Pricing Your Product on a Research API

Pricing a product built on a research API comes down to three numbers: the input cost you pay the API for each completed interview, the resale price you charge your own customers, and the spread between them. That spread is your business. The input cost is transparent and usage-based, the resale price is yours to set, and the model you wrap around them should match how your customers already buy — cost-plus, value-based, tiered, usage, or subscription. Because the input cost is a known per-interview number, you can model the entire economic picture before you write a line of integration code.

This guide is written for builders and platforms embedding qualitative research and selling it on to their own customers. It is the pricing companion to the research infrastructure platform, which explains the stack you are building on, and to the pillar on building customer research into your product, which explains the strategy. Here the focus is narrower and more practical: how to turn a per-interview input cost into a product your customers pay for at a margin you can defend.

The three numbers behind every research-API business

Every product built on a research API has the same underlying economic shape, no matter what the front end looks like. There is what you pay, what you charge, and what you keep.

1. Input cost — what the API bills you per completed, usable interview. This is your cost of goods for the research feature. On a usage-based platform it is a known per-interview number rather than a bundled license fee, which is what makes the rest of the math tractable.
2. Resale price — what you charge your own customer for the same research, expressed in whatever unit your product uses. This is entirely yours to set. Your customer pays for the outcome your product delivers, not for the raw interview.
3. Spread — the difference between the two, per unit. Multiply it by volume and subtract your own delivery costs, and you have the contribution your research feature makes to the business.

The mistake most teams make is skipping straight to the resale price without pinning down the input cost first. If the input cost swings unpredictably — because you are billed for attempts, dropped calls, or thin sessions — the spread swings with it, and no pricing model can rescue a floor you cannot forecast. Nail the input cost, and every model below becomes a decision about positioning rather than a gamble on cost. For a deeper treatment of the cost structure itself, the sibling breakdown on research API economics works the numbers in detail.

What pricing model should you use for embedded research?

There is no universally correct model. The right one mirrors how your customers already buy from you, because that is the friction-free path to revenue. A workflow tool that sells seats should not suddenly meter research by the interview; a developer platform that sells credits should not force a flat subscription. Below are the five models that fit embedded research, and where each one earns its keep.

Cost-plus is the simplest to explain and the easiest to defend to a procurement team, which is why it dominates agency rebilling and transparency-first relationships. Its weakness is that it teaches your customer exactly what you pay, which caps how much value they will accept you capturing.

Value-based pricing detaches the price from the interview count entirely and ties it to the decision the research informs. One well-run study that stops a bad launch is worth far more than the interviews it took to run it. This model produces the strongest margins, but only if your product can articulate and prove the outcome — otherwise the customer defaults to counting interviews.

Tiered and seat models suit workflow SaaS where research is one capability inside a broader product. You fold a research allowance into each plan, and the recurring nature smooths revenue. The discipline is sizing the allowance so a heavy user does not quietly turn a profitable seat into a loss.

Usage and credit models fit pay-as-you-go and developer-facing products where consumption varies widely between customers. The spread scales cleanly with volume, and customers appreciate paying only for what they draw down. The trade-off is that a visible meter makes buyers cost-conscious and makes your own revenue lumpier.

Subscription models work when the research is continuous — always-on brand tracking, a monitoring feature, a standing panel. You capture recurring, compounding revenue, but you take on the risk of usage running ahead of the flat fee, which puts the predictability of your input cost front and center.

How do you model the unit economics?

Modeling the economics is a bounded exercise once you have a real input-cost number. Work through it in order, and you will know your margin before you build anything.

1. Start from the input cost per completed interview. On the Professional plan that is $25 per quality interview, and only interviews that pass the quality bar are billed. This is your cost of goods, and it is a single known number rather than a range.
2. Choose your billing unit. Decide whether your customer pays per interview, per study, per seat, or per credit. The unit should match the pricing model you selected above, not the unit the API happens to bill you in.
3. Set your resale price. This is the number that is entirely yours. Anchor it to the value your product delivers and to what your customers already pay you for adjacent features, not to a fixed multiple of the input cost.
4. Compute the spread and gross margin. Subtract input cost from resale price for the per-unit spread, then divide by the resale price for gross margin. The worked example below shows the shape.
5. Layer in your own delivery costs. Support, integration maintenance, storage, and the engineering time to keep the feature running are real costs that sit between gross and net margin. Subtract them before you call a price profitable.
6. Forecast at volume. Because quality-only billing gives you a predictable input-cost floor, you can extend the model to one, ten, and a hundred studies with confidence. This is where you discover whether a generous seat allowance or a thin cost-plus markup survives scale.
7. Pressure-test against willingness to pay. Hold the modeled price against what your customer will accept. If the value-based ceiling sits well above your cost-plus floor, that gap is the margin headroom your positioning can capture.

Here is the per-unit math made concrete. The resale price shown is illustrative — a 2x example, not a recommendation — because the resale number is yours to choose.

