Salesforce — Funding the Next Layer of the Customer Platform
Download case PDF (5 pages)Request teacher’s pack
Decision date and information cutoff: 29 November 2023, 4:05:29 p.m. Eastern Time
Decision scope: Hypothetical FY2025 engineering and partner-enablement capacity
Case convention: This is an author-created decision exercise. Salesforce had not disclosed this allocation, a dedicated budget, or these alternatives. Company statements are attributed; no private deliberations or later outcomes are represented.
The decision
At 4:05 p.m. on November 29, Salesforce made its third-quarter FY2024 results public. Revenue had grown 11% year over year, subscription and support revenue 13%, and current remaining performance obligation 14%. The company also reported a 17.2% GAAP operating margin and a 31.2% non-GAAP margin. The figures offered a stronger operating backdrop than a year earlier, but none identified the economics of a particular product. [S1]
The next challenge was less visible in the release. Salesforce had announced Einstein 1, a platform bringing Data Cloud, Einstein AI and CRM applications together on a metadata framework. It had also announced an expanded partnership with Google, including intended connections between Workspace and Salesforce. These moves suggested a richer, more integrated customer platform. Whether customers would adopt it broadly, pay more for it, renew because of it, or use it to reduce costs was not yet established in the public record. [S2–S3]
Salesforce had another credible claim on scarce engineering capacity: the Sales and Service products that supported its installed base and core workflows. Microsoft reported 28% year-over-year growth in Dynamics 365 for its quarter ended September 30, 2023. That rate used a different fiscal period and product scope, but it underscored competitive momentum. [S4]
Assignment: You advise Salesforce on a hypothetical FY2025 allocation of 100 incremental capacity points between two uses: (1) core Sales/Service reliability, workflow and customer success; and (2) Data Cloud/Einstein 1 and partner integrations. One point represents one percent of this invented capacity pool; it is not a headcount, dollar amount, Salesforce budget, or disclosed management plan. State how many points go to each use. For the platform allocation, specify how many points to release initially and how many to hold behind customer and delivery milestones. A staged release is a way to govern platform points, not a third use. Recommend what evidence should trigger the next release or a change in direction. Use only information available by the cutoff above. The scheduled 5:00 p.m. ET earnings call is outside the case boundary.
What Salesforce sells and how it reports it
Salesforce sells subscription-based customer relationship management and related software. Its product portfolio spans Sales, Service, Platform and Other, Marketing and Commerce, and Data offerings. Products can overlap in features, and the company records offering revenue according to the products customers order, not a complete measurement of which application or feature each customer uses. [S5], MD&A, printed pp. 39–40.
FY2023 revenue was $31.352 billion, of which $29.021 billion, or 93%, came from subscription and support. [S5], MD&A, “Sources of Revenues,” printed p. 45. The service-offering table showed Sales revenue of $6.831 billion and Service revenue of $7.369 billion, each up 14%; Platform and Other was $5.967 billion, up 32%; Marketing and Commerce was $4.516 billion, up 16%; and Data was $4.338 billion, up 15%. Salesforce defined the Data category as Analytics and Integration. It was not a separately reported measure of the newer Data Cloud product. Platform and Other included Slack. [S5], “Subscription and Support Revenues by Service Offering,” printed p. 46.
The FY2023 filing reported $5.055 billion in research and development expense, equal to 16% of revenue. The amount included employee costs, stock-based compensation and allocated overhead across the company. Salesforce expected R&D expense to rise in absolute dollars in the near term, but disclosed no product-level allocation or incremental FY2025 pool. [S5], MD&A, “Operating Expenses,” printed pp. 47–48.
