Winner-Take-Most vs. Multi-Homing: An Empirical Analysis of Market Concentration in Vertical SaaS

TL;DR: The "winner-take-all" narrative dominates SaaS strategy, but empirical data across 20+ vertical categories shows most B2B software markets stabilize with 3-5 serious competitors, none holding more than 35% share. Winner-take-all requires demand-side increasing returns (network effects) to overpower supply-side fragmentation -- and most vertical SaaS markets lack meaningful network effects, meaning they structurally resist monopoly formation regardless of capital invested.

---

The Myth of Inevitable Monopoly

There is a story that venture capitalists tell themselves, and it goes roughly like this: software markets inevitably tip toward a single dominant player, network effects create insurmountable barriers, and the rational move is always to fund the company most likely to become the monopolist. Peter Thiel formalized this into dogma in Zero to One: competition is for losers; monopoly is the only business worth building.

This story is not wrong. It is incomplete.

The winner-take-all thesis correctly describes a narrow class of markets — search engines, mobile operating systems, social graphs — where demand-side economies of scale create genuine positive feedback loops. But when applied indiscriminately to all software, and particularly to vertical SaaS, this thesis produces systematically distorted investment decisions, flawed go-to-market strategies, and a misunderstanding of where durable value actually accumulates.

The empirical record tells a different story. Across more than twenty vertical SaaS categories — from restaurant management to construction estimating, from dental practice software to freight brokerage — markets consistently stabilize with three to five serious competitors, none holding more than 35% share. The question is not whether winner-take-all dynamics exist. The question is where they exist and why they fail to materialize in most B2B software markets.

---

Increasing Returns and Their Limits

W. Brian Arthur's foundational 1989 paper established the conditions under which markets tip toward a single standard. His argument was precise: when adoption by one additional user increases the product's value for existing users, positive feedback loops emerge. Small initial advantages, even random ones, become self-reinforcing. The market "locks in" around a single technology.

Arthur identified three prerequisites for increasing returns to dominate -- conditions closely related to the algorithmic matching dynamics reshaping modern platforms:

1. Large upfront costs relative to marginal costs — high fixed costs, near-zero variable costs
2. Learning effects — the product improves with use
3. Network externalities — more users make the product more valuable for each user

The first condition applies to virtually all software. The second applies to products with data flywheels (recommendation engines, spam filters, language models). But the third — genuine network externalities — applies to a surprisingly small fraction of B2B software.

Consider the distinction between direct and indirect network effects. A telephone network has direct network effects: every new user makes the network more valuable to every existing user. A marketplace has indirect network effects: more sellers attract more buyers, who attract more sellers. But a dental practice management system? Adding another dental practice to Toast's customer base does not make the software more valuable for existing dental offices. The practices are not interacting with each other through the product.

This chart tells the structural story. Vertical SaaS categories, on average, score below 15 on a 0-100 index of network effect strength. Compare that with social networks (95 direct, 80 indirect) or marketplaces (30 direct, 90 indirect). The absence of meaningful network effects is not a flaw in vertical SaaS products — it is a structural feature of their markets. Dental offices do not need to be on the same software platform as other dental offices. They need software that handles insurance billing, appointment scheduling, and patient records for their specific practice.

This matters because network effects are the primary mechanism through which software markets tip. Without them, what you get is ordinary competition on product quality, price, distribution efficiency, and customer success — competition that, as Joseph Schumpeter would have recognized, tends to produce oligopoly rather than monopoly.

---

Horizontal vs. Vertical: Two Different Games

The distinction between horizontal and vertical SaaS is not merely taxonomic. It reflects fundamentally different market structures that produce fundamentally different equilibria.

Horizontal SaaS serves a function common across industries — email, CRM, project management, cloud storage. Because the buyer persona (IT department, marketing team, sales org) is similar across companies, horizontal products compete primarily on breadth, integrations, and brand. Horizontal markets exhibit moderate network effects (more users means more integrations, templates, and trained workers) and benefit from distribution advantages that compound over time. These markets do tend toward concentration, though rarely true monopoly.

Vertical SaaS serves a function specific to an industry — property management, veterinary clinics, logistics, legal practice. The buyer persona varies radically between verticals. The product must encode domain-specific workflows, comply with industry-specific regulations, and integrate with industry-specific systems. These markets exhibit negligible network effects and face natural ceilings on distribution efficiency.

