SaaSCron: Metrics & Pricing

SaaS revenue recognition: ASC 606 and IFRS 15 basics for subscriptions

Ricardo Machado — Tue, 27 Oct 2026 21:00:00 GMT

You closed an annual contract, the cash is in the bank, and your accountant says you earned only a small part of it this month. Your investor update shows bookings, your income statement shows revenue, and the two never match. You need to know why, so you can read your own numbers and ask your accountant the right questions.

Revenue recognition is the set of rules that decides when money from a customer counts as revenue. Under U.S. GAAP the rules sit in ASC Topic 606; under international standards, in IFRS 15. Both apply to subscription software as described in What is SaaS?. This guide explains the shared model in plain words. It is not accounting advice: how the rules apply to your contracts is a judgment for your accountant and auditor.

Revenue recognition in plain terms

Think of a gym that sells a year’s membership for $600 in January. The gym has the cash on day one, but it still owes you eleven more months of access. Each month it delivers, it earns $50. Until then, the unearned part sits on its balance sheet as something it owes you.

Two terms carry the whole topic:

Deferred revenue: money billed or received for service not yet delivered. Some filings, such as Salesforce’s, call it unearned revenue.
Performance obligation: a distinct promise in the contract, such as software access or an onboarding project.

The five steps

The IFRS Foundation summarizes IFRS 15 in five steps. FASB’s Topic 606, issued as Accounting Standards Update 2014-09 in May 2014, uses the same model. The IFRS Foundation says IFRS 15 was issued “together with” Topic 606 and applies to annual periods beginning on or after 1 January 2018. The two are not identical in every detail, and FASB publishes a comparison of them, so check which standard your company reports under.

Step	What the standard asks	The SaaS question
1. Identify the contract	Which agreement or agreements exist with the customer	Is the contract the order form, the online terms, or both?
2. Identify the performance obligations	Which distinct promises the contract makes	Is onboarding, support or a term license separate from access?
3. Determine the transaction price	How much consideration you expect to receive	What about discounts, credits and usage fees you cannot know yet?
4. Allocate the price	Split it across the promises by relative stand-alone selling price	Does a “free” onboarding package still take a share of the price?
5. Recognize revenue	As each promise is fulfilled, when the customer obtains control	Is access delivered over the term or at one point in time?

The core principle, in the IFRS Foundation’s words, is that a company “recognises revenue to depict the transfer of promised goods or services to the customer”, in the amount it expects to be entitled to.

How a public SaaS company applies it

Salesforce’s annual report for the year ended January 31, 2026, shows the model at scale (Form 10-K):

Revenue from cloud services is generally recognized ratably over the contract term.
Term licenses are “generally recognized at the point in time when the software is made available to the customer.”
Customers are generally invoiced “in advance, in annual installments”, which is why unearned revenue builds up on the balance sheet.
Remaining performance obligation, the contracted revenue not yet recognized, was about $72.4 billion. About $35.1 billion of it, or 48%, was expected to become revenue in the next 12 months.

The last figure shows the gap between signing and earning: $35.1 billion / $72.4 billion = 0.485, so roughly half of what customers had committed would turn into revenue within a year.

Worked example: a 12-month contract with onboarding

Example: a labeled hypothetical, for illustration only. On January 1 you sign a 12-month contract. Your list prices are $12,000 for a year of access and $3,000 for onboarding, $15,000 in total. You give a 10% discount, so the customer pays $13,500 up front. Assume your accountant concludes that onboarding is a distinct promise, completed in January, and that list prices are the stand-alone selling prices. If the conclusion is instead that onboarding is not distinct, it is combined with access. Its fee is then recognized with the access, not in January, and the table below would change.

Step 4 splits the $13,500 by relative stand-alone selling price:

Access: $12,000 / $15,000 = 80%, and 80% x $13,500 = $10,800, or $10,800 / 12 = $900 a month.
Onboarding: $3,000 / $15,000 = 20%, and 20% x $13,500 = $2,700, recognized when delivered in January.

Date	Event	Revenue this month	Deferred revenue after the event
January 1	Invoice paid, $13,500 cash	$0	$13,500
January 31	Onboarding done, one month of access	$2,700 + $900 = $3,600	$13,500 - $3,600 = $9,900
February 28	One month of access	$900	$9,900 - $900 = $9,000
Each month to December	One month of access	$900	falls by $900
December 31	Last month of access	$900	$0

Check the total: $3,600 + 11 x $900 = $3,600 + $9,900 = $13,500, the full contract. The discount did not land only on onboarding or only on access. Both carry 10% less than list, because step 4 spreads it in proportion.

Bookings in January are $13,500. Revenue in January is $3,600. Neither number is wrong; they answer different questions. Metrics such as MRR and ARR are management measures and are not defined by either standard, so label them when you report them. For how billing events like upgrades and proration show up in those metrics, see SaaS subscription.

Usage fees and multi-year deals

Usage fees. A usage charge is unknown when you sign. Step 3 asks for the amount you expect to be entitled to, so estimates enter the analysis. This is a judgment area: document how you bill usage and take the contracts to your accountant.

Multi-year contracts. A three-year deal billed annually does not change when access is delivered. Revenue still follows delivery, month by month, while billing follows the payment schedule. Remaining performance obligation counts all contracted revenue not yet recognized, billed or not, so it also includes the later years not yet invoiced. The license metric and term in the contract, covered in SaaS licensing, feed directly into steps 2 and 3.

Common mistakes

Booking an annual invoice as revenue in the month it is paid. It makes one month look large and the next eleven look empty. Why it happens: a cash-based ledger records money when it arrives. Instead: record deferred revenue and release it as the service is delivered.
Treating a bundled “free” service as worth nothing. If the package is a distinct promise, step 4 allocates by stand-alone selling price, so it can still take a share of the contract. Why it happens: the invoice line shows $0. Instead: keep a list price for every promise you make.
Reporting bookings or ARR as revenue. Investors and auditors read them as different numbers. Why it happens: bookings are the figure sales celebrates. Instead: show bookings, revenue and deferred revenue side by side, each labeled.
Deciding setup-fee or usage treatment alone. These are the judgment areas of the standards. Why it happens: they look like bookkeeping. Instead: write down the contract terms and get your accountant’s view before the first audit or due diligence.

Sources

IFRS Foundation, IFRS 15 Revenue from Contracts with Customers, checked October 17, 2026.
FASB, Accounting Standards Update 2014-09, Revenue from Contracts with Customers (Topic 606), May 2014.
Salesforce, Inc., Form 10-K for the fiscal year ended January 31, 2026, SEC EDGAR, checked October 17, 2026.

In three sentences

ASC 606 and IFRS 15 share one five-step model: identify the contract and its promises, set and allocate the price, and recognize revenue as each promise is delivered. In the example, a $13,500 prepaid contract produces $3,600 of revenue in January and $900 a month after that, with deferred revenue falling to zero by December. Keep bookings, revenue and deferred revenue apart, and take setup fees, usage charges and bundles to your accountant, because those are the judgment calls.

