Ramp: developing the ESS business model
The Core Parallel
Ramp monetizes through interchange (~1.5-2% of spend) and gives away the intelligence layer — expense management, duplicate invoice detection, vendor benchmarking, policy automation. That’s structurally similar to Costco: make your money from the relationship (interchange / membership), then optimize every transaction for the customer. The more clients on the platform, the better the cross-company benchmarking data, the better the savings recommendations, the stickier the product.
If Costco practices economies of scale shared, Ramp practices economies of intelligence and scale shared. As the financial copilot and intelligence layer for businesses, Ramp uses its cross-client data to help each business save money — passing the economies of intelligence back to them. As it scales, it becomes more intelligent, cross-referencing data points from different clients, which translates into further savings for everyone on the platform.
The Compounding Paths
The flywheel compounds in 2-3 directions simultaneously:
- Client saves → retains longer (lower churn, higher LTV)
- Client saves → grows faster (more margin to reinvest, more runway) → more spend volume through Ramp
- More clients → better data → better savings for everyone → clients save more → grow faster + retain longer → even more spend volume
As clients save, they retain longer and grow faster — more money to invest in the business, more margin to spend. Ramp compounds through faster client growth, more spend per client, and higher retention — all feeding the same loop.
The SKU and Inventory Turnover Analogy
The analogy extends beyond revenue mechanics into operational structure.
In Costco’s world: Physical inventory sitting on shelves is dead capital. Costco turns stock ~12x per year versus ~6x for supermarkets. Every dollar of inventory works twice as hard. Limited SKU count (~4,000 vs. 30,000+ at a supermarket vs 100,000+ at a Walmart) means each item gets more buyer attention, better supplier terms, and better quality control. This creates a negative working capital dynamic — goods sell in under 30 days, suppliers are paid in 30-60 days — so growth generates cash rather than consuming it.
In Ramp’s world: The “inventory” is your finance team and your software stack — paid-for capacity sitting on the shelf. Every manual expense report, every duplicate invoice review, every reconciliation task is inventory waiting to be turned. Ramp’s automation clears the friction that blocks productive work from reaching the people. It creates “demand” for the idle asset inventory, freeing up human capital to do higher-value work — increasing asset turnover.
The Structural Parallels
| Costco | Ramp |
|---|---|
| Reduces SKUs to increase velocity per SKU | Consolidates tools (card + expense + bill pay + accounting) to increase velocity per workflow |
| Low SKU count means each item gets more buyer attention, better terms, better quality | Single integrated layer means each process gets more automation attention, better data, fewer handoffs |
| Negative working capital: sell before you pay suppliers | Value delivered before full cost of adoption is incurred — automation saves time before the client fully onboards every workflow |
| ~4,000 SKUs vs. 30,000+ at competitors | One platform vs. 5-8 disconnected finance tools at most companies |
Where the Analogy Has Limits
The incentive alignment is close but not identical. Costco’s membership fee means they literally profit when goods are sold cheaper. Ramp profits from interchange — they benefit when clients spend more through the card, not necessarily when they spend less overall. Those aren’t contradictory (cut waste → redeploy into growth spend → more interchange), but the incentive vector isn’t as perfectly aligned as “we collect a flat fee, then everything is at cost.”
Data network effects are real but often weaker than scale economies in physical retail. The jump from 100 to 10,000 clients is transformative for benchmarking. The jump from 10,000 to 100,000 may be marginal. Costco’s scale economies are mechanical — 2x volume = concrete supplier leverage. Intelligence economies hit diminishing returns faster. Ramp’s deeper moat may be in switching costs (embedded in accounting stack) more than in the data itself.