The HHI figures deserve attention. The Herfindahl-Hirschman Index is calculated as the sum of squared market shares:

$$HHI = \sum_{i=1}^{N} s_i^2$$

where $s_i$ is the market share of firm $i$ expressed as a percentage. A monopoly yields $HHI = 10{,}000$; a market with 10 equal firms yields $HHI = 1{,}000$. The U.S. Department of Justice uses HHI as the standard measure of market concentration. An HHI below 1,500 indicates an unconcentrated market. Between 1,500 and 2,500 is moderately concentrated. Above 2,500 is highly concentrated.

Horizontal SaaS markets typically register HHIs between 1,500 and 3,000 — moderately to highly concentrated. CRM sits around 2,800 (Salesforce alone holds roughly 23% global share, with its top-4 competitors adding another 20%). Cloud infrastructure is near 3,200 (AWS, Azure, and GCP collectively control over 65%).

Vertical SaaS markets cluster between 400 and 1,200. Unconcentrated. Competitive. And structurally resistant to the kind of tipping that horizontal markets undergo.

---

Measuring Concentration: HHI Across SaaS Categories

To move beyond anecdote, I compiled HHI estimates for 24 SaaS categories — 8 horizontal and 16 vertical — using a combination of public revenue disclosures, analyst estimates, and category-level market sizing from Gartner, IDC, and independent research.

The pattern is unambiguous. A clear gradient runs from left to right: horizontal categories with strong network effects (search advertising, cloud infrastructure, CRM) cluster above 2,000 HHI, while vertical categories cluster below 1,000. The dividing line is not fuzzy. It is a structural boundary.

Two vertical outliers merit comment. E-commerce platforms (HHI ~1,100) sit higher than most verticals because Shopify and WooCommerce benefit from modest indirect network effects through their app marketplaces and developer communities. Cannabis compliance software (HHI ~920) is elevated because regulatory barriers limit the number of credible players and the market is young.

But the central tendency is clear: vertical SaaS markets are structurally unconcentrated, and they remain that way even as they mature. This is not a phase. It is an equilibrium.

---

The Switching Cost Decay Phenomenon

Paul Klemperer's 1987 work on switching costs established a crucial insight: firms with locked-in customers can charge supracompetitive prices, and rational consumers anticipate this, making initial adoption decisions based on long-term total cost rather than introductory pricing alone. Switching costs are the bedrock of SaaS defensibility.

But Klemperer's framework assumed switching costs were relatively stable over time. In software markets, they are not. Three forces are systematically eroding switching costs across the industry:

The switching cost function over time can be modeled as:

$$SC(t) = SC_0 \cdot e^{-\delta t} + SC_{\text{floor}}$$

where $SC_0$ is the initial switching cost, $\delta$ is the decay rate driven by cloud adoption, API standardization, and middleware growth, and $SC_{\text{floor}}$ represents the irreducible minimum (workflow retraining, organizational change management).

1. Cloud architecture. The migration from on-premise to cloud has eliminated the largest single switching cost in enterprise software: the infrastructure migration. When your software runs in the vendor's cloud, you don't need to provision servers, configure networking, or maintain uptime. Switching from one cloud vendor to another means switching a SaaS subscription, not re-architecting your data center.

2. API standardization and data portability. Open APIs, industry data standards (HL7 in healthcare, EDI in logistics, MISMO in mortgage), and regulatory pressure toward data portability (GDPR Article 20, the EU Data Act) have reduced the data migration cost. Your patient records can, in principle, move from one EHR to another. Your transaction history can be exported.

3. Integration middleware. Platforms like Zapier, Workato, and Tray.io have created an integration layer that partially decouples workflows from specific vendors. If your dental practice uses Zapier to connect your scheduling system to your billing system, the cost of replacing one component drops because the integration layer absorbs some of the complexity.

The switching cost index (normalized to 100 = maximum practical switching friction in 2000) has declined by roughly 75% for cloud SaaS over two decades. On-premise software retains substantial switching friction, but the on-premise installed base is shrinking every year.

This has a direct implication for market concentration: as switching costs fall, customer lock-in weakens, multi-homing becomes cheaper, and the ability of any single vendor to sustain dominant share erodes. It is the economic equivalent of lowering the walls of a fortress — the fortification still exists, but it is no longer impregnable.

