Every agency proposal has one. A line item near the top, before design starts, before anyone touches a CMS. ‘Discovery phase’ or ‘discovery and scoping’ or, occasionally, the more grandiose ‘strategic discovery workshop’. It sits between £3,000 and £12,000 on most mid-market proposals, and most buyers have no framework for deciding whether it’s essential planning or an expensive way to be told what they already know.
The short answer: legitimate discovery work is one of the few line items that reliably saves money downstream. But not all discovery is legitimate, and the difference between the two is specific enough to evaluate before you sign.
What discovery work actually contains
Strip away the language and a discovery phase is a structured period (typically two to six weeks) where the agency maps the gap between what you’ve asked for and what needs to be built. That gap is always larger than either side expects.
On a well-run discovery, the agency is doing some combination of the following: auditing your existing site or system (content, analytics, technical debt, integrations), interviewing stakeholders who will use or manage the finished product, mapping user journeys against actual behaviour data, documenting the technical requirements that your brief didn’t specify (because you shouldn’t have to), and producing a specification detailed enough that any competent agency could build from it.
The deliverable set matters. At the end of discovery, you should have documents you own and can take elsewhere: a written specification, a sitemap or information architecture, wireframes or prototypes for key pages, a technical architecture document if integrations are involved, and a realistic project plan with dependencies mapped. If the agency can’t tell you what you’ll walk away with, that’s your first red flag.
Why it exists and the cost of skipping it
The uncomfortable truth behind discovery is that your brief, however thorough, is not a specification. Briefs describe outcomes. Specifications describe what needs to be built to achieve them. The translation between the two is where most project failures originate.
McKinsey’s widely cited research with the University of Oxford, covering more than 5,400 IT projects, found that large technology projects run 45 per cent over budget on average, overrun schedules by 7 per cent, and deliver 56 per cent less value than predicted. The study identified unclear requirements and shifting scope among the primary drivers of cost overruns — precisely the problems a discovery phase is designed to surface before the build starts, when changes are cheap rather than ruinous.
The same dynamic plays out at smaller scale. A £30,000 digital project that skips discovery and hits a fundamental information architecture problem in week six doesn’t cost £30,000. It costs £30,000 plus the weeks of rework, plus the strained relationship, plus the missed launch date.
We’ve seen this pattern repeatedly: agencies that skip formal discovery on complex projects don’t eliminate the work — they end up conducting it anyway, reactively and unpaid, once the build stalls. The discovery happens whether you plan for it or not. The only question is whether it happens before the budget is committed or after.
What should it cost?
Discovery pricing maps roughly to project complexity, not to agency prestige. As a proportion, expect discovery to sit between 8 and 15 per cent of the total project budget. In absolute terms, for the kinds of web projects that most small businesses are evaluating:
A brochure website or straightforward redesign with limited integrations rarely needs a standalone paid discovery phase. A good agency folds this into its proposal process. If you’re being charged £5,000 / $6,000 for discovery on a £12,000 / $15,000 build, the ratio is wrong.
A complex project — a major website with multilingual requirements, an ecommerce platform handling £1m+ in annual revenue, a membership portal with third-party integrations — typically warrants discovery fees start at, and usually significantly exceeding, £5,000 / $6,000. This is the range where the work genuinely reduces downstream risk and the deliverables are substantial enough to be independently useful.
Large-scale builds, platform migrations, or projects touching multiple internal systems can justify discovery fees that surpass £15,000 / $20,000. At this level, discovery often involves technical architects alongside designers and strategists, and the specification it produces may run to dozens of pages.
The proportionality test is the simplest one: if discovery costs more than 15 per cent of the quoted build, ask why. If it costs less than 5 per cent on a complex project, ask what’s being skipped.
When discovery is worth paying for
Not every project needs a paid discovery phase. The decision rests on complexity, not budget.
Pay for discovery when:
- the project involves significant integrations (CRM, ERP, payment gateways, third-party data)
- multiple stakeholder groups will use the finished product in different ways
- you’re migrating from one platform to another and need to understand what transfers and what doesn’t
- the brief involves functionality that doesn’t exist on your current site
- the total build budget exceeds £25,000 and the specification hasn’t been written by someone technical
Question it when:
- the project is a straightforward redesign/reskin of an existing site with no new functionality
- the agency has already built something nearly identical for you before
- the scope is narrow enough that a detailed brief and a single scoping meeting cover the same ground.
In these cases, a reputable agency should be able to fold its planning work into the project price rather than billing it separately.
How to evaluate a discovery proposal
The difference between useful discovery and expensive throat-clearing shows up in the proposal. Look for these specifics:
Named deliverables with formats. ‘A comprehensive discovery document’ is meaningless. ‘A written functional specification, a sitemap, annotated wireframes for five key templates, and a technical architecture document covering your CRM integration’ tells you what you’re buying and lets you hold the agency to it.
A stated duration with a defined team. Discovery that runs ‘two to four weeks with a strategist and a senior developer’ is evaluable. Discovery that runs ‘as long as it takes’ is an open chequebook.
Separation from the build contract. Good discovery should be structured so that you own the deliverables and could, in theory, take them to a different agency for the build. If the proposal ties discovery to a mandatory build contract, the agency’s incentive is to use discovery to lock you in rather than to give you clarity. Some agencies offer discovery as a standalone engagement for exactly this reason — and that’s worth paying a modest premium for.
Evidence of similar work. Ask to see a redacted discovery deliverable from a comparable project. The quality of the output tells you more than the proposal language ever will.
How do you spot a discovery phase that’s padding?
A few patterns that suggest discovery is being used as padding rather than planning:
The agency can’t explain what you’ll receive at the end, only what activities they’ll undertake. Activities without outputs aren’t a phase — they’re billable hours dressed up.
Discovery is quoted as a percentage of the build with no independent justification. If the discovery fee moves in lockstep with the build quote, it’s a margin line, not a scope line.
The team assigned to discovery is entirely junior. Discovery is where senior experience earns its fee — understanding which questions to ask, which assumptions to challenge, and which requirements the client hasn’t thought of yet. A discovery phase run by account managers rather than architects or senior designers is unlikely to surface the technical risks that matter.
The proposal includes discovery but no specification in the deliverable list. If the agency isn’t committing to a written spec, what is the discovery phase actually producing?
The one question that tells you whether discovery is worth it
Ask the agency a direct question: ‘If we pay for discovery and then decide not to proceed with the build, what do we walk away with? The answer tells you everything.
If the deliverables are concrete, owned by you, and useful to any agency you might work with next, the discovery phase is doing its job. If the answer is vague — ‘a better understanding of the project’, ‘alignment on objectives’ — you’re paying for a conversation, not a specification.
Discovery done well is one of the few genuinely protective investments in a digital project. It turns assumptions into documented decisions before those assumptions become expensive. But it earns that status through specifics, not through the word ‘discovery’ on an invoice line.
