Paid discovery vs a free scoping call: which to choose and when
What is the difference between paid discovery and a free scoping call?
A free scoping call produces a conversation, a rough range and a proposal. Paid discovery produces artefacts you own: a validated use case, a data assessment, an architecture, an evaluation plan and a fixed-price build quote. Choose by how much uncertainty your project is carrying.
A free scoping call produces a conversation, an indicative price range and a proposal written from what you managed to explain in an hour. Paid discovery produces artefacts you own: a validated use case, a data assessment, an architecture, an evaluation plan and a build quote that can be fixed. The difference is evidence.
Both are legitimate, and we offer both. This article sets out what each actually delivers, a side-by-side of the two, the situations where paying would be a waste of your money, the situations where skipping it costs a quarter, and what discovery costs at published Eazyware prices.
What a free scoping call actually is
A free scoping call is a sales qualification conversation that is genuinely useful to both sides. In forty-five to ninety minutes a competent team should be able to tell you whether your problem is a known shape, roughly what similar work has cost, which parts of your description worry them, and whether anyone should be spending money on this at all. A good one includes an honest no.
What it cannot do is reduce uncertainty. Nobody discovers in an hour that your customer data lives in three systems with conflicting identifiers, that half your documents are scanned images, or that the workflow you described has an undocumented exception path handling a third of volume. Those findings arrive later, usually mid-build, and they are what turns a firm price into a change request.
What paid discovery actually is
Paid discovery is a short, scoped engagement with named deliverables and a fixed fee. At Eazyware it takes three shapes. A ten-day AI Discovery Sprint answers whether a specific use case is worth building and produces the decision, the data findings and a build plan. A two-to-four-week AI product strategy and use-case discovery engagement ranks a portfolio of candidate use cases rather than testing one. A three-week AI POC Sprint goes further and proves the hardest technical assumption on your real data.
The point of paying is not access to smarter people. It is that paid time can be spent on your systems rather than on your description of them: reading schemas, sampling documents, sitting with the person who does the work today, running a model against a hundred real records. Joel Spolsky's argument for writing the spec before the code applies unchanged to AI work: problems that cost an afternoon to fix in a document cost weeks to fix in a shipped system.
A side-by-side: what you get for what you pay
| Dimension | Free scoping call | Paid discovery |
|---|---|---|
| Duration | 45 to 90 minutes | Ten days to four weeks |
| Who does the work | One or two senior people, on your account of the problem | A working team, on your actual data and systems |
| Access to your systems | None | Schemas, samples, stakeholders, sometimes a sandbox |
| Output | A proposal and a price range | Documents and, in a POC sprint, running code |
| Who owns the output | The vendor's proposal template | You, including findings that say do not build |
| Price confidence | Wide range, caveated | Narrow enough to fix a price and a date |
| Comparable across vendors | Hard; every proposal assumes something different | Yes, because the scope is written down |
| Honest exit | Politeness stops most vendors saying no | A documented recommendation not to proceed |
| Cost | Nothing | $3,250 to $10,500, often credited to the build |
When a free call is genuinely enough
Take the free call and stop there when the work is a known shape and the stakes are bounded. A WhatsApp support bot over an existing help centre, a straightforward integration, a copilot over a documented API: these have been built hundreds of times, and a vendor who has done them can price them from a conversation. If your budget is under roughly twenty thousand dollars, spending a fifth of it on discovery is poor arithmetic.
A free call is also the right move when you are still choosing a partner. Three scoping calls cost you three hours and tell you a great deal about who listens, who asks about your data before your budget, and who says no. Paying three vendors to discover the same thing is not diligence, it is indecision with an invoice. The signals worth watching are listed in questions to ask a vendor before you sign.
When you should pay for discovery
Pay when the uncertainty sits in your systems rather than in the technology. If nobody can say how clean the data is, how many exception paths the workflow has, or whether the legacy system exposes an API at all, no proposal built on your description will survive contact with reality. Pay when the build is large enough that a twenty per cent scoping error is bigger than the discovery fee, which for most teams means anything above forty thousand dollars.
Pay when you need the output for someone else: a board, a risk committee, a regulator, or a procurement process that requires comparable bids. A discovery document you own can be shown to three build vendors to get quotes on identical scope. A vendor's free proposal cannot, because it was written to win the work rather than to describe it neutrally.