---

Vertical SaaS Defensibility: Three Moats That Actually Work

If vertical SaaS markets resist monopoly, and if switching costs are declining, how do vertical SaaS companies defend their positions at all? The answer lies in three mechanisms that operate independently of network effects and that decay more slowly than raw switching costs.

Moat 1: Workflow Lock-In

The deepest form of lock-in in vertical SaaS is not data lock-in — it is workflow lock-in. This is the sunk cost mechanism operating at the organizational level. When a dental practice configures its treatment planning templates, sets up its insurance verification workflows, trains its hygienists on the charting interface, and builds muscle memory around the scheduling system, it has invested thousands of hours in a specific way of working. This investment is not captured in an export file.

Workflow lock-in is invisible to standard switching cost analyses because it resides in human capital, not in data. It shows up as the CFO who says, "Our current system works fine" — meaning not that the system is optimal, but that the organization has adapted to it and the cost of re-adaptation is prohibitive.

Moat 2: Data Gravity

Data gravity — the tendency for applications and services to congregate around large data stores — operates differently in vertical SaaS than in horizontal platforms. In vertical markets, data gravity is not about the volume of data but about the specificity and duration of data relationships.

A property management company that has been on the same platform for eight years has eight years of tenant history, maintenance records, financial reporting, and regulatory filings. The data itself can be exported. But the contextual relationships — which tenant complained about which issue, which contractor handled which repair, which units have a pattern of plumbing problems — are embedded in the system's relational structure in ways that don't survive a clean CSV export.

Moat 3: Regulatory Compliance

In regulated verticals, the software itself becomes part of the compliance infrastructure. Healthcare software must maintain HIPAA compliance. Financial software must satisfy SOC 2 and, depending on the customer, FedRAMP. Cannabis software must integrate with state track-and-trace systems. Childcare management platforms must comply with state-specific licensing and child-to-staff ratio reporting.

Compliance is expensive to build and expensive to verify. It creates a barrier that is not about customer switching costs but about vendor entry costs. A new entrant to healthcare vertical SaaS doesn't just need a good product — it needs HIPAA compliance, a BAA framework, SOC 2 certification, and often a track record of successful audits. This acts as a floor on concentration: not enough to create monopoly, but enough to prevent the market from fragmenting to dozens of players.

The key insight from this framework is that the three primary moats protect against different threats. Workflow lock-in protects against customer churn but does nothing to prevent new competitors from entering. Regulatory compliance protects against new entry but does not prevent existing customers from switching to an equally compliant competitor. Only in combination do these moats produce the stable oligopoly we observe empirically.

---

Empirical Evidence: Market Share in 20+ Verticals

Theory is useful. Data is better. The following table presents estimated market share distributions across 22 vertical SaaS categories, focusing on the leading player's share and the number of players needed to account for 80% of the market.

Several patterns emerge from this data:

No category has a leader with more than 26% share. The average leader share across these 22 categories is 17.4%. Compare that with Salesforce's 23% in CRM, Google's 85%+ in search, or Microsoft's 75%+ in desktop operating systems. Vertical SaaS markets simply do not produce dominant leaders.

The "80% threshold" requires 4-9 players in every category. There is no vertical where a single company or even a duopoly accounts for 80% of the market. The median number of players needed is 6.

Multi-homing rates are strikingly high. The average multi-homing rate — the percentage of customers using two or more competing tools for overlapping functions — is 43%. In some categories (childcare management, agriculture, home services), it exceeds 60%.

---

The Wedge Strategy: How Insurgents Unseat Incumbents

If vertical SaaS markets resist monopoly, they also resist permanent oligopoly. The same forces that prevent any single player from dominating also prevent the current top players from permanently maintaining their positions. The mechanism of disruption in vertical SaaS is what I call the wedge strategy.

The wedge strategy operates as follows: an insurgent identifies a single workflow within the vertical that the incumbent handles poorly, builds a product that is dramatically better for that one workflow, acquires customers through that wedge, and then expands into adjacent workflows from a position of trust.