ARPA vs ARPU in SaaS: definitions, formulas and when to use each

Ricardo Machado — Sat, 24 Oct 2026 16:00:00 GMT

Your board deck shows “ARPU”, your investor’s template asks for “ARPA”, and the two numbers can differ by a factor of nine for the same company. You need to know which one to report, how each is calculated and why they can move in opposite directions. Average revenue per account (ARPA) divides recurring revenue by the number of paying customers; average revenue per user (ARPU) divides it by the number of people using the product.

The choice is not cosmetic. It decides whether a lost customer, a new small customer or a price change shows up in your numbers at all. This guide gives both formulas, works them through with the same company over two months, reads a real ARPU from an annual report and adds committed monthly recurring revenue (CMRR). For the basics of the SaaS model, see What is SaaS?.

ARPA and ARPU in plain terms

Think of it like a restaurant’s takings. You can divide the evening’s revenue by the number of tables served, or by the number of people who ate. A table of eight and a table of two are both one table, but 10 diners. The first number tells you what a booking is worth. The second tells you what a seat is worth.

In SaaS, the table is the account: the company or person that pays the invoice. The diner is the user: each person with a login.

ARPA = monthly recurring revenue (MRR) / number of paying accounts.
ARPU = MRR / number of paying users.

MRR is the recurring revenue you bill in a month, normalized to a monthly amount, without one-time fees. Use the same moment for the numerator and the denominator: end-of-month MRR with end-of-month counts, or a period’s revenue with the period’s average count.

When to use each

Situation	Use	Why
You sell to companies, one invoice per company	ARPA	The account is what you win, lose and spend on acquiring
You price per seat and want to test the seat price	ARPU	It isolates what one seat earns
Consumers, one person per subscription	Either	Accounts and users are the same count
You calculate customer lifetime value (LTV)	ARPA	Churn is counted in accounts, so revenue must be too
You compare yourself with a public company	Whatever the company defines	Read the filing’s definition first

In B2B SaaS, ARPA is usually the better fit for one reason: acquisition cost and churn are both measured per customer. Mixing per-user revenue with per-customer churn produces an LTV that means nothing. For that calculation, see LTV in SaaS.

Worked example: same company, two months

Example: a labeled hypothetical. A project-management tool charges $30 per user per month.

Month one: 200 paying accounts with 1,800 users in total, so MRR is 1,800 x $30 = $54,000.

During month two:

It wins 20 small accounts with two users each: 40 users, 40 x $30 = $1,200 of new MRR.
It loses one large account with 100 users: 100 x $30 = $3,000 of lost MRR.

Measure	Month one	Month two
Paying accounts	200	200 + 20 - 1 = 219
Paying users	1,800	1,800 + 40 - 100 = 1,740
MRR	$54,000	$54,000 + $1,200 - $3,000 = $52,200
ARPA	$54,000 / 200 = $270.00	$52,200 / 219 = $238.36
ARPU	$54,000 / 1,800 = $30.00	$52,200 / 1,740 = $30.00

ARPU did not move. Every user still pays $30, so a per-user metric cannot see the change. ARPA fell by ($270.00 - $238.36) / $270.00 = 11.7%, because the company swapped one large account for 20 small ones. Account count went up, MRR went down, and only ARPA tells that story.

The reverse also happens. If no customer leaves, a price rise from $30 to $33 per user moves ARPU and ARPA by the same 10%. Report both when you change prices, and ARPA alone when you report on your customer mix.

A real ARPU from a public filing

Dropbox reports ARPU in its annual report and defines it as revenue for the period “divided by the average paying users during the same period”. Average paying users is the count at the start of the period plus the count at the end, divided by two (Dropbox Form 10-K for 2025).

The filing’s figures for the year ended December 31, 2025:

Revenue: $2,521.0 million (2024: $2,548.2 million).
Paying users at year end: 18.08 million (end of 2024: 18.22 million).
Reported ARPU: $138.91 (2024: $140.23).

You can redo the calculation. Average paying users: (18.22 + 18.08) / 2 = 18.15 million. ARPU: $2,521.0 million / 18.15 million = $138.90. The cent of difference from the reported $138.91 comes from user counts rounded to two decimals. Per month, that is $138.91 / 12 = $11.58 per paying user. ARPU fell by $1.32, or 0.9%, from 2024.

Note what Dropbox counts. Paying users are “users who have active paid licenses”, so a team plan with 10 licenses counts as 10 users. That makes ARPU the natural metric for a product sold largely by the license. A company that sells mainly by the account would report ARPA.

CMRR: committed monthly recurring revenue

CMRR is MRR adjusted for what is already signed. Add contracts that are signed but not yet live; subtract revenue from customers who have given notice to cancel. It answers “what will MRR be if nothing else changes?”

Example: the same hypothetical company ends month two with $52,200 of MRR. It has signed $4,000 of new monthly contracts that start next month, and customers worth $1,500 a month have given notice.

CMRR = $52,200 + $4,000 - $1,500 = $54,700.

Keep ARPA on live MRR, not on CMRR. The signed accounts are not paying yet, and counting them would raise ARPA before any money arrives. Show CMRR next to MRR as a forward-looking line, and say which contracts it includes.

Common mistakes

Calling ARPA “ARPU” in a B2B deck. Readers compare your number with companies that define it differently. Why it happens: ARPU is the better-known acronym. Instead: write the definition under the number, as Dropbox’s filing does.
Mixing periods. Annual revenue divided by end-of-year accounts gives an annual ARPA that looks 12 times larger than a monthly one. Why it happens: finance reports yearly, dashboards monthly. Instead: label every ARPA with its period and the counting method.
Including one-time fees. Setup or onboarding fees in the numerator inflate ARPA in a month with many new customers. Why it happens: total revenue is the easiest number to pull. Instead: use recurring revenue only.
Reading a flat ARPU as “nothing changed”. As the example shows, ARPA can fall 11.7% while ARPU stays at $30. Why it happens: one metric gets all the attention. Instead: track both, plus the account and user counts behind them.

For how ARPA feeds into the rest of the model, see the SaaS business model.

Sources

Dropbox, Inc., Form 10-K for the fiscal year ended December 31, 2025, key metrics and definitions.

In three sentences

ARPA is monthly recurring revenue divided by paying accounts, and ARPU is the same revenue divided by paying users. In the example, swapping one 100-user account for 20 two-user accounts left ARPU at $30 but cut ARPA by 11.7%, from $270.00 to $238.36. Use ARPA when you sell to companies and calculate LTV, use ARPU to test a per-seat price, and keep CMRR as a separate line for contracts that are signed but not yet live.

SaaS valuation multiples: how revenue multiples work and what moves them

Ricardo Machado — Thu, 22 Oct 2026 11:00:00 GMT

You read that a software company sold for “five times revenue”, and you want to know what that means for your own company, or for one you are thinking of investing in. You also suspect that two people quoting a multiple are not always dividing the same numbers. They often are not, and the difference can move a valuation by millions.

A SaaS valuation multiple is the price of a company divided by a measure of its size, most often revenue or annual recurring revenue (ARR). It is a shorthand for comparing companies of different sizes. If SaaS itself is new to you, start with What is SaaS?.

