Flat-rate vs per-user: the pricing split that decides Japanese B2B SaaS

While building AgentDoor's comparison guides we did something unglamorous: we read the published pricing pages for ten expense-management and electronic-contract tools across the US and Japan, and wrote down the actual numbers with their sources.

One pattern showed up in both categories, and it isn't a feature difference. It's a billing-shape difference — and it decides more purchases than feature checklists do.

The split

Broadly, the US-built tools charge per user. The Japanese-built tools charge a flat monthly fee, often with unlimited users, and recover volume elsewhere.

In expense management:

• Ramp is free per user on its entry tier if you adopt its card program; Expensify runs $5–9 per member per month; Zoho Expense $3; Navan $15 past the first five users.
• 楽楽精算 (Rakuraku Seisan) charges a flat ¥30,000+/month with a ¥100,000 setup fee — and stops scaling with headcount.
• TOKIUM charges from ¥10,000/month with unlimited accounts, then meters by receipt volume.

In electronic contracts the same shape repeats:

• DocuSign charges ¥3,300 per user per month on Standard (Japan storefront), with 100 envelopes per user annually and a hard 50-user ceiling.
• クラウドサイン (CloudSign) charges ¥11,000/month flat with unlimited users, plus ¥220 per document sent.
• GMOサイン (GMO Sign) charges ¥8,800/month flat with unlimited users, plus ¥100 per witness-type send.

Why this matters more than the feature grid

Because the two models cross over, and the crossover point is often inside the range of a normal company.

Take electronic contracts. At five people who need to send documents, DocuSign is roughly ¥16,500/month and CloudSign is ¥11,000 plus per-send volume — close enough that features decide it. At thirty senders, DocuSign is around ¥99,000/month while CloudSign's base fee hasn't moved. Nothing about either product changed; only your org chart did.

The same logic runs the other way. Flat-rate pricing looks expensive to a five-person startup and cheap to a 200-person company. A tool that was obviously wrong at your seed round can be obviously right at Series B, and vice versa — which is a real argument for re-checking pricing structure at each growth stage rather than assuming last year's answer holds.

The three questions worth asking first

1. Does our cost scale with people or with activity? Count the humans who need to send or submit, not total headcount. These are frequently very different numbers, and vendors quote against whichever flatters them.
2. What's the billing basis, not just the rate? Expensify's Control plan bills only users active that month. TOKIUM bills receipts, not seats. These change the arithmetic more than a per-unit discount would.
3. Where are the caps? DocuSign's Standard and Business Pro stop at 50 users; its envelope allowance is 100 per user per year. Caps are where a cheap-looking plan turns into a quoted enterprise conversation.

Why this is exactly the wrong work for humans

Notice what answering those questions requires: reading several pricing pages, normalising incompatible units (per seat, per month, per receipt, per send), and modelling each against your own numbers. It is arithmetic and translation. It is also most of what the early stage of a software evaluation actually is.

None of it requires judgement. Judgement is deciding whether the compliance gap matters, whether you trust the vendor, and what to negotiate — and that arrives only after the comparable numbers exist.

That gap is the reason AgentDoor exists. You describe the constraint once — "200 people, 40 of them send contracts, we need party-type signatures, we're on freee" — and agents collect comparable answers from each vendor in parallel, so the modelling is done before anyone books a meeting.

Read the sourced comparisons

Every figure above is on a page with its source URL and the date we checked it:

• Expense management tools, compared
• Electronic contract & e-signature tools, compared
• Total cost of ownership — why list price is rarely the deciding number
• Vendor evaluation — where the time actually goes

A caveat we apply to ourselves: pricing moves. Each record carries the date we verified it, and where a vendor doesn't publish something — compliance certifications, in several cases — we say so rather than inferring it. If you're making a decision, confirm against the vendor's own page.