The point of the table is not the multiple. It is that every cell is derivable the moment your input cost is a fixed number. Change the resale price and the spread moves; the input cost stays put. That stability is the whole reason a usage-based, per-interview foundation is easier to build a business on than an opaque license where your cost of goods is a mystery until the invoice arrives.

Why quality-only billing makes your input cost predictable

Quality-only billing is the mechanism that turns the model above from an estimate into a forecast. Interviews are billed only when they pass automatic Length, Depth, and Coverage checks, so a session that ends early, wanders off-topic, or fails to cover the guide does not count against you. You pay for signal, not attempts.

For a product that resells research, this matters more than for an internal research team. When you are marking up interviews to your own customers, an unpredictable input cost is a hole in your margin that widens exactly when volume grows. If you were billed for every attempt, a spike in dropped calls or low-quality sessions would silently compress your spread across every customer at once. Quality-only billing removes that variance: your cost of goods per usable interview is a constant you can put in a spreadsheet and defend to a finance team.

That predictability is also what lets you commit to a price before you build. You do not need a live integration and three months of production data to know your floor — you know it from the published per-interview rate and the quality-only guarantee. The same property makes agent-driven products viable, where an autonomous workflow triggers studies on its own; agentic research depends on costs that stay bounded no matter how the agent behaves, and quality-only billing supplies that bound.

Validate the loop on the free Starter tier before you scale

You do not have to take the model on faith. The free Starter tier exists to let you prove the entire economic loop with real interviews before you write production code or turn on billing.

1. Sign up for Starter. It is $0 per month with 3 interviews included and no credit card, which is enough to run a genuine study rather than a demo.
2. Run one real study end to end through your integration path. Create the study, recruit, and retrieve results the way your product eventually will, so you are testing the actual loop and not a happy path.
3. Confirm the structured results map to your value. Check that the preference splits, ranked themes, and verbatim quotes returned as JSON are what your customer is paying for. If the output does not support your resale price, discover it now.
4. Measure your true input cost against the quality bar. Because only completed, usable interviews bill, your Starter run tells you exactly what your cost of goods will be per usable interview at scale.
5. Model the spread at one, ten, and a hundred studies. Extend the unit economics from your real numbers, not assumptions, and see where each pricing model holds or breaks.
6. Write production integration code and turn on billing. Only after the loop is proven do you commit engineering time — by which point your pricing is grounded in evidence, not hope.

This sequence inverts the usual order, where teams build first and price later. When the input cost is transparent and a free tier lets you run the real loop, pricing becomes the first thing you validate rather than the last thing you guess. The research API integration checklist covers the engineering phases that follow once the economics check out.

Common mistakes when pricing resold research

A handful of errors show up repeatedly when products first price a research feature. Each one is avoidable once you have named it.

1. Pricing off attempts instead of completed interviews. If your model assumes you pay per attempt, you will either over-price defensively or get squeezed when quality varies. Build on a completed-interview cost so your floor is real.
2. Anchoring your customer to your raw input cost. Cost-plus is fine as a model, but exposing the input number in every context trains customers to see your margin as overhead. Reserve full transparency for relationships that reward it.
3. Forgetting your own delivery costs. Support, maintenance, and storage sit between gross and net margin. A 50% gross spread that ignores them is not the number you take to the bank.
4. Choosing a model that fights how your customer buys. A seat-based SaaS that bolts on per-interview metering adds friction at exactly the moment of value. Match the research model to the buying motion you already have.
5. Publishing volume discounts you have not secured. Never advertise wholesale or bulk rates you have only assumed. Standard per-interview pricing is public and predictable; volume and partnership economics are set directly with our partnerships team, so base your published prices on numbers you have agreed.

How does User Intuition price research you can resell?

User Intuition prices for embedding, not for one-off buying. The per-interview cost is transparent and usage-based — $25 per quality interview on the Professional plan, with studies starting at $150 — and only interviews that pass automatic Length, Depth, and Coverage checks are billed. That combination gives you a known cost of goods and a predictable floor, which is precisely what a product needs to set its own prices with confidence. The free Starter tier, at $0 per month with 3 interviews and no card, lets you validate the whole loop before you spend.

The foundation underneath is the same one described on the research infrastructure platform: recruitment from a 4M+ vetted panel across 50+ languages, AI-moderated interviews that ladder to the decision driver, and automated analysis returned as structured JSON. Your product builds on that and owns the customer relationship. Participant satisfaction averages 98%, and the platform carries 5/5 ratings on G2 and Capterra, which are proof points you can pass through to your own customers as evidence of the quality behind your feature.

For standard integrations, the published per-interview pricing is all the math you need — the input cost is known, the resale price is yours to set, and the spread is your business. For high-throughput resale, volume and partnership economics are set directly with our partnerships team rather than through a public rate card, so your unit economics rest on agreed numbers rather than assumptions. That is the difference between building research infrastructure and pricing a product on top of it: the panel, the moderation engine, and the analysis pipeline already exist at a transparent, usage-based rate, so your team spends its energy on the part that is yours — the product your customers pay for, and the margin you build into it.

Start free, model your margin before you build. Three interviews, no card, transparent per-interview pricing. Get an API key → · See the infrastructure model →