Usage-Based Pricing: 43% Have No Monthly Starting Price

Ask a usage-based vendor what it costs and you get a rate card, a calculator, or a quote form. What you rarely get is a number. We pulled every product in our directory that prices by consumption and checked what each one actually prints for a starting price, and the headline is blunt: 77 of 178 have no monthly starting price in our data at all. That is 43%, and it changes how you should design (and read) usage-based pricing.

Everything below is current pricing as of the directory’s latest fetch, taken from the product listings and two full pricing records. The usage-based label is our classification, so treat the shares as a snapshot.

How big is usage-based pricing right now?

Across the directory’s latest stats, 178 of 1,194 products are labeled usage-based, about 15%. Hybrid (356) and per-seat (328) are larger, with flat-rate at 153 and custom at 75. Our earlier cuts of the model mix, in SaaS pricing models and seat-based pricing, put per-seat first. The labels move as pages change and as our classifier gets sharper, so the safe reading is directional: the two models that bill on consumption, usage-based and hybrid, together cover far more of the directory than per-seat does.

The boundary between them is blurry. Resend, covered below, is labeled usage-based here yet sells three monthly bundles with an overage rate on top. Ask whether the buyer pays for consumption or pays a subscription with a consumption ceiling. They behave very differently on a bill.

The starting price is missing for 43%

Of the 178 usage-based products, 101 carry a numeric starting price and 77 do not. Typical cases are a pure rate card (Twilio, Fly.io, Replicate and Scaleway have no tier table at all), a pay-as-you-go plan that starts at zero and bills past a free allowance, or a quote. PostHog is the clean example: our record says the pay-as-you-go tier has no fixed price and users pay only for usage above the free limits.

The 101 that do print a number need careful reading too. Eleven of them are under $1, which is a unit rate (MongoDB Atlas at $0.011, Courier at $0.005, Upstash at $0.2) rather than a monthly fee. Drop those and the median starting price across the remaining 90 is $22 a month, against a directory-wide median of $25. Fifty-two of the 90 start at $25 or less, so usage-based pricing is cheap to enter. At the top, 15 start at $100 or more, including Plivo at $1,000, Chameleon at $750, Maxio at $599 and Telnyx at $500. Those are commitment-style entry plans, not pay-as-you-go.

So the entry point is bimodal. Either there is no monthly number to anchor on, or there is a low one ($22 median) that tells you almost nothing about what you will pay at scale. That gap is where buyer anxiety lives.

Free plans do the selling

Usage-based products lean on free plans more than the directory as a whole. Of the 178, 95 offer a free plan (53%), against 45% across the full directory, while 89 offer a free trial (50%) against 52%. Cross the two and the shape is clear:

• Free plan only: 59 (33%)
• Free trial only: 53 (30%)
• Both: 36 (20%)
• Neither: 30 (17%)

This fits the model’s logic. If price scales with consumption, the cheapest way to prove value is to hand out a metered allowance and let the first bill arrive only after the customer has built on you. Our freemium vs free trial cut found the same category-driven pattern across the wider directory.

Where it clusters, and where the number hides

Six categories hold 78 of the 178: Monitoring (18), Analytics (15), AI (14), Databases (11), Developer Tools (10) and Data (10). They differ a lot in how openly they price.

Databases and Developer Tools are the most legible: 8 of 11 and 7 of 10 print a starting price. AI is the opposite. Only 3 of 14 show a monthly figure (Manus at $20, Weaviate at $45, Modal at $250), even though 10 of the 14 offer a free plan. Data is similar, with 3 of 10 priced. Payments is the starkest: nine products, five free plans, no trials, and a single starting price (Superwall at $49).

Two smaller pockets are worth noting. All five identity products (Frontegg, FusionAuth, Stytch, SuperTokens, WorkOS) offer a free plan. And in cloud infrastructure, only DigitalOcean offers a free plan and none of the five offers a trial, so those buyers have to commit to a price list without trying anything first.

Teardown 1: Resend sells bundles, then charges overage

Resend, an email API, is the cleanest example of usage-based pricing that behaves like a subscription. Per our record on Resend’s pricing page, the ladder is Free at $0 for 3,000 emails a month (capped at 100 a day), Pro at $20 for 50,000, Scale at $90 for 100,000, and a Custom tier with no price. Extra emails cost $0.90 per 1,000 on both Pro and Scale. Add-ons include extra domains at $20 a month, dedicated IPs at $30 and single sign-on at $150.