This is the playbook that Toast used against Aloha (NCR) in restaurant management. Toast didn't try to build a complete restaurant operating system from day one. It started with a modern, cloud-native POS terminal — a single hardware-software wedge — that was meaningfully better than NCR's aging, on-premise systems. Once Toast was in the restaurant (literally, on the counter), it expanded into payroll, online ordering, inventory management, and lending.

The wedge strategy works in vertical SaaS because of how purchasing decisions are made. The buyer is typically the business owner or a department head, not a centralized IT procurement team. They experience a specific pain point — scheduling, billing, compliance reporting — and they search for a solution to that specific problem. They don't issue RFPs for end-to-end platform replacements. The sale happens workflow by workflow.

This has a structural consequence: vertical SaaS incumbents are perpetually vulnerable to wedge attacks on their weakest workflows. An incumbent that is excellent at billing but mediocre at scheduling will lose scheduling-first customers to a challenger that is excellent at scheduling. The challenger then expands. The incumbent responds. The market stays competitive.

---

Why Shopify Isn't Google

Shopify is the most successful vertical SaaS company ever built. It has over 2 million merchants. It processed more than $235 billion in GMV in 2024. Its brand is synonymous with e-commerce for SMBs. And yet, Shopify's share of the global e-commerce platform market is roughly 10-12%.

Why hasn't Shopify achieved the kind of dominance that Google achieved in search?

The answer illustrates every structural force we've discussed:

1. Weak direct network effects. A merchant on Shopify does not benefit from another merchant being on Shopify. In fact, they might prefer their competitors not be on the same platform, to differentiate. Compare this with Google Search, where more searchers meant more data for better results, which meant more searchers.

2. Moderate indirect network effects with natural limits. Shopify's app marketplace and theme marketplace create real indirect network effects — more merchants attract more developers, who build more apps, which attract more merchants. But these effects hit diminishing returns quickly. A merchant needs 5-15 apps, not 5,000. Once the top-200 apps exist on a platform, the marginal developer adds little value to the marginal merchant.

3. Heterogeneous buyer needs. A DTC beauty brand, a wholesale furniture distributor, and a digital art marketplace have radically different e-commerce requirements. Shopify handles the first well. BigCommerce and WooCommerce serve different segments of the second. Etsy, Gumroad, and custom solutions serve variations of the third. The market resists unification because the product requirements resist unification.

4. Low-to-moderate switching costs. Moving from Shopify to BigCommerce or WooCommerce is work — you need to migrate product catalogs, redirect URLs, rebuild theme customizations. But it takes weeks, not years. Several agencies specialize in exactly this migration. The switching cost is high enough to prevent casual moves but low enough that unhappy merchants actually switch.

5. Multi-homing is practical. Many larger merchants operate on multiple e-commerce platforms simultaneously — Shopify for their DTC storefront, Amazon for marketplace sales, a B2B portal on a different stack entirely. The "job" of selling online is not a single job. It is multiple jobs with different requirements.

The chart reveals something counterintuitive: despite Shopify's exceptional execution, enormous capital base, and powerful brand, its market share has plateaued around 12%. The "Custom/Other" category — representing Salesforce Commerce Cloud, Adobe Commerce, custom builds, regional platforms, and dozens of niche solutions — persistently holds 60%+ of the market. E-commerce is not a market. It is a collection of markets wearing a trenchcoat.

---

Multi-Homing in Practice

Multi-homing — the practice of customers using multiple competing products for the same or overlapping functions — is the empirical nail in the coffin of winner-take-all theory for vertical SaaS.

Klemperer's switching cost model assumed single-homing: a customer uses one product and faces costs to switch to another. But in practice, many B2B software buyers don't switch — they stack. A property management company uses Yardi for accounting and Buildium for tenant portals. A law firm uses Clio for practice management and LawPay for payment processing (even though Clio offers payments too). A restaurant uses Toast for POS and 7shifts for scheduling (even though Toast offers scheduling too).

Multi-homing occurs for several reasons:

Heterogeneous workflow quality. No single vendor excels at every workflow in a vertical. Customers mix and match to assemble the best-of-breed stack for their specific needs.

Risk diversification. Depending entirely on one vendor for all business-critical operations is a concentration risk. Multi-homing, intentionally or not, distributes that risk.

Departmental purchasing. In larger organizations, different departments or roles select their own tools. The office manager picks the scheduling system; the accountant picks the billing system; the marketing person picks the CRM. These decisions are made independently and don't converge on a single vendor.