A revenue multiple in plain terms

Think of houses priced per square foot. Two houses of the same size can sell for very different prices because of the street, the school district and the state of the roof. The price per square foot lets you compare them quickly. It does not explain why one is worth more. A revenue multiple does the same job for companies: it summarizes the price, and the reasons sit elsewhere, in growth, retention and margins.

Three definitions come first:

Equity value: what the shares are worth, the share price times the number of shares, or the price paid for all of them in a sale.
Enterprise value (EV): equity value plus debt minus cash. It is the price of the business itself, whoever finances it.
ARR (annual recurring revenue): the yearly value of active subscriptions at a point in time. Revenue, by contrast, is what was recognized over a period and can include services and one-time fees.

The formulas

The common multiples divide enterprise value by a size measure:

EV / revenue (trailing): enterprise value divided by revenue over the last 12 months or the last fiscal year.
EV / revenue (forward): enterprise value divided by expected revenue for the next 12 months.
EV / ARR: enterprise value divided by current ARR, used often for private SaaS companies.
EV / EBITDA: enterprise value divided by earnings before interest, taxes, depreciation and amortization. It only works for companies with positive earnings, which many growing SaaS companies do not have.

Use enterprise value, not equity value, on top. Revenue belongs to the whole business, whether it is financed by shareholders or by lenders, so the price you divide must include both.

Worked example: two companies with the same ARR

Example: two labeled hypothetical companies, each with $10 million of ARR.

Item	Company A (fast growth)	Company B (slow growth)
ARR today	$10.0M	$10.0M
ARR growth, next 12 months	40%	10%
Equity value	$60.0M	$28.0M
Debt	$5.0M	$4.0M
Cash	$15.0M	$2.0M
Enterprise value	$60.0M + $5.0M - $15.0M = $50.0M	$28.0M + $4.0M - $2.0M = $30.0M
EV / ARR today	$50.0M / $10.0M = 5.0x	$30.0M / $10.0M = 3.0x
ARR in 12 months	$10.0M x 1.40 = $14.0M	$10.0M x 1.10 = $11.0M
EV / forward ARR	$50.0M / $14.0M = 3.6x	$30.0M / $11.0M = 2.7x

Read the table from top to bottom. Same ARR, different prices. Company A’s higher multiple is mostly paying for growth: on next year’s ARR its multiple falls from 5.0x to 3.6x. Also look at equity against enterprise value. Company A’s equity value of $60.0M divided by ARR gives 6.0x, a full turn higher than its EV multiple, because A holds $10.0M more cash than debt. Quote equity over revenue as if it were EV over revenue, and your comparison is wrong before it starts.

How much one turn is worth

A “turn” is one point of multiple. On a company with $10 million of ARR, each turn is $10 million of enterprise value. For Company A, whose cash exceeds its debt by $10.0M, the shareholders’ value is enterprise value plus that $10.0M.

EV / ARR	Enterprise value ($10.0M ARR)	Equity value for Company A (EV + $10.0M net cash)
3.0x	$30.0M	$40.0M
5.0x	$50.0M	$60.0M
8.0x	$80.0M	$90.0M

The jump from 3.0x to 8.0x more than doubles the equity value, from $40.0M to $90.0M, with no change in ARR. That is why the factors behind the multiple matter more than the multiple itself.

A real deal, with the arithmetic shown

Public filings let you rebuild a multiple yourself. In May 2025, Salesforce agreed to acquire Informatica. Salesforce’s quarterly report on Form 10-Q states that Informatica’s Class A and Class B-1 holders would receive $25 in cash per share. It adds that the transaction “represents an equity value of approximately $8 billion”, net of the Informatica shares Salesforce already held (Salesforce Form 10-Q, quarter ended July 31, 2025).

Informatica’s results for fiscal 2024, filed with the SEC, give the denominators (Informatica Q4 and full-year 2024 results, Exhibit 99.1):

Total revenue for 2024: $1,640.0 million.
Total ARR at December 31, 2024: $1,725.5 million.
Net debt at December 31, 2024: $1,823 million term loan minus $1,232 million of cash and short-term investments = $591 million.

Example: an approximate multiple from those figures.

Enterprise value: $8,000M equity + $591M net debt = $8,591M.
EV / 2024 revenue: $8,591M / $1,640.0M = 5.2x.
EV / ARR: $8,591M / $1,725.5M = 5.0x.
Equity value / revenue, for comparison: $8,000M / $1,640.0M = 4.9x.

Treat these as approximations. The equity value is rounded in the filing and excludes the shares Salesforce already owned, and the balance sheet is from December 2024, five months before the agreement. This is one deal’s price, not a benchmark for the market or for your company.

What moves a SaaS multiple

The multiple summarizes what buyers believe about future cash. These factors explain most of it:

Growth. Faster ARR growth means more revenue in later years, as Company A shows.
Retention. Net revenue retention above 100% means existing customers grow on their own. Low retention means growth has to be bought again each year.
Gross margin. More of each dollar is left after hosting and support.
Profitability. A common heuristic, the Rule of 40, adds the growth rate and the profit margin. In two labeled hypotheticals, 40% growth with a -5% free cash flow margin and 10% growth with a 25% margin both score 35. The score is the same; the businesses are not.
Revenue quality. Recurring subscription revenue is more predictable than services and one-time fees, which is why EV / ARR and EV / revenue can differ.
Size and market conditions. Buyers price scale and the cost of borrowing at the moment of the deal, and both change outside your control.

For the retention and unit economics behind these factors, see LTV in SaaS and the SaaS business model. For the share of subscription revenue at five large public companies, see What is a SaaS company?.

Common mistakes

Dividing equity value by revenue and calling it a revenue multiple. In the example, Company A shows 6.0x on equity and 5.0x on enterprise value. Why it happens: the share price is the number in the headline. Instead: compute EV first, then divide.
Mixing trailing and forward multiples. A forward multiple looks lower for a growing company. Why it happens: reports do not always say which one they quote. Instead: label every multiple with its denominator and period.
Applying a public company’s multiple to a small private one. Public shares can be sold any day; a small private stake cannot, and its size and risk differ. Why it happens: public multiples are the only ones easy to find. Instead: use them as context and discount for size and liquidity, with the reason stated.
Counting services and one-time fees as ARR. It raises the denominator and flatters the business. Why it happens: total revenue is easier to pull from the accounts. Instead: count only active recurring subscriptions in ARR.

Sources

Salesforce, Form 10-Q for the quarter ended July 31, 2025, checked October 2026.
Informatica, Fourth quarter and full-year 2024 results, Form 8-K Exhibit 99.1, checked October 2026.

In three sentences

A SaaS revenue multiple is enterprise value (equity plus debt minus cash) divided by revenue or ARR, and it must say which denominator and period it uses. Two hypothetical companies with $10 million of ARR are worth 5.0x and 3.0x because one grows 40% a year and the other 10%, and Informatica’s filings give about 5.2x revenue. Compute EV yourself, label every multiple and look at growth, retention and margins to explain it.

