a real GL with 400 accounts and inconsistent historical conventions.

Shipping without native ERP connectors. A nightly CSV export is not an integration, and the finance systems owner will say so. Prioritize connectors by where your ICP actually is: QuickBooks and Xero for SMB, NetSuite and Sage Intacct for mid-market, Workday Financials, Dynamics 365 Finance, and Oracle Cloud ERP for enterprise. Certified marketplace listings matter more than API availability, because they signal that someone else already validated the integration.

Under-tuned auto-coding at pilot review. If your model codes at a low accuracy rate against the customer's historical data, the controller concludes the product creates work rather than removing it, and no amount of interchange economics recovers the deal. The fix is process, not model: budget a tuning pass inside the sandbox window, train on the customer's own historical coding decisions, and report accuracy honestly, including which categories fail. A vendor who says "we're at 71 percent on your travel categories and here's the fix" is more credible than one claiming universal accuracy.

Treating compliance as a late-stage checklist. PCI DSS scope, SOC 2 Type II, and SOX-relevant control documentation are gating items in enterprise, and assembling them reactively adds weeks. If you issue cards, KYC and KYB obligations apply to your customers' entities, and a clumsy onboarding flow there will kill deals that were already won.

Pricing per seat in a zero-software-cost market. If your competitive set includes interchange-funded platforms and you have no card story, you must either move upmarket into segments where card issuance is impractical, or differentiate on something the interchange players genuinely do not do — multi-entity consolidation, VAT reclaim, grant-fund accounting, project-level billable allocation.

Ignoring the employee as a user. The economic buyer is finance, but the daily user is every employee who submits an expense. Low mobile adoption sinks the deployment, the controller ends up chasing the same people they chased before, and the renewal conversation goes badly. Measure submission rate within 48 hours of transaction as a health metric and intervene when it drops.

Selling expense as a standalone forever. A single-module footprint caps expansion. Plan the attach sequence into the initial deal narrative even if you sell expense alone at first.

Operating model and cadence that keeps the motion honest

Team build sequence. The first five hires: founder-led selling until roughly the first million in recurring revenue, then a lead enterprise AE recruited out of an incumbent in this category (they arrive with the buyer map), a customer success lead who has actually been a controller, a solutions engineer who owns ERP integration and the sandbox, and a product marketer with a real finance-community network. Hires six through fifteen add enterprise and mid-market AE capacity, SDRs targeting CFO and controller titles, a partner manager for the accounting-firm channel, implementation managers, and — if you issue cards — a card-network and sponsor-bank partnership lead plus a machine-learning engineer owning coding accuracy. Beyond that: a VP of sales, a VP of customer success, regional leadership if you go international, and a compliance officer with genuine authority once card issuance is material.

Weekly cadence. Monday enterprise pipeline review, where every deal in a sandbox reports its current auto-coding accuracy number — that single metric predicts close probability better than stage. Midweek sandbox review with the solutions team on which accounts need a tuning pass. End-of-week partner-channel touch log, because the accounting-firm channel decays without deliberate contact.

Monthly cadence. Module-attach review: what percentage of the base runs expense only versus expense plus adjacent modules, and which accounts are attach-ready. Coding-accuracy scorecard per live customer, with anything below your threshold triggering a re-tune ticket rather than a support ticket. Implementation-hours-per-deal review. Renewal-risk board driven by usage signals — submission rate, admin login frequency, and integration error volume — not by relationship sentiment.

Quarterly cadence. A customer advisory council of controllers, run as a working session on close-process pain rather than a roadmap presentation. Analyst and community briefings. If you issue cards, a partnership health review with your issuing and network partners. A model retraining cycle with a documented before-and-after accuracy comparison.

The expansion loop. Land expense, then attach in a deliberate sequence: AP automation once the first close cycle is clean (roughly month four), procurement or vendor management once spend data has accumulated enough to make classification useful (roughly month nine), travel where the customer's travel volume justifies it (roughly month twelve), and treasury or working-capital products last. Each attach should be triggered by a usage signal, not a calendar reminder — pitching AP automation to a customer whose expense deployment is still struggling burns credibility you need later.

What compounds. The durable advantages in this market are integration breadth, coding accuracy trained on a large corpus of real general-ledger decisions, and the accounting-firm channel. All three get better with scale and none can be bought quickly, which is why the operating cadence above prioritizes measuring them weekly rather than reviewing them annually.