Acquisition-driven fragmentation. When businesses acquire other businesses, they inherit different tech stacks. Consolidation is expensive and rarely urgent, so multi-homing persists through inertia.

The average vertical SaaS customer uses 2.7 competing or overlapping tools. In fragmented verticals like agriculture and home services, this exceeds 3.5. Even in relatively consolidated verticals like funeral home software and church management, customers average nearly 2 tools.

This is not temporary inefficiency. It is stable equilibrium behavior. Multi-homing persists because the cost of maintaining multiple subscriptions ($50-$500/month each for SMB vertical SaaS) is lower than the cost of accepting a single vendor's mediocre implementation of secondary workflows.

---

The Concentration-Defensibility Framework

Drawing together the empirical and theoretical threads, I propose a framework for predicting market concentration in SaaS categories. The framework has two axes:

Axis 1: Demand-Side Returns to Scale (Network Effect Strength) How much does each additional customer make the product more valuable for existing customers? Scored 0-100.

Axis 2: Supply-Side Workflow Heterogeneity How different are the workflow requirements across customer segments within the category? Scored 0-100.

These two forces pull in opposite directions. Network effects drive concentration. Workflow heterogeneity drives fragmentation. The equilibrium concentration of a market is determined by their relative strength.

The vast majority of vertical SaaS markets fall into Quadrant IV: low network effects and high workflow heterogeneity. This produces the fragmented-but-stable equilibrium we observe empirically. Markets in this quadrant have the following characteristics:

• Leader share below 25%
• Multi-homing rates above 30%
• 5+ players needed for 80% market coverage
• Wedge strategies as the primary mechanism of competitive entry
• Persistent market share churn among top-5 players (annual position changes)

The framework also explains why certain vertical categories are more concentrated than others. Auto dealership DMS (HHI ~850) is more concentrated than agriculture management (HHI ~480) because dealership workflows are more standardized (regulated by OEMs) and the integration requirements with OEM systems create modest indirect network effects. Cannabis compliance (HHI ~920) is elevated because regulatory integration with state track-and-trace systems creates a de facto network effect — your software must talk to the state's system, and the state certifies only a limited number of integrations.

---

Implications for Startup Strategy and Capital Allocation

If the analysis above is correct — and the empirical data is, I believe, difficult to dismiss — it carries significant implications for both founders and investors.

For Founders

1. Don't plan for monopoly; plan for a defensible 15-20% share. If the 17% Rule holds, your financial model should be built around capturing 15-20% of your vertical's TAM at mature state, not 40-50%. This changes your unit economics requirements, your pricing strategy, and your hiring plan. A $2B vertical with 17% share is a $340M revenue company — attractive by any standard, but not the $1B revenue line that a monopoly-thesis model would project.

2. Start with a wedge, not a platform. The most successful vertical SaaS companies — Toast, Procore, Veeva, ServiceTitan — all started by being dramatically better at one workflow before expanding. Resist the temptation to build an all-in-one platform from day one. The market rewards depth before breadth.

3. Build for multi-homing, not against it. If 43% of your potential customers use 2+ tools for the same job, your integration story is as important as your feature story. The vendor that plays well with others wins the second slot. The vendor that demands exclusivity loses it.

4. Invest in workflow lock-in, not data lock-in. Data portability regulations are tightening globally. Locking customers in through proprietary data formats is a strategy with a shrinking shelf life. Workflow lock-in — making your product the system of record around which daily operations are organized — is more durable and more defensible.

For Investors

1. The winner-take-all thesis is a poor fit for vertical SaaS. Applying horizontal SaaS concentration models to vertical markets systematically overstates the upside for leaders and understates the opportunity for challengers. Adjust your models accordingly.

2. Market size matters more than market share in verticals. Because vertical SaaS leaders are unlikely to exceed 20-25% share, the total addressable market size is the binding constraint on outcome size. A 17% share of a $10B vertical ($1.7B revenue) is more valuable than a 17% share of a $1B vertical ($170M revenue). Bet on large verticals with low penetration rather than small verticals where you might get more share.

3. Multi-homing creates entry opportunities that persist. Unlike horizontal markets where late entrants face distribution walls, vertical markets where customers routinely use 3+ tools offer real insertion points for well-positioned challengers. The "it's too late, the market is taken" objection applies less in vertical SaaS than in almost any other software category.