SaaS licensing explained: seats, usage, enterprise agreements and contract terms

Ricardo Machado — Tue, 20 Oct 2026 21:00:00 GMT

A renewal quote for one of your SaaS tools lands next month. It lists more seats than people who log in, a line for “overage” nobody remembers agreeing to, and a price increase. Finance wants to know what you are actually licensed for before anyone signs. You need to read the license the way the vendor wrote it.

A SaaS license is the contract right to use software that the vendor hosts: who may use it, how much, for how long and at what price. Unlike old installed software, you never own a copy. You rent access, and the license metric decides what you pay for. For the basics of the model, see What is SaaS?.

A SaaS license in plain terms

Think of parking at an office. One company buys a named permit for each employee’s car, whether the car comes in or not. Another pays the meter for the hours it actually parks. A large company leases the whole lot for three years at a negotiated rate. The parking spaces are the same. What changes is what gets counted, and who pays for the empty spaces.

SaaS licensing works the same way. The vendor picks a license metric, the unit it counts and bills. The contract then adds the term (how long you commit), the renewal rules and what happens when your count goes up or down.

The four common license metrics

The table compares the four metrics you will meet most often. Each example comes from the vendor’s own public page, as of October 2026.

Metric	What is counted	Who pays for unused capacity	Public example
Named user (seat)	Each person given access, active or not	You	GitHub lists Team and Enterprise plans per user per month (GitHub pricing)
Active user	Only people who used the product in the period	The vendor	Slack credits members inactive for over 28 days (Slack fair billing policy)
Usage	Units of work: tasks, API calls, events, volume	Nobody, but cost is harder to predict	Zapier counts “tasks” per month (Zapier pricing); Stripe Billing charges 0.7% of billing volume (Stripe Billing pricing)
Enterprise agreement	A negotiated bundle: seat bands, usage pools, term, price protection	Depends on the bands you commit to	Usually “contact sales”; terms are private

Named seats are simple. They are easy to budget and easy to over-buy. Active-user billing moves the risk of idle seats to the vendor. Usage metrics fit products whose value grows with volume, but your bill moves with your business. For how each metric shapes the vendor’s price list, see SaaS pricing models.

How active-user billing works

Slack’s fair billing policy is a public, precise example. A member counts as inactive when they have not used Slack “in over 28 days”. The workspace then receives a credit for the unused part of the billing period. Slack’s own formula is the cost per member divided by the days in the month, times the days left. Its help page works it through as ($8.75 / 30 days) x 15 days = $4.38. If the member comes back, Slack charges the prorated cost for the rest of the period.

Not every vendor works this way. Under a named-seat license, an employee who left in March still costs you a full seat until someone removes them, and sometimes until the next renewal.

Contract terms that change the bill

The metric sets the unit. These terms decide how much of it you pay for:

Term and renewal. Annual and multi-year terms usually come with a discount and an automatic renewal. Note the notice period: miss it and you renew at the vendor’s terms.
True-up. A clause that bills you for seats added during the term, often at a review date or at renewal.
Seat reduction rights. Many contracts let you add seats at any time but reduce only at renewal. Some do not let you reduce at all within a multi-year term.
Price increase cap. A limit on how much the per-unit price can rise at renewal. Without one, the renewal price is whatever the vendor quotes.
Feature gating. Some features sit only in higher tiers. GitHub’s plan cards list SAML single sign-on on Enterprise and not on Team. A security requirement can force the whole company onto a higher tier.
Caps by plan. Some plan families have a seat ceiling. The Microsoft 365 Business plans stop at 300 users, as worked through in Microsoft 365 as SaaS.

Worked example: an annual seat license with a true-up

Example: a labeled hypothetical. You license 200 named seats at $12 per user per month on a 12-month contract, paid annually.

Annual license cost: 200 x $12 x 12 = $28,800.
A login report in month 6 shows 160 active users. Idle seats: 200 - 160 = 40.
Cost of idle seats for the year: 40 x $12 x 12 = $5,760, or $5,760 / $28,800 = 20% of the contract.
In month 7 a new team joins and you add 30 seats. The true-up clause bills them for the remaining 6 months: 30 x $12 x 6 = $2,160.
Total for the year: $28,800 + $2,160 = $30,960.

Now look at what happened. You paid $2,160 for 30 new people while 40 paid seats sat unused. If the contract had allowed reassigning idle seats, the new team would have cost nothing extra. Reassigning 30 of the 40 idle seats to the new team would have saved the full $2,160, and the 10 left over show what to cut at renewal.

The same 40 idle users under active-user billing like Slack’s would have earned credits instead. At $12 per member and a 30-day month, Slack’s formula gives ($12 / 30) x 30 = $12.00 per idle member per full month, or 40 x $12.00 = $480 a month.

Checklist before you sign or renew

Find the license metric in the order form, not on the pricing page. They can differ.
Pull a login report for the last 90 days and count active users against licensed seats.
Remove or reassign seats of people who left before the true-up date.
Read the seat reduction clause: can you go down at renewal, or never within the term?
Find the renewal notice period and put it in a calendar 30 days earlier.
Ask for a price increase cap in writing.
List the features your security team requires, such as single sign-on, and check which tier includes them.
For usage metrics, take your last three months of usage and price the busiest month, not the average.

For how the same per-seat logic plays out in a CRM purchase, see CRM SaaS.

Common mistakes

Paying for seats of people who left. Named seats cost the same whether anyone uses them. Why it happens: offboarding removes email access but not every SaaS account. Instead: tie seat removal to the offboarding checklist and run a login report each quarter.
Buying a seat for everyone who might need access. Every read-only viewer costs a full seat on many plans. Why it happens: it is easier than handling access requests. Instead: check for viewer, guest or lower-cost seats, and share reports by export where you can.
Missing the renewal notice period. Auto-renewal at the vendor’s new price is the default. Why it happens: the notice date sits in the contract, not in anyone’s calendar. Instead: record the date when you sign, with a reminder 30 days before it.
Budgeting usage licenses at the average month. Usage bills follow your busiest periods, and overage rates are often higher than the included rate. Why it happens: averages look tidy in a spreadsheet. Instead: price the peak month and ask for the overage rate in writing.

Sources

Slack, Fair billing policy, checked October 2026.
GitHub, Pricing, checked October 2026.
Zapier, Pricing, checked October 2026.
Stripe, Billing pricing, checked October 2026.

In three sentences

A SaaS license sets a metric (named seats, active users, usage or a negotiated enterprise bundle) plus the term, renewal and change rules that decide what you pay. In the example, 40 idle seats out of 200 waste $5,760 a year, and a $2,160 true-up for 30 new users could have been avoided by reassigning them. Read the metric in the order form, count active users before every renewal and get seat reduction, notice period and price cap terms in writing.

SaaS subscriptions: how the model works and how to manage one

Ricardo Machado — Tue, 20 Oct 2026 16:00:00 GMT

Your product has paying customers, and the billing questions have started. One customer wants to upgrade halfway through the month, another wants an annual plan, and a third asks why the invoice shows a credit and a charge on the same day. Each one is a subscription question, and your answers need rules you can repeat.

