SaaS Pricing Models and How to Move Between Them

Your pricing page is a guess about how your customer's bill is supposed to grow.

Five structures cover almost every SaaS product: flat rate, tiers, per seat, usage, and freemium. Which one you pick comes down to a single question, and it is not what your competitors charge. Does your heaviest customer get ten times the value of your lightest one? If the honest answer is yes, you need a model that can bill for it. If it is no, seats or tiers will serve you better and be far easier to explain.

Most founders answer that question from the pricing page of a company four stages ahead of them, then spend a year defending a structure their buyer cannot repeat to a colleague.

So each model here comes with the numbers its best-known user publishes today, the failure that shows up around month nine, and the migration path, which is the part the guides skip.

The SaaS Pricing Models, Priced Side by Side

The table is the summary. The sections after it are where the decisions get made.

Every figure in that column came off the company's own page this week: Basecamp, Notion, Slack.

Two columns decide more than the other three. Revenue shape is what the model does to your forecast. Where it breaks is what it does to your renewal calls.

Read down the breakage column and the trade is obvious: a model that grows revenue without a conversation also grows the chance of a bill somebody resents. A model that caps the bill stops scaling with the customer.

Here is the decision, in order. Do it on your own data.

1. Export 90 days of usage per account. Sort by units consumed and compare your top account to your median. Ten to one or wider, and usage belongs in the model.
2. Count how many people joined each account in the same window. If the median account adds people every quarter, per seat grows on its own.
3. Write the sentence a customer will say when they open the bill. "It costs what it costs" is flat. "We added three people" is per seat. "We sent 40,000 more messages" is usage. If none sounds natural, the model is wrong.
4. Pick one backbone and stop. Add a second dimension only when the math demands it, because two axes is a pricing page and three is a spreadsheet with a login.

That fourth rule is where hybrid models come from, and it is worth seeing one at work. Notion charges per member for the workspace and then charges again for agent credits, because the second meter measures something seats never could: how much the software did on its own. The customer can hold both numbers in their head at once. Add a third axis, say storage plus seats plus credits plus a support tier, and the buyer stops estimating and starts asking for a call.

Flat and Tiered: Capping the Bill Instead of Metering It

Basecamp publishes a clean example of both at once, and it changed recently enough that the guides have not caught up.

Its page on 23 October 2026 lists four packages. Freelancer at $29 a month for 3 active projects and up to 5 users. Pro at $79 for 8 projects and 10 users. Pro Plus at $99 for 12 projects and 15 users.

Then Pro Unlimited at $199 a month billed annually, or $329 monthly, for unlimited projects and unlimited people. No per-user fees anywhere on the page.

That last clause is the trick. Basecamp sells access, then caps the account by project count and headcount, so the cap does the growth work a seat price would normally do.

The guides get this wrong in a way that costs you a decision. Moesif's guide, updated May 2026, puts Basecamp Pro Unlimited at $299 a month billed annually and $399 monthly. That was the price before the current one. I trust Basecamp's own page from this week over a secondary source from five months ago, and I would trust your own page over both.

Tiering fails when the middle tier is designed to be outgrown rather than chosen. A team on Pro at $79 needs a ninth project, gets pushed to $99 for four more projects and five more seats, and feels punished for succeeding. Nine times out of ten the fix is the cap, not the price.

1. Write down your limits, one line each. Projects and seats, plus whatever else a customer could run out of.
2. Find the 90th percentile of your current customers on each limit. That is your top tier's cap, not a number you invent.
3. Price the middle tier near twice the low tier and give it at least twice the capacity. A middle tier that only adds features reads as a toll.
4. Walk the upgrade path looking for a false ceiling. If a customer can hit a cap and gain nothing by moving up, the cap is wrong.

The failure to watch is cap gaming: the customer who deletes three projects a month to stay on the cheaper plan. That is a customer doing unpaid admin to avoid paying you, and it usually means the cap sits on the wrong unit.

Per Seat: A Model That Stalls When Hiring Stalls

Per seat is the most legible model in software and the one most exposed to the customer's headcount.