What does discovery cost, and what do you get back?
The AI Discovery Sprint is $3,250 or ₹2,00,000, fixed, and credited against the build that follows, so choosing to proceed makes it free in effect. AI product strategy and use-case discovery is $4,250 or ₹2,80,000 across two to four weeks. An AI POC Sprint runs $6,250 to $10,500, or ₹4,00,000 to ₹6,80,000, over three weeks and ends with working code against your data. Every published figure sits on the pricing page, and rough build ranges are on the estimate page.
Set that against what it buys. A build mis-scoped by a quarter on a $45,000 programme is an eleven-thousand-dollar problem plus the schedule damage, and mis-scoping is the normal outcome when nobody looked at the data. What a discovery engagement actually costs, including your own team's time, is broken down in AI discovery sprint cost in 2026.
The order that works: both, in sequence
These are not competing purchases. The free call decides whether to talk to this vendor at all; paid discovery decides whether to build the thing and at what price. Running them in that order is normal and correct, and any vendor who will not take a free call first is telling you something.
The transition point is specific. When a scoping call ends with the vendor naming three assumptions that would change the price materially, and neither side can resolve them by talking, that is the moment discovery has become the cheaper option. If the call ends with no such assumptions, discovery would be theatre.
How to decide: five questions
- Can you name the data the system will read, and has anyone looked at a sample? No on either count means pay.
- Is the workflow documented, including the exceptions? If the only documentation is the person who has done it for nine years, pay.
- Would a twenty per cent price error matter? On a small build it is annoying. On a large one it is a re-approval.
- Do you need to compare vendors on identical scope? Then you need a specification you own, not three proposals.
- Is the hardest part technical or organisational? Technical uncertainty needs a POC sprint; organisational uncertainty needs a strategy engagement instead.
- Has an earlier attempt already failed? A second attempt priced from a conversation usually fails the same way.
When paid discovery is the wrong choice
Discovery is wrong when it is used to defer a decision. Teams that cannot get approval sometimes buy a sprint to look busy, then buy another. If the output would not change what you do next, do not commission it. A discovery engagement that cannot end in a recommendation against building is not discovery.
It is also wrong when the vendor will not credit it and will not let you take the artefacts elsewhere. Discovery whose output is a proposal only that vendor can execute is a sales document you paid for. Ask before signing whether the findings, the architecture and the evaluation plan are yours to use, and ask the same question about the code, prompts and infrastructure of any build that follows.
What you lose by skipping it
The reversal cost is asymmetric, and it lands late. Skipping discovery does not usually surface as a failure; it surfaces as change requests at week six, when the first real data reaches the system. By then you have committed a team, told your stakeholders a date, and the cheapest remaining option is to reduce scope, which is how projects quietly become the thing nobody wanted.
Buying discovery you did not need costs you the fee and about two weeks of calendar, and it is usually credited. The two errors are not the same size, which is the whole argument. Where the uncertainty is real, discovery is the cheapest insurance in the engagement, and it is also what makes scope lock possible on the build.
Related reading
What a fixed-price AI quote should contain tells you what discovery should produce, AI proof of concept vs demo explains why a convincing demo is not evidence, and the comparison hub holds the rest of these decisions. If you want the free call first, talk to us and we will tell you which of the two you actually need.
Pay for discovery when the unknowns live inside your systems, take the free call when they live inside the brief, and be suspicious of anyone who insists it is always one or the other.
Frequently asked questions
Is a free scoping call worth taking if I already know what I want?
▾
Yes, because it tests the vendor rather than the idea. In under an hour you learn whether they ask about your data before your budget, whether they have built this shape before, and whether they are willing to say no. That is useful even when your requirements are settled.
How much does paid discovery cost, and is it refundable?
▾
Eazyware's AI Discovery Sprint is $3,250 or ₹2,00,000 fixed and is credited against the build that follows. Product strategy and use-case discovery is $4,250 or ₹2,80,000, and an AI POC Sprint runs $6,250 to $10,500. Ask any vendor whether their discovery fee is credited and whether you keep the artefacts.
Can I use one vendor's discovery output to get quotes from others?
▾
Only if your contract says the artefacts are yours. Insist on it before signing. A discovery document you own, covering scope, data findings, architecture and evaluation plan, lets several vendors quote identical work, which is the only way to compare AI proposals on anything other than headline price.