A SaaS subscription is an agreement in which a customer pays a recurring fee, monthly or yearly, for continued access to software that the vendor hosts and runs. This guide covers how the model works, the arithmetic of discounts and mid-cycle changes, and what a billing system does for you. If SaaS itself is new to you, start with What is SaaS?.

A subscription in plain terms

Think of a magazine subscription. You pay for a year up front or one month at a time. Issues keep arriving until you stop paying. If you switch to the premium edition halfway through the year, the publisher credits the issues you had not received yet and charges you for the rest of the premium year.

A SaaS subscription works the same way, with software access in place of issues. Five terms carry most of the weight:

Billing period: the stretch of time one payment covers, usually a month or a year.
Renewal: the automatic start of the next billing period, with a new charge, unless the customer cancels.
Proration: charging or crediting only the share of a period that a change affects.
MRR (monthly recurring revenue): the monthly value of all active subscriptions. An annual plan counts as one twelfth of its price each month.
Dunning: the process of retrying a failed payment and reminding the customer before access ends.

The life of a subscription, step by step

Sign-up. The customer picks a plan and a billing period. A trial may come first.
First invoice. In the usual setup, the fixed fee is charged at the start of the period it covers.
Renewal. At the end of the period, the next invoice is created and charged automatically.
Change. The customer adds seats, upgrades or downgrades. The billing system prorates.
Failed payment. A card expires or is declined. Dunning retries the charge and emails the customer.
Cancellation. Access ends now or at the end of the paid period, depending on your terms.

Every step changes either the invoice, the MRR or both. The table below shows how, and it is the reference to keep next to your finance reports.

Event	What happens to the invoice	What happens to MRR
New monthly subscription at $50	$50 charged now	+$50
New annual subscription at $480	$480 charged now	+$40 ($480 / 12)
Upgrade from $50 to $80 mid-month	Credit for unused $50 time, charge for remaining $80 time	+$30 from the change
Downgrade from $80 to $50 mid-month	Credit for unused $80 time, usually not refunded in cash	-$30 from the change
Failed renewal payment	Invoice stays open while dunning retries	Unchanged until you decide it is churn
Cancellation at period end	No new invoice	-$50 when the period ends

The figures are a labeled hypothetical. The rule they show is general: cash arrives on the invoice date, while MRR spreads the value over the months it covers.

Monthly or annual billing

Annual billing gives the vendor cash up front and a 12-month commitment. In exchange, vendors usually discount it. Zapier’s pricing page is a clear public case: its Professional plan with 750 tasks is listed at $19.99 a month billed annually and $29.99 billed monthly (Zapier pricing, as of October 2026).

Example: the same plan, paid for one year in each way.

Billed annually: $19.99 x 12 = $239.88 a year.
Billed monthly: $29.99 x 12 = $359.88 a year.
Difference: $359.88 - $239.88 = $120.00 a year.
Discount for paying annually: $120.00 / $359.88 = 33.3%.

For the vendor, that annual customer adds $239.88 / 12 = $19.99 of MRR, not $239.88 in the month the money arrives. Reporting the full prepayment as one month’s revenue inflates a growth chart by accident.

The customer’s side of the trade is simple. Annual billing saves a third here, but it ties the money for a year. Pick monthly while you are still testing whether a tool fits, and annual once you know it does.

Upgrades, downgrades and proration

When a customer changes plan in the middle of a period, a fair bill charges only for the time on each plan. Stripe’s documentation gives the standard case. A customer upgrades from a $10 monthly plan to a $20 plan halfway through the period. The invoice credits $5 for the unused half of the old plan and charges $10 for the remaining half of the new one, so the customer pays $5 more (Stripe, Prorations).

Two details from the same page matter in practice. Stripe prorates to the second by default, so the amount changes slightly between a preview and the actual change unless you fix the proration date. And negative prorations, the credits from a downgrade, are not refunded or paid out automatically: they reduce future invoices unless someone refunds them by hand.

Example: a team plan priced per seat, a labeled hypothetical at $15 per seat per month on a 30-day billing period. The customer has 10 seats and adds 5 with 12 days left.

Full-month price of the new seats: 5 x $15 = $75.
Share of the period remaining: 12 / 30 = 0.4.
Prorated charge now: $75 x 0.4 = $30.
Next full invoice: 15 x $15 = $225.
MRR change: +$75, from $150 to $225.

The customer pays $30 today, and MRR rises by the full $75. Mixing those two numbers up is a second reporting error to avoid.

What subscription billing software does

You can bill a handful of customers by hand. Past that, a billing system runs the schedule for you. It stores the plans, creates invoices on each renewal date, calculates proration, retries failed cards, sends receipts and reports MRR. Some also handle sales tax and the customer portal where people update their card.

These systems usually charge a share of what they bill. Stripe Billing lists a fee of 0.7% of billing volume, on top of payment processing fees (Stripe Billing pricing, as of October 2026). At $50,000 of subscriptions billed in a month, that is $50,000 x 0.7% = $350 for the month, before card fees. Compare that with the hours someone would spend on invoices and failed-payment emails.

For how the price itself is structured, flat, per seat, tiered or usage-based, see SaaS pricing models.

Renewal and cancellation rules

Automatic renewal is convenient for both sides until a customer feels trapped by it. For consumer subscriptions, renewal and cancellation are regulated, and the rules are moving. The Federal Trade Commission’s page on its Negative Option Rule lists a February 2026 revision “to conform these rules to federal court decisions”. It also lists a March 2026 request for public comment on new rulemaking (FTC, Negative Option Rule). State laws add their own requirements.

If you sell to consumers, have a lawyer check your renewal notices and your cancellation flow. If you sell to businesses, put the renewal date, the notice period and the price-change terms in the contract in plain words.

Common mistakes

Counting an annual prepayment as one month’s revenue. It makes a single month look twelve times bigger than it is. Why it happens: the payment report shows cash, not recurring revenue. Instead: divide annual plans by 12 for MRR, as in the table above.
Promising refunds on downgrades that the system does not make. Credits from a downgrade usually reduce the next invoice rather than return cash. Why it happens: support agents assume a credit means a refund. Instead: write down your downgrade policy and preview the invoice before confirming the change.
Treating failed payments as cancellations. A declined card is often an expired card, not a customer who wants to leave. Why it happens: no dunning process exists, so access simply stops. Instead: retry the charge on a schedule and email the customer before you cut access.
Hiding the cancellation path. A hard-to-find cancel button produces chargebacks and complaints, and can break consumer rules. Why it happens: it looks like a cheap way to cut churn. Instead: make cancellation as easy as sign-up and ask one question about why the customer is leaving.

For how churn turns into customer lifetime value, see LTV in SaaS. For how subscription revenue, churn and acquisition cost fit together, see the SaaS business model.

Sources

Zapier, Pricing, checked October 2026.
Stripe, Prorations, checked October 2026.
Stripe, Billing pricing, checked October 2026.
Federal Trade Commission, Negative Option Rule, checked October 2026.