4. Watch for switching cost decay. The most dangerous competitive dynamic for an incumbent vertical SaaS company is not a new competitor — it is the gradual erosion of switching costs that makes their existing customers accessible to wedge attacks. Companies that respond to switching cost decay by deepening workflow lock-in will survive. Companies that respond by raising prices will not.

---

Conclusion: The Calm Mathematics of Competition

The winner-take-all narrative has a gravitational pull because it maps cleanly onto the most visible technology outcomes of the past two decades. Google won search. Facebook won social. AWS leads cloud. These examples are real. But they are also exceptional — products of specific structural conditions (strong direct or indirect network effects, low workflow heterogeneity, high demand-side returns to scale) that do not generalize to most B2B software markets.

The empirical reality of vertical SaaS is less dramatic but more economically interesting. These are markets that find and maintain a competitive equilibrium — not the equilibrium of perfect competition (too many switching costs and regulatory barriers for that) and not the equilibrium of monopoly (too little network effect and too much workflow heterogeneity for that), but the equilibrium of stable oligopoly with persistent multi-homing.

For those building and investing in vertical SaaS, the implication is a kind of liberation. You do not need to win the entire market. You do not need to crush all competitors. You need to build something that twenty percent of the market will find indispensable, and then you need to defend that position through depth, compliance, and workflow integration rather than through scale advantages you will never possess.

This is not a lesser ambition. Some of the most durable, profitable software companies in the world — Veeva, Procore, Tyler Technologies, Toast — are precisely this kind of business. They didn't win everything. They won enough.

As Spinoza might have observed: the nature of a thing is not what we wish it to be, but what the structure of reality requires it to be. The structure of vertical SaaS markets requires competitive equilibrium. The wise builder works with that structure, not against it.

---

Further Reading

• Herfindahl-Hirschman Index — Market concentration measurement
• Shopify — Vertical SaaS case study
• Increasing Returns on Wikipedia — W. Brian Arthur's theory

References

1. Arthur, W.B. (1989). "Competing Technologies, Increasing Returns, and Lock-In by Historical Events." The Economic Journal, 99(394), 116-131.

2. Klemperer, P. (1987). "Markets with Consumer Switching Costs." The Quarterly Journal of Economics, 102(2), 375-394.

3. Thiel, P. & Masters, B. (2014). Zero to One: Notes on Startups, or How to Build the Future. Crown Business.

4. Shapiro, C. & Varian, H.R. (1999). Information Rules: A Strategic Guide to the Network Economy. Harvard Business Press.

5. Eisenmann, T., Parker, G., & Van Alstyne, M. (2006). "Strategies for Two-Sided Markets." Harvard Business Review, 84(10), 92-101.

6. Armstrong, M. (2006). "Competition in Two-Sided Markets." The RAND Journal of Economics, 37(3), 668-691.

7. Farrell, J. & Klemperer, P. (2007). "Coordination and Lock-In: Competition with Switching Costs and Network Effects." Handbook of Industrial Organization, Vol. 3, 1967-2072.

8. Evans, D.S. & Schmalensee, R. (2016). Matchmakers: The New Economics of Multisided Platforms. Harvard Business Review Press.

9. U.S. Department of Justice & Federal Trade Commission (2023). Merger Guidelines. Section on HHI thresholds and market concentration.

10. Cusumano, M.A., Gawer, A., & Yoffie, D.B. (2019). The Business of Platforms: Strategy in the Age of Digital Competition, Innovation, and Power. Harper Business.

11. Bresnahan, T.F. & Greenstein, S. (1999). "Technological Competition and the Structure of the Computer Industry." The Journal of Industrial Economics, 47(1), 1-40.

12. Gartner (2025). "Market Share Analysis: CRM Software, Worldwide." Gartner Research.

13. IDC (2025). "Worldwide Public Cloud Services Market Shares." IDC MarketScape.

14. Acemoglu, D. & Robinson, J.A. (2012). Why Nations Fail: The Origins of Power, Prosperity, and Poverty. Crown Business. [For institutional analysis framework applied to regulatory moats.]

15. Taleb, N.N. (2012). Antifragile: Things That Gain from Disorder. Random House. [For the conceptual basis of multi-homing as risk diversification.]