Run the arithmetic and something odd shows up. Pro includes 50,000 emails for $20, which is $0.40 per 1,000. Scale includes 100,000 for $90, which is $0.90 per 1,000, exactly the overage rate. Above 100,000 emails, Pro costs $20 plus $0.90 for every extra thousand beyond 50,000, and Scale costs $90 plus $0.90 for every extra thousand beyond 100,000. The difference is a constant $25 in Pro’s favor at any volume. On email count alone, Scale never wins.

What Scale actually sells is everything around the volume: 1,000 domains instead of 10, a dedicated Slack channel, 10 webhook endpoints instead of 5, and 500 AI credits a month instead of 100. That is a smart structure, because it moves the upgrade decision off the meter and onto capabilities, which is where buyers are least price-sensitive. It is also a trap for a buyer who assumes the next plan up is a volume discount. This calculation assumes the listed overage rate applies the same way on both plans, so check it against the vendor’s page before you budget on it.

Teardown 2: PostHog publishes the free tier, not the price

PostHog sits at the other end. Our record shows two tiers: Free at $0, with the full monthly free tier on every product, unlimited team members, one project and one-year data retention, and Pay-as-you-go with no fixed price, unlimited usage, six projects, seven-year retention and email support (Slack over $2,000 a month). Discounts are listed for startups and non-profits, with bigger ones for annual commitments above a certain spend. See the full record on PostHog’s pricing page.

There is no sticker price, and for this buyer that is fine. The free allowance is the product’s pricing page, and the paid plan is a promise that the meter is the only cost. It works because the metered units (events, recordings, requests) are things an engineer can estimate before signing up. Where the unit is vague, the same structure turns into a black box.

What to copy, and what to skip

If you sell on consumption, the directory data points to four decisions.

First, give the buyer one number to anchor on, even if it is only the cost of a realistic month. A median entry price of $22 means a buyer expects to start small. A missing price means they expect a catch.

Second, pick a unit a customer can forecast. Emails, events and API calls pass that test. Credits and compute-seconds often do not, and it is no accident that AI and Data, where only 3 of 14 and 3 of 10 products show a monthly figure, lean on the vaguest units.

Third, if you bundle, make the upgrade about capabilities, as Resend does, and be honest that the bigger plan is not a volume discount. If you do not bundle, say so plainly, as PostHog does.

Fourth, use a free allowance as your demo. Over half of usage-based products do, and the ones that do not need a stronger reason for asking a stranger to commit before they have built anything. For more on testing that choice, see SaaS pricing benchmarks. To see how your own page reads to a first-time buyer, run it through the pricing page analyzer.

The one move worth making this week

Write down what a customer at 1x, 10x and 100x your typical usage would pay, using only the numbers on your public page. If you cannot do it in two minutes, neither can a buyer. Fix the page first, and only then argue about the meter.

Outmano is an AI-powered competitive intelligence platform that watches competitor pricing pages, flags every change and explains what it means, so you hear about a rival moving from seats to credits when it happens. See how it works at outmano.com or read about competitor pricing monitoring.

Frequently Asked Questions

What is usage-based pricing?

Usage-based pricing charges customers for what they consume, such as emails sent, events tracked or API calls made, rather than for seats or a flat fee. Many vendors pair it with a free allowance and bill only above it. In our directory, a majority of usage-based products also offer a free plan.

Is usage-based pricing cheaper than per-seat pricing?

At small scale it usually is, since a typical entry price among products with a monthly figure is about $22. At large scale it depends on how fast your consumption grows, because a metered bill rises with volume while a seat price rises with headcount. Estimate the bill at 10x your current usage before you decide.

Why do usage-based pricing pages often hide the starting price?

Because the price depends on volume, and many vendors would rather show a calculator or free allowance than a number that anchors buyers low or high. In our data, 43% of usage-based products have no monthly starting price on record. The cost to the vendor is buyer uncertainty, which shows up as slower evaluations.

What is the difference between usage-based and hybrid pricing?

Usage-based pricing bills mostly on consumption, while hybrid pricing combines a fixed fee, usually seats or a platform charge, with a metered component. In practice the line blurs: a vendor selling monthly bundles with overage can fit either label. Check whether the buyer pays a subscription first or a meter first.

Which software categories use usage-based pricing the most?

Monitoring, Analytics, AI, Databases, Developer Tools and Data hold 78 of the 178 usage-based products in our directory. Databases and Developer Tools are the most open about starting prices, while AI and Data are among the least.