In three sentences

A SaaS subscription charges a recurring fee per billing period, renews automatically and prorates any change made in the middle of a period. An annual plan at $19.99 a month against $29.99 monthly saves the customer 33.3%, but it adds only $19.99 of MRR, never the full $239.88 in one month. Use a billing system for proration, retries and renewals, keep cash and MRR apart in your reports, and make cancellation as clear as sign-up.

NDR and GRR in SaaS: formulas, a cohort example and reported figures

Ricardo Machado — Sun, 18 Oct 2026 16:00:00 GMT

Your board pack shows NDR at 107% and GRR at 91%, and someone asks why one is above 100% and the other is not. You need to explain both and recompute them from your own billing data. Both measure what happened to revenue from the customers you already had a year ago. Gross revenue retention (GRR) counts only what you kept. Net dollar retention (NDR), also called net revenue retention (NRR), adds what those same customers bought on top.

Investors read the two numbers next to growth because they show whether existing customers are a base to build on or a bucket with a hole in it. This guide gives both formulas, works one cohort through step by step, shows what three public companies reported and lists the usual calculation errors. For the basics of SaaS, see What is SaaS?. For how retention feeds customer lifetime value, see LTV in SaaS.

NDR and GRR in plain terms

Think of a gym that looks at last October’s members one year later. Some have left. Some have dropped to a cheaper off-peak pass. Some have added personal training. GRR asks how much of last October’s membership fees is still being paid, ignoring the extras. NDR asks how much those same people pay now, extras included.

Two rules make both numbers mean something:

A fixed cohort. Only customers who were paying at the start of the period count. Customers won during the year are left out of both numbers.
Recurring revenue only. Use annual recurring revenue (ARR) or monthly recurring revenue. One-off fees, such as setup or consulting, stay out.

The formulas

Take the cohort’s recurring revenue at the start of the period. Twelve months later, sort every change in that revenue into three buckets:

Expansion: upgrades, added seats and added products.
Contraction: downgrades and removed seats, from customers who stayed.
Churn: revenue from customers who left entirely.

Then:

GRR = (starting revenue - contraction - churn) / starting revenue.
NDR = (starting revenue + expansion - contraction - churn) / starting revenue.

GRR can never exceed 100%, because expansion is left out. NDR can, whenever expansion outweighs the losses. The gap between the two is the expansion rate.

A cohort worked through

Example (hypothetical): a cohort of 200 customers paying $1,200,000 in ARR on October 1, 2025. On October 1, 2026, the same 200 accounts show:

Churn: 18 customers left. Together they paid $72,000.
Contraction: customers who stayed removed seats worth $36,000.
Expansion: customers who stayed added $192,000.

Step by step:

Losses: $36,000 + $72,000 = $108,000, or $108,000 / $1,200,000 = 9.0% of the base.
Expansion: $192,000 / $1,200,000 = 16.0% of the base.
GRR: ($1,200,000 - $36,000 - $72,000) / $1,200,000 = $1,092,000 / $1,200,000 = 91.0%.
NDR: ($1,092,000 + $192,000) / $1,200,000 = $1,284,000 / $1,200,000 = 107.0%.
Cross-check: 100% - 9% + 16% = 107%.

Read the two together. NDR of 107% says the cohort grew. GRR of 91% says it lost $108,000 of its base and refilled the hole with expansion from the customers who stayed. If expansion slows, the same losses turn the cohort negative: with $96,000 of expansion instead of $192,000, NDR would be ($1,092,000 + $96,000) / $1,200,000 = 99.0%.

Notice also the two churn rates. Customers lost: 18 / 200 = 9.0%. Revenue lost to churn: $72,000 / $1,200,000 = 6.0%. The customers who left were smaller than average, so logo churn and revenue churn tell different stories.

Why the numbers carry so much weight

NDR tells you how fast the base grows before you win a single new customer. At 107%, last year’s customers bring 7% more revenue this year. At 93%, you need new customers worth 7% of the base just to stay flat. GRR tells you how much of the base is solid. A high NDR with a low GRR means a few large customers expand while many others leave, which is harder to sustain than it looks.

What public companies report

Companies choose their own definitions, so read the definition before the number. Four disclosures from filings and results releases, checked October 17, 2026:

Company	Metric	Figure	As of	How the company calculates it
Workday	GRR	About 97%	January 31, 2026	ARR of customers at the prior period end against the same customers’ ARR now; churn counted, add-ons and expansion not
Weave	GRR	89% (91% a year earlier)	December 31, 2025	Revenue still billed to the same customer locations 12 months later; expansion, contraction and new locations not reflected
Weave	NRR	93% (98% a year earlier)	December 31, 2025	Weighted average of monthly rates over 12 months; new locations excluded
Snowflake	NRR	126% (125% on January 31, 2026)	July 31, 2026	Product revenue of a starting cohort in the second of two trailing years, divided by the first year

Weave states that its 2025 figures exclude the impact of its TrueLark acquisition. Snowflake’s product revenue depends on how much customers consume, so its rate moves with usage rather than with seats.

Three lessons from the table:

Same name, different arithmetic. Weave’s GRR does not reflect contraction, so a customer who downgrades does not lower it. The formula in this guide counts downgrades.
Different windows. Snowflake compares two trailing years; Weave averages twelve monthly rates; Workday compares two period ends.
Not benchmarks. Four figures from three companies on three dates do not make a target for your business. Compare your rate with your own history first.

For why reported ARR and reported revenue differ, see What is a SaaS company?.

Common mistakes

Letting new customers into the cohort. Revenue from accounts won during the year inflates NDR. Why it happens: the billing report shows all customers by default. Instead: freeze the list of customer IDs at the start date and track only those.
Counting one-off fees. Setup and consulting revenue makes the cohort look bigger at the start or stronger at the end. Why it happens: the invoice total is easier to pull than the recurring part. Instead: use ARR or monthly recurring revenue only.
Mixing logo churn and revenue churn. In the example, 9.0% of customers left but only 6.0% of revenue did. Why it happens: both are called “churn”. Instead: label which one you report, and show both when customer sizes vary.
Comparing your NDR with another company’s without reading its definition. A monthly-weighted rate, a two-year usage rate and a point-in-time ARR rate are different measures. Why it happens: the headline number is all a news story quotes. Instead: line up the definitions, as the table does, before you compare.

For how retention fits into the monthly arithmetic of a subscription business, see the SaaS business model.

Sources

Workday, Inc., Form 10-K for fiscal 2026.
Weave Communications, Inc., Form 10-K for 2025.
Snowflake Inc., second quarter fiscal 2027 results, September 2, 2026, and fourth quarter fiscal 2026 results, February 25, 2026.

In three sentences

GRR is the share of a cohort’s starting recurring revenue still billed a year later without counting expansion, and NDR is the same share with expansion added back. In the example, a $1,200,000 cohort keeps $1,092,000 (GRR 91%) and grows to $1,284,000 once expansion is added (NDR 107%). Read the two together, and read each company’s definition before comparing, because Workday, Weave and Snowflake each calculate theirs differently.