Slack lists Pro at $8.75 per user a month billed monthly, or $7.25 billed annually, with Business+ at $18 monthly or $15 annually. Notion runs Plus at $10 per member and Business at $20. Both sell the same axis: how many people are in the room.

Notion's answer to the model's worst instinct is worth copying. Guests are free, and removing a member frees a seat you can hand to somebody else inside the same billing period, so swapping people costs nothing while the billable count holds.

Slack priced the behaviour itself. From its Fair Billing Policy on the pricing page:

Members who take an action in Slack at any time within a 28-day period will be considered active for billing purposes. If someone you've already paid for becomes inactive, we'll add a prorated credit to your account for the unused time.

That one definition answers the objection that kills per-seat deals inside big companies. IT can provision 400 accounts without promising to pay for 400, because inactive is defined, measured, and credited.

The failure shows up quietly. Your customer signs 20 seats, hires nobody for three quarters, and revenue from that account goes flat while your support load does not. Meanwhile the buyer coaches the team not to invite the contractor, because every invitation is money. You have made your own product smaller.

Run the arithmetic on one account to see how much smaller. Forty seats on Slack Business+ at $18 is $720 a month. If 22 of those people touch the product in a 28-day window, the Fair Billing Policy credits the other 18, and the bill lands at $396. The vendor keeps the account in a 200-person rollout that would otherwise have leaked to a competitor. Without the credit, the same rollout gets capped at a pilot, and the vendor earns $396 in the best case because the buyer is protecting a budget line.

1. Pull paid seats and activated seats for your last 20 accounts. Activated means the definition you are willing to write down.
2. Divide one by the other. Below about 60%, your customers are paying for chairs nobody sits in, and that is the renewal conversation coming at you.
3. Move the billable unit to active seats with a 28-day window and a prorated credit, then put the definition on the pricing page rather than in a help article.
4. Re-check a week after the change. If paid seats fell by less than the dormant seats you found, the definition is too tight and you are giving revenue away.

Usage: The Bill That Can Surprise You

Usage tracks value better than any model here, and it is the one most likely to sit on billing software that cannot keep up.

Notion already publishes a metered price beside its seats: custom agents are free to try, then $10 per 1,000 monthly credits. That is usage bolted to seats, and it works because the unit is countable and visible, and small enough for a customer to estimate.

MGI Research's note from 30 March 2026 puts the operational gap bluntly. Fewer than 10% of organizations have the skills and tools to manage usage-based pricing today. More than 60% of software companies plan to deploy usage billing within 12 months. A billing stack that fails after go-live costs $1M to $5M or more to remediate.

Those figures are a warning about sequencing. Most of the market is about to build metering it is not staffed for.

The requirements are narrower than the tooling market suggests.

1. Every billable event lands in your metering system within seconds. A delayed event becomes a disputed invoice later.
2. The customer sees their own usage before the invoice. A visible counter prevents more churn than any discount.
3. Usage reconciles to the invoice line by line, so finance can answer the only question customers ask: why is this number this number.

The unit is the other half of the work, and picking it wrong is worse than picking a flat fee. A workflow tool that bills per run gets punished by retries, because a run that failed three times cost the customer nothing and still shows up on the invoice. Bill per completed task instead, count retries as zero, and the same month of activity produces a number the buyer recognizes. Write the definition in one sentence on the pricing page before you build anything, because the sentence is what your engineers will implement.

Then cap the damage. A hard ceiling you can raise is friendlier than a soft warning nobody reads. If your unit is $0.004 and a customer normally runs 400,000 units, their bill is $1,600. At 2 million units it is $8,000, which is the invoice that ends the relationship. Put the ceiling at 1.5 times their trailing average and route anything above it to a person.

Usage also decides what you can spend to acquire. What customer acquisition cost has to cover is the calculation that should sit beside the unit price, because a model whose revenue arrives monthly and whose bill can spike needs a payback window you can actually survive.