Related questions

How long should the sandbox stage take?

Thirty days is the right default: long enough for a tuning pass against real general ledger data, short enough that the deal does not lose momentum. Cap concurrent sandboxes per solutions engineer at roughly three, and require a named controller-side owner and a scheduled readout before starting one.

Should a new entrant issue cards or stay software-only?

Issue if you can absorb PCI scope, KYC and KYB obligations, sponsor-bank relationships, and fraud reserve — interchange is the revenue model competitors are pricing against. Stay software-only if you target regulated entities, complex multi-entity multinationals, or organizations with entrenched banking contracts.

Which ERP integrations should ship first?

Sequence by ICP: QuickBooks and Xero for SMB, NetSuite and Sage Intacct for mid-market, Workday Financials, Dynamics 365 Finance, and Oracle Cloud ERP for enterprise. Certified marketplace listings carry more weight with finance systems owners than raw API availability.

What is the earliest reliable signal that a deal will close?

Auto-coding accuracy against the buyer's own historical data during the sandbox. It predicts close probability better than pipeline stage, because it is the number the controller uses internally to justify the change.

When does the accounting-firm channel start paying off?

Typically two to three quarters after you build a real partner console and revenue share — one regional firm standardizing on your platform can deliver dozens of client accounts at near-zero incremental acquisition cost, with the firm's staff doing much of the implementation.

FAQ

Who is the real decision-maker on an expense platform purchase?

The controller. The CFO is the economic approver and owns the narrative internally, but the controller runs month-end close, owns the general ledger coding conventions, and lives in the product daily. A deal that has CFO enthusiasm and controller skepticism does not close; the reverse frequently does.

Why does the spend-data sandbox compress the cycle?

Because it replaces an opinion contest with arithmetic on the buyer's own data. Every vendor claims accurate automated coding; only one shows the actual percentage against this customer's historical general ledger, names the categories where it underperforms, and demonstrates a tuning pass that improves them.

How do you compete against a deeply entrenched legacy incumbent?

On implementation speed, mobile submission experience, coding automation, and total cost including card economics. Time the outreach to the renewal window, and target the controller who inherited the incumbent rather than the executive who chose it. Do not lead with feature comparison charts — lead with the sandbox.

What is a realistic implementation timeline?

Two to eight weeks for mid-market with a native ERP connector and a reasonably clean chart of accounts. Enterprise with multiple entities, currencies, and a custom coding structure runs longer. If your median routinely exceeds this, the cause is usually incomplete connectors or under-scoped discovery, and it will show up as CAC payback drift.

How should module expansion be sequenced after go-live?

AP automation once the first close cycle runs clean, procurement once enough spend data has accumulated to make vendor classification useful, travel where volume justifies it, and treasury products last. Trigger each attach on a usage signal rather than a calendar date.

What compliance work must be finished before selling enterprise?

SOC 2 Type II, ISO 27001 where you sell internationally, PCI DSS attestation if you touch card data, documented SOX-relevant controls, and GDPR and data-residency answers. Assemble these before the first enterprise conversation — reactively answering a security questionnaire adds weeks to an already long cycle.

Sources

• https://www.gartner.com/en/finance
• https://www.pcisecuritystandards.org/document_library/
• https://www.aicpa-cima.com/topic/audit-assurance/audit-and-assurance-greater-than-soc-2
• https://www.federalreserve.gov/paymentsystems/regii-interchange-fee-standards.htm
• https://www.afponline.org/
• https://www.iso.org/standard/27001
• https://www.sec.gov/spotlight/sarbanes-oxley.htm
• https://gdpr.eu/
• https://www.g2.com/categories/expense-management
• https://www.netsuite.com/portal/products/erp.shtml

Related on PULSE

• [How do you build a forestry management software go-to-market motion in 2027?](/knowledge/gp0122)
• [How do you build a livestock management software go-to-market motion in 2027?](/knowledge/gp0120)
• [How do you build an AgTech crop management (Climate FieldView / Granular) go-to-market motion in 2027?](/knowledge/gp0119)
• [How do you build a hospital revenue cycle management (Epic / Cerner adjacent) go-to-market motion in 2027?](/knowledge/gp0113)
• [How do you build an EAM and CMMS (enterprise asset management) software go-to-market motion in 2027?](/knowledge/gp0090)
• [How do you build a construction tendering and bid management software go-to-market motion in 2027?](/knowledge/gp0089)