LTV in SaaS: formula, worked examples and the LTV:CAC ratio

Ricardo Machado — Sat, 17 Oct 2026 11:00:00 GMT

An investor, a board member or your finance lead asks for your LTV, and the number you produced last quarter looks too good to defend. You need a formula you can explain and numbers you can redo in front of them. Customer lifetime value (LTV) is the gross profit you expect from an average customer before they leave. The simplest honest version is average monthly revenue per account, times gross margin, divided by monthly churn.

LTV matters because it caps what you can afford to spend to win a customer. This guide shows two formulas, works them through at three churn rates, adds the LTV:CAC ratio and then shows the four ways the number misleads. For how recurring revenue, churn and acquisition cost fit together month by month, see the SaaS business model. For the basics of SaaS, What is SaaS?.

LTV in plain terms

Think of a gym. A member pays every month until one day they stop coming. If you know what a member pays, how much of it the gym keeps after costs, and how many months a member stays on average, you know what a new member is worth. That is the most the gym should spend on the poster that brought them in.

Three inputs, each defined once:

ARPA (average revenue per account): monthly recurring revenue divided by the number of paying customers. Use ARPA rather than average revenue per user when you sell to companies, because one account may hold many users.
Gross margin: revenue minus the cost of delivering the service (hosting, support, third-party fees), as a percentage of revenue.
Churn: here, monthly logo churn, the share of customers who leave in a month.

Two formulas

The average customer lifetime, in months, is one divided by monthly churn. At 2% monthly churn, the average customer stays 1 / 0.02 = 50 months. That follows from assuming the same share of customers leaves every month.

Revenue LTV = ARPA / monthly churn. It counts every dollar the customer pays.
Gross-margin LTV = ARPA x gross margin / monthly churn. It counts only the dollars you keep after the cost of serving them.

Use the second one. Acquisition spend is repaid out of gross profit, not revenue, so comparing revenue LTV with acquisition cost overstates the result by the cost of delivery.

Worked example at three churn rates

Example: a SaaS product with ARPA of $100 a month, an 80% gross margin and a customer acquisition cost (CAC) of $1,200. CAC is everything spent on sales and marketing in a period divided by the new customers won in it. Only the churn rate changes between columns.

Input or result	1% monthly churn	2% monthly churn	5% monthly churn
Average lifetime	1 / 0.01 = 100 months	1 / 0.02 = 50 months	1 / 0.05 = 20 months
Revenue LTV	$100 / 0.01 = $10,000	$100 / 0.02 = $5,000	$100 / 0.05 = $2,000
Gross-margin LTV	$80 / 0.01 = $8,000	$80 / 0.02 = $4,000	$80 / 0.05 = $1,600
LTV:CAC	$8,000 / $1,200 = 6.7	$4,000 / $1,200 = 3.3	$1,600 / $1,200 = 1.3
CAC payback	$1,200 / $80 = 15 months	15 months	15 months

The gross profit per month is $100 x 80% = $80, which is why the margin row divides $80.

Read the table from left to right. Moving churn from 1% to 5% cuts LTV by 80%, from $8,000 to $1,600. CAC payback does not move at all: it is 15 months in every column, because it depends only on CAC and monthly gross profit. Payback tells you how long the money is at risk. LTV tells you whether it comes back at all. At 5% churn the average customer stays 20 months, only five months longer than it takes to repay their acquisition cost.

The LTV:CAC ratio

LTV:CAC divides gross-margin LTV by CAC. A ratio of 1 means a customer repays their acquisition cost and nothing more. Below 1, every new customer loses money over their lifetime. Well above 1, you may be spending too little on growth.

If someone quotes a target ratio, such as 3, treat it as a convention, not a law. The ratio says nothing about how long the money takes to come back, which is why the payback row sits next to it. In the example, 2% churn gives a ratio of 3.3 with a 15-month payback. The same ratio with a 40-month payback would need far more cash in the bank.

Where the formula misleads

Blended churn across different customers. Suppose 80 customers are small businesses with 5% monthly churn and 20 are larger companies with 1% churn, all at the same $80 monthly gross profit.

Blended churn: (80 x 5% + 20 x 1%) / 100 = (4 + 0.2) / 100 = 4.2% a month.
LTV from blended churn: $80 / 0.042 = $1,905.
LTV by segment: small businesses $80 / 0.05 = $1,600; larger companies $80 / 0.01 = $8,000.
Weighted average of the segment LTVs: 0.8 x $1,600 + 0.2 x $8,000 = $1,280 + $1,600 = $2,880.

The blended number understates the average by about a third, and it hides that one segment is worth five times the other. Calculate LTV per segment and spend on acquisition per segment.

Very low churn. At 0.5% monthly churn the formula gives a 200-month lifetime, almost 17 years. Few young companies have data that can support that claim. Cap the lifetime at a period you can defend, such as 60 months, and say so.

Expansion revenue. Customers who add seats or upgrade raise ARPA over time, which the simple formula ignores. If you include expansion, use your measured net revenue retention, and say which version you report.

Early churn. Many products lose more customers in the first months than later. An average monthly rate spreads that early loss evenly and makes the first year look better than it is.

Common mistakes

Reporting revenue LTV against CAC. It overstates the ratio by the cost of delivery, 25% in the example ($10,000 against $8,000 at 1% churn). Why it happens: revenue is the number on the dashboard. Instead: multiply by gross margin first.
Using annual churn as if it were monthly. 10% annual churn is not 10% monthly; it is roughly 0.9% a month. Why it happens: reports mix periods. Instead: convert to monthly before dividing, and label the period.
One blended LTV for different segments. The example shows an understatement of about a third and a hidden five-fold gap. Why it happens: one number is easier to present. Instead: show LTV per segment next to CAC per segment.
CAC that leaves out salaries. Counting only ad spend makes CAC look small and the ratio look large. Why it happens: salaries sit in a different budget line. Instead: include the sales and marketing team’s pay and tools in CAC.

For how price structure affects ARPA in the first place, see SaaS pricing models.

In three sentences

Gross-margin LTV is ARPA times gross margin divided by monthly churn, and it is the version to compare with customer acquisition cost. With $100 ARPA, 80% margin and $1,200 CAC, LTV falls from $8,000 at 1% monthly churn to $1,600 at 5%, while the 15-month payback stays the same. Calculate it per customer segment, cap very long lifetimes and label every churn rate with its period, or the number will mislead the people you show it to.

SaaS pricing models: per-seat, usage-based, tiered and freemium compared

Ricardo Machado — Fri, 16 Oct 2026 15:00:00 GMT

You are about to put a price on a SaaS product, and every founder forum says something different: charge per user, no, charge per usage, no, give it away and upsell. The answer is that there are four models, each behaves differently as customers grow, and the right one depends on how your customers get value from the product. This guide prices one fictional product four ways so you can see the behavior in numbers.

The models are the same ones the 30 products in SaaS examples by category use, and the subscription mechanics behind them are explained in What is SaaS?.

The four models in plain terms

You already pay for things in all four ways. A gym membership is a flat price per person per month, whether you go twice or twenty times: that is per-seat pricing. A taxi meter charges by the distance you travel: that is usage-based pricing. A phone plan offers small, medium and large bundles, and you pick the bundle that fits: that is tiered pricing. A free app that nags you to upgrade for the good features is freemium.