Freemium: Do the Arithmetic Before You Open the Door

A free tier is a marketing budget with a support queue attached, and the wall is the product decision.

Slack's free plan gives 90 days of message history. Basecamp's gives one project and three users. Neither wall is arbitrary. Slack's hits when a team starts relying on search across old decisions. Basecamp's hits when a second project appears. Both sit where the customer has already felt the value and can see what they lose.

The arithmetic takes an afternoon and almost nobody runs it. Say 10,000 accounts sit on your free tier and hosting plus support costs $0.35 per free account a month. That is $3,500 a month before one of them pays.

Your paid plan is $20 a month and your free-to-paid rate is 3%. Three hundred accounts convert, worth $6,000, so the free tier clears $2,500 and pays for itself. Push that rate to 1.5% and the same tier loses $500 a month, quietly, for as long as you leave it up.

1. Cost the free tier: hosting and storage, plus support hours at a real hourly rate, divided by free accounts. Leave your whole infrastructure bill out of it.
2. Pull your free-to-paid rate from the last two quarters. If you launched recently, use the number you have and mark it provisional.
3. Multiply rate by price. Break-even is that product beating the cost per free account, and the formula is short enough to keep in a note.
4. Move the wall, not the price. Tightening what free includes usually converts better than raising what paid costs.

Where the wall goes matters more than how tall it is, and there is a way to test it without guessing. Gate the feature that appears in the fifth week of real use, not the first. Then watch the free cohort for 30 days after the gate: the number you want is how many free accounts hit it, because a wall nobody reaches is a wall that converts nobody. If fewer than a fifth of free accounts touch the gated feature in a month, the wall is sitting on a feature your customers do not need yet.

The signup page carries the rest of the load, since a free tier spends its budget on strangers. Conversion tests that survive low traffic covers the testing that works when you have hundreds of visitors rather than hundreds of thousands.

How to Move From One Model to Another

Changing structure is a heavier job than setting it, and the cost is not the pricing page. It is everything downstream of the invoice.

MGI's $1M to $5M remediation figure is the reason to phase rather than switch. Billing configuration, proration, revenue reporting, and the dashboards finance already trusts all have to agree with the new model on the day it goes live.

Four steps, in this order.

1. Pick the new value metric from your own data. One unit, countable and visible to the customer.
2. Price the new model so your median account pays what it pays today. Build the ladder backwards from that number.
3. Run it on new signups only for 60 days. Compare conversion and revenue per account across the two cohorts.
4. Grandfather everyone else, give 30 to 60 days of notice, and show the old bill beside the new one in plain language.

Worked example. You charge $29 flat and move to a $19 base plus $0.004 per unit. The median account runs 1,800 units, so it lands at $26.20, just under what it paid. The heavy account runs 9,000 units and lands at $55, which is the account you were subsidizing all along. New signups convert at the same rate and revenue per account rises without a price increase on the base.

That move is also a positioning decision, which is why it belongs next to the four decisions a go-to-market strategy has to make rather than in a finance deck on its own.

Whatever you charge has to leave room for being seen, and that is the line pricing pages never show. One of the places we price reach is our own: The Board is a leaderboard of brands, every listing on it is a paid placement ranked strictly by what has been paid and by nothing else, and rank decides how many impressions a listing gets. Rank 01 costs $206 right now, and a dollar over the holder takes it. That is a fixed line you can put in a budget, priced in public, which beats an auction you cannot see.

The Number to Type First

Open your billing system and pull three figures for the last 90 days: revenue per account, units consumed per account, seats activated per account.

Sort each one. If revenue tracks units, you are already pricing usage whether the page admits it or not. If it tracks seats, your growth is a hiring proxy. If neither moves, you are on flat rate and your best customers are subsidizing your worst.

Then write the sentence your customer will say at renewal. Everything above is a way of making that sentence true.

If you want the neighbouring work, SaaS marketing when the team is three people covers the demand side and growth marketing without a growth team covers the rest of the week. The SaaS category is where this one lives.

A price is a claim about who your customer becomes. The model is how you get paid while they become it.