Exhibit 1. Selected company results
USD billions except growth and margins; reported periods and categories.
| Measure | FY2023 | FY2024 Q3 |
|---|---|---|
| Revenue | 31.352 | 8.72 |
| Revenue growth, year over year | — | 11% (10% constant currency) |
| Subscription and support revenue | 29.021 | 8.14 |
| Subscription and support growth | 18% | 13% |
| Services and other revenue | 2.331 | 0.58 |
| GAAP operating margin | — | 17.2% |
| Non-GAAP operating margin | — | 31.2% |
| Current remaining performance obligation | — | 23.9 (+14%; +13% constant currency) |
| Research and development expense | 5.055 | — |
Sources: FY2023 Form 10-K and FY2024 Q3 earnings release. [S1], [S5]. FY2023 category percentages and Q3 values are company-reported. Non-GAAP margin excludes stock-based compensation, acquired-intangible amortization and restructuring. Current remaining performance obligation is contracted, unrecognized revenue expected within twelve months; it is not revenue or bookings, and the company cautions that RPO is not necessarily indicative of future growth. [S5], Note 2. These company-wide measures do not establish Data Cloud/AI returns or the available capacity pool.
In the Q3 release, Salesforce narrowed its FY2024 revenue guidance and raised full-year operating-margin and operating-cash-flow guidance. It also reported $1.9 billion of share repurchases through the quarter. Repurchases were another use of capital, but neither the guidance nor the buybacks disclosed what product teams should receive. [S1]
The platform proposition
Salesforce described Einstein 1 as bringing Data Cloud, Einstein AI and CRM applications together over a metadata framework. It said Data Cloud could harmonize structured and unstructured customer information to create a unified view, with AI and automation available within customer workflows. This was Salesforce's product and intended-benefit description, not independent evidence of adoption, accuracy, productivity or customer economics. [S2]
The Google announcement proposed two-way connections between Salesforce and Workspace and use of Google-hosted models through Salesforce's Copilot Studio and Trust Layer. The announcement made partner coordination central to the product thesis. It did not show that every announced integration had shipped, how many customers would use it, or how the partners would divide costs and value. [S3]
The platform-led argument was that connected data, CRM workflows, AI and ecosystem integrations might reinforce one another. A customer could gain more from a model when it had relevant, permissioned information and could act on the result inside a workflow. Salesforce could potentially make more of its portfolio useful together. Each step in that chain required execution, adoption and customer value that the available disclosures did not quantify.
The core-product claim
Sales and Service together represented $14.2 billion in FY2023 revenue, and each offering grew 14%. These were meaningful established businesses. Reliability, workflow depth, implementation and customer-success work could support retention and expansion. Yet public filings did not isolate how much engineering would sustain core quality, which customer frictions mattered most, or the marginal return on another increment of core capacity. [S5]
Microsoft's Dynamics 365 growth provided a competitive signal, not a clean comparison. Microsoft reported 28% growth for the quarter ended September 30, 2023 (26% in constant currency), versus Salesforce's Q3 results for the quarter ended October 31. The product category, currency treatment and period differed. The data could motivate defensive investment, but could not establish that Salesforce was losing particular customers or share. [S1], [S4]
What the research can and cannot answer
Studies of IT investment suggest that returns depend on the kind of asset, the intended outcome and complementary organizational capabilities. One panel study found different associations for informational, strategic, transactional and infrastructure investment; its infrastructure result varied substantially with measured IT capability. Research on CRM similarly points to customer knowledge and information sharing as mechanisms between software and customer satisfaction. A study of information-management capability links data capability to customer, process and performance-management capabilities. [S6–S8]
These studies offer ways to frame a test. They do not estimate Salesforce's AI return, Data Cloud adoption, retention lift or customer-specific economics. They are observational, use older settings and different measures, and cannot supply a forecast coefficient for this decision. Company disclosures at the cutoff likewise offered no isolated product revenue, margin, attach, renewal effect, AI inference cost, or trial-to-production conversion. A disciplined recommendation must make the missing evidence part of its design.