In SaaS terms:

Per-seat (seat-based) pricing: a fixed price per user per month or year. Slack and Notion charge this way.
Usage-based pricing: a price per unit consumed, such as a message, a gigabyte or an API call. Twilio and Datadog charge this way.
Tiered pricing: two to four fixed plans with rising limits and features, often per company rather than per user. QuickBooks Online and Xero charge this way.
Freemium: a permanent free plan plus paid plans. Not a free trial, which ends after a set time. GitHub, Trello and Zoom have one.

Most real products mix two of them: a per-seat plan with usage charges above an included allowance, or a tiered plan with a free entry tier. The mixes are easier to judge once you can see each pure model on its own.

The same product, priced four ways

Example: a fictional product, Ledgerly, helps small teams track expenses. Round numbers throughout. The assumptions are the same for every model so that only the pricing changes:

The average customer has 5 users.
The average customer records 600,000 expense events a month.
Customers come in three sizes: 60% small, 30% medium, 10% large.
For freemium, “customers” means sign-ups, and 4% of sign-ups upgrade to a paid plan.

The four price lists:

Per-seat: $12 per user per month. Average customer: 5 x $12 = $60.
Usage-based: $0.10 per 1,000 events. Average customer: 600 x $0.10 = $60.
Tiered: Small $29, Medium $79, Large $199 per month. Weighted average: 0.6 x $29 + 0.3 x $79 + 0.1 x $199 = $17.40 + $23.70 + $19.90 = $61.
Freemium: free plan, then the per-seat price of $12 per user. Per sign-up: 4% x $60 = $2.40.

Monthly revenue	10 customers	100 customers	1,000 customers
Per-seat ($60 per customer)	$600	$6,000	$60,000
Usage-based ($60 per customer)	$600	$6,000	$60,000
Tiered ($61 per customer)	$610	$6,100	$61,000
Freemium ($2.40 per sign-up)	$24	$240	$2,400

The first three rows are set to land in the same place on purpose. The point is not which earns more at the average; it is what happens when customers stop being average.

When a customer grows. Under per-seat pricing, a customer that goes from 5 to 15 users triples its bill. Under usage-based pricing, a customer that triples its events triples its bill, even with the same 5 users. Under tiered pricing, that customer pays nothing extra until it hits the next tier, then jumps from $79 to $199. Under freemium, growth inside the free plan earns nothing until the upgrade trigger fires.

When a customer shrinks. Per-seat revenue falls with layoffs. Usage revenue falls with a quiet quarter. Tiered revenue holds until the customer downgrades, which customers do more slowly than they cut seats. That stickiness is why finance teams like tiers and why customers resent them.

What freemium needs. With a 4% conversion rate, 1,000 sign-ups earn what 40 paying customers earn. Freemium only works when sign-ups are cheap to acquire and cheap to serve, and when the free plan has a wall that paying users hit on their own. Without that wall, the free users stay free. Change the conversion rate to 10% in the table and freemium earns $6,000 at 1,000 sign-ups, the same as per-seat at 100 customers. The model is a bet on volume.

How to choose

The rule that survives contact with real products: charge for the thing that grows when your customer gets more value.

If value grows with the number of people using the product, per-seat fits. Collaboration tools, CRMs, design tools.
If value grows with volume that is independent of headcount, usage fits. Messaging APIs, payments, hosting, monitoring. Twilio prices an outbound US text message at $0.0083 per segment (Twilio pricing, October 2026) because a customer with two developers can send ten million of them.
If customers fall into a few recognizable sizes and dislike variable bills, tiers fit. Accounting software, e-commerce platforms.
If the product is easy to try alone and spreads inside a company, freemium fits as the entry point, with per-seat pricing above it.

Then test the price. Two numbers tell you whether the model is working: what share of customers upgrade or expand within a year, and how many complain about the bill being unpredictable. The second number is the usage model’s weak point.

AI features are pushing vendors away from pure seats

AI assistants changed one assumption behind per-seat pricing: that value scales with people. An AI agent that answers support tickets does work instead of a person, so vendors have started charging for the work. Three dated, sourced examples:

Salesforce moved Agentforce from $2 per conversation to “Flex Credits” metered per action. “Each action costs $0.10 and you only pay for what you use,” the company wrote on its blog on May 23, 2025. Its example: an exchange that would have cost $2 under conversation billing runs 3 to 6 actions, or $0.30 to $0.60.
Intercom sells per-seat plans (Essential $29, Advanced $85, Expert $132 per seat per month) and prices its Fin AI Agent separately, “from $0.99 per Fin outcome”, counted when the customer confirms resolution, stops asking, or Fin completes a workflow (Intercom pricing, October 2026).
Zendesk includes a number of AI “automated resolutions” per agent in its suites and sells additional resolutions on committed and pay-as-you-go rates, priced separately from the seat (Zendesk pricing, October 2026).

The pattern is a hybrid: seats for the humans, usage or outcomes for the AI. If your product has an AI feature that does work on the customer’s behalf, expect the same question from your customers: why pay per person for something that replaces a person?

Common mistakes

Per-seat pricing on a product that punishes adding seats. If inviting a colleague to a shared budget costs $12 more a month, customers stop inviting colleagues, and the product never spreads. Why it happens: per-seat is the default everyone copies. Instead: make viewers or light users free (Airtable charges only editors; Vercel’s Pro plan includes unlimited viewer seats) or charge by the thing that is used, not by the people who look at it.
Usage-based pricing with bills the customer cannot predict. A finance team that cannot forecast a bill will cap it, or cancel. Why it happens: pure metering looks fair to the vendor. Instead: sell committed usage with a discount, publish a calculator, and alert customers before they cross a threshold. Datadog’s price list shows the shape: a lower rate billed annually ($15 per host per month) and a higher rate on demand ($18), per its pricing page in October 2026.
Freemium with no upgrade trigger. A free plan that does everything a small team needs is a charity, not a funnel. Why it happens: founders fear scaring users away. Instead: pick one limit that a serious user will hit (history, members, volume) and make the upgrade path obvious. Slack’s free plan keeps 90 days of message history (Slack pricing, October 2026); the wall is the archive.
Pricing from your costs instead of the customer’s value. Cost tells you the floor, not the price. Why it happens: costs are the only number a new company knows for sure. Instead: find what the customer saves or earns, and price a fraction of it. Then check the floor.

Sources

Salesforce, Flex Credits announcement on the Salesforce blog, May 23, 2025.
Intercom, Pricing; Zendesk, Pricing; Twilio, US SMS pricing; Datadog, Pricing; Slack, Pricing; Airtable, Pricing; Vercel, Pricing. All checked October 2026.

In three sentences

SaaS products charge per seat, per unit of usage or per plan tier, often with a free entry plan, and the right model is the one that charges for whatever grows when the customer gets more value. On an average customer the three paid models earn about the same, so the difference shows up only when customers grow, shrink or stay free. AI features that do work on the customer’s behalf are pushing vendors toward hybrids, with seats for people and metered prices for the AI.