Allocation approaches for discussion
- Platform-heavy allocation. Assign most points to Data Cloud, Einstein 1 and partner integrations. State which customer use cases, data-readiness requirements and partner commitments justify the allocation, how many platform points to release first, and how core service levels will be protected.
- Core-heavy allocation. Assign most points to Sales/Service reliability, workflow and customer-success needs. Define how the company will learn which frictions affect adoption, renewals or expansion, and how many platform points remain available for learning.
- Staged platform release. Choose any explicit core/platform split, protect a core floor, then release platform points in tranches against customer-use, willingness-to-pay, delivery, cost-to-serve and trust milestones. Specify milestones and a review date; do not assume pilots will prove causality by themselves.
These are classroom approaches, not actual management proposals. Every recommendation must allocate all 100 points between core and platform uses; staging changes when platform points are released, not where they are allocated. The point pool is an allocation aid, not a forecast. Students may combine or refine the approaches if the split and release plan remain explicit and feasible.
Questions
- What must be true for Data Cloud and Einstein 1 to create customer and company value? Which required complements are within Salesforce's control, and which depend on customers or partners?
- How many of the 100 points should go to core work and how many to platform work? For the platform allocation, how many points should be released initially and what should gate later tranches? What is the strongest argument against your choice?
- Which leading indicators would justify increasing, holding or redirecting platform investment? Separate product use, customer value, commercial conversion, cost-to-serve and trust.
- How much should Microsoft's reported Dynamics growth affect the choice, given the differences in timing and scope?
- What evidence is missing, and how would you collect it without mistaking correlation or pilot enthusiasm for durable economics?
Prepare a concise decision memo with an allocation, strategic rationale, implementation sequence, milestones, risks and conditions that would change your recommendation.
Sources and conventions
S1. Salesforce, FY2024 Q3 Form 8-K and Exhibit 99.1, accepted 29 November 2023 at 4:05:29 p.m. ET. SEC filing index; earnings release. Release, financial tables, guidance and non-GAAP reconciliation. Cutoff excludes the 5:00 p.m. ET call.
S2. Salesforce, “Salesforce Announces the New Einstein 1 Platform,” 12 September 2023. Issuer announcement. Product positioning and intended architecture, attributed to issuer.
S3. Salesforce and Google, “Salesforce and Google Expand Partnership…,” 12 September 2023. Joint announcement. Announced intended integrations and model access; not proof of delivery or economics.
S4. Microsoft, FY2024 Q1 results, 24 October 2023. Issuer release. Dynamics 365 growth for quarter ended 30 September; different scope and period.
S5. Salesforce, FY2023 Form 10-K, accepted 8 March 2023. SEC filing. Revenue discussion, MD&A “Sources of Revenues,” printed p. 45; service-offering table and Data-category definition, “Subscription and Support Revenues by Service Offering,” p. 46; R&D expense and expectations, MD&A “Operating Expenses,” pp. 47–48; product-ordering revenue-recognition statement, MD&A, pp. 39–40; RPO definition and limitations, Note 2.
S6. Aral, S. & Weill, P. (2007), “IT Assets, Organizational Capabilities, and Firm Performance,” Organization Science 18(5), 763–780. DOI. Observational panel; asset-outcome associations vary with capabilities; not transferable Salesforce returns.
S7. Mithas, S., Krishnan, M.S. & Fornell, C. (2005), “Why Do Customer Relationship Management Applications Affect Customer Satisfaction?” Journal of Marketing 69(4), 201–209. DOI. Cross-sectional CRM/customer-knowledge and satisfaction evidence; limited sample and nonrandomized design.
S8. Mithas, S., Ramasubbu, N. & Sambamurthy, V. (2011), “How Information Management Capability Influences Firm Performance,” MIS Quarterly 35(1), 237–256. Article. Observational capability relationships in one large business group; not product-level causal returns.
Historical disclosures, issuer claims, academic evidence and invented classroom assumptions are kept distinct. No post-cutoff company outcome or earnings-call statement is included.