Skip to content
Interface language: English
Imperium / Signal

Paid discovery before a project: a separate service needs a useful outcome

When can understanding a problem be a paid service in its own right? Define a usable outcome, the client's contribution and the boundary of any later project before work starts.

Article language: English

An illustration of a gold-coloured lantern lighting stone steps and cracks in a wall.
In this article

Description: When can understanding a problem be a paid service in its own right? Define a usable outcome, the client's contribution and the boundary of any later project before work starts.

“To price a solution, we first need to see what is happening.” That can be a reasonable way to begin working together. It can also mean asking the client to pay for work that helps only the person writing the next proposal. The difference lies in the outcome: what will the client receive, and how could they use it if the engagement ends with discovery?

An introductory conversation helps establish whether there is a fit. A self-contained discovery assignment involves defined work on the client's material and gives the client a basis for a decision. The boundary is not determined by the length of a meeting or the appeal of calling it a “workshop”. Set it out before anyone starts preparing data.

What is the client really trying to find out?

Imagine a distributor whose team prepares answers to product enquiries. They use several AI tools and company documents. The owner wants a shared solution because the current work feels chaotic. This is a hypothetical example. We do not yet know whether the main difficulty lies in the sources, the choice of tool, the flow of material or the approval of answers.

In an initial conversation, a provider can identify the type of task, the tools in use and the decision the client wants to make. They do not have to assess every document or reconstruct the full journey of an enquiry at that stage. They can say honestly that a concrete proposal needs more information and offer a separate discovery assignment. That is different from offering an implementation while concealing the work needed to understand the problem.

The GOV.UK Service Manual's account of discovery separates understanding the problem from building a service. The phase ends with a decision about whether to proceed, and the findings may justify stopping. This is a model for UK public services, not a rule on charging in a Polish business. It does show that discovery can have an ending of its own.

In our example, the discovery question could be: what needs to be put in order before the team compares ways of working together with AI? It does not assume a new system will be bought. If the assignment starts as “prove that we need a new assistant”, the provider has been given a predetermined result rather than a problem to investigate. The client should know whether they are buying an analysis or further development of an idea already chosen.

A standalone outcome must be usable

Test the proposed deliverable with a simple scenario. The person responsible receives the material, ends the engagement with the provider and returns to their team. Can they explain what was learned, what supports it and what could happen next? If every sentence requires the author's presence, the result is not yet clear enough.

For the distributor, a useful deliverable might describe the observed path of selected enquiries, confirmed difficulties and the limits of the analysis. It should distinguish a change possible within the current setup from a question needing another trial. It need not contain a complete design for a future tool. It should, however, enable the agreed decision without making the client guess what a recommendation means.

The Australian Digital Transformation Agency describes documenting discovery, including who was consulted, the problems identified, what was out of scope and the reasons for decisions. We refer to its description of a public-sector phase, not to a template for a commercial report. For your own assignment, choose the information needed for the decision at hand.

Suppose the analysis finds that no source of current product data has been agreed. The client can first appoint an owner for that information and organise a selected set of materials. That is useful in its own right, even if they buy no new tool. The provider should show the basis for the conclusion and the limits of the material reviewed, rather than treating that answer as a failed implementation sale.

By contrast, “the data should be organised” alone offers little practical value. Which data, in which activity, and what in the sample revealed the problem? The discovery proposal should specify a level of detail that lets the client take the next step. The number of pages says little about that.

What are you commissioning, and what must you provide?

The client has tasks in a separate discovery assignment too. They prepare materials, identify people to involve and explain cases that cannot be understood from a file alone. If the proposal describes only the consultant's work, the client may learn what they must contribute only after work has begun. A delay can then result from inputs that were never agreed, rather than a lack of goodwill.

Here the parties need to establish who may share enquiry examples and documents within the permitted scope. There is no reason to demand the entire correspondence history by default. An agreed sample may be a suitable starting point if it answers the research question. Access and the people involved should follow from the work the provider will actually do.

The article on materials for an AI audit (in Polish) discusses preparing inputs for such work. The business decision comes first: can the company share them, and does it have time to take part in the analysis? Do not commission a phase that depends on someone unavailable for its entire duration without agreeing a different way to proceed.

The provider should also explain what happens if the material is incomplete. Will they narrow the outcome, ask for more information or propose a change of scope? “We will see as we go” can honestly describe uncertainty, but it does not replace an agreement on how to handle it. The client needs to recognise when the original assignment can no longer be carried out as planned.

Separate preparing existing material from creating new material for the analysis. If the company has no register of tools in use, compiling one takes work. Do not list it as an available input merely to make the discovery proposal look simpler. Agree who will create it and what information is needed, rather than silently shifting that obligation.

Aeneia conceptual photography

What changes once the result is delivered?

Look more closely at the fictional distributor's assignment. The company selects enquiries about one product group. It wants to know why preparing answers requires going back to several people. One type of enquiry concerns availability; another asks for an explanation of a specification. This distinction matters: assessing help with reading documentation will not, by itself, answer a question about current stock.

Suppose the result identifies two different gaps. For availability, the team has no agreed route to up-to-date information. For specifications, it has the right document but cannot identify the correct version. The manager can assign two different tasks on that basis. They do not have to choose one technology immediately to fix both problems. That is how discovery starts to work for the client after the conversation ends.

The recipient should know that the finding concerns the product group and cases examined. It cannot be presented as proof of conditions across the whole company. If the client wants broader discovery, they now have a specific reason and question. They may instead decide to address only the identified point for now, because that is where they can make a change.

Before accepting the deliverable, check one of the cases with someone who knows it. Does the finding refer to the same situation? Does an omitted exception change the conclusion or merely limit where it applies? An agreed way to raise such feedback allows an analytical mistake to be corrected. Acceptance should not become an endless exercise in adding processes that were outside the assignment.

The person approving the purchase need not attend every conversation, but should know the decision that ends the phase. In our example, that is the choice of the first area to put in order and the condition for a possible tool trial. If the authorised person appears only after the presentation, the provider may discover that they answered a different question from the one the client needs resolved. Check alignment before work begins.

Four different endings for the same discovery

One outcome is that the present way of working needs organising before a technology is selected. A second is that a limited trial of a chosen solution is worthwhile. A third is that a simpler change will suffice for the identified use than the project first considered. A fourth is that the available material still does not support a decision, but the missing information is now known. These are different outcomes and must be assessed against the purpose of the assignment.

The last case calls for particular honesty. A lack of resolution does not become valuable merely because it is called the discovery of uncertainty. Check what the provider was supposed to establish, which inputs were agreed and what work was actually done. If available material was not analysed, do not present that as a limitation imposed by the outside world.

When uncertainty is apparent at the start, define in advance what will count as the deliverable: for example, the difference between two versions of an enquiry workflow and the information needed to compare them. The client then understands what work they are paying for. They do not receive a new definition of success after the fact, fitted to whatever happened to be achieved.

A comparison of building, buying or keeping the work manual (in Polish) may be a later decision. Discovery need not itself complete the entire choice of provider and technology. It is enough to state honestly which options were examined, which need a trial and what remains outside the present scope.

Can the client leave with the result?

Test the phase's independence organisationally too. Will the client retain the agreed material and understand how they may use it? Do they know where the discovery result ends and the detailed design of the next phase begins? Agree what will be handed over and the terms of use before work starts. Neither should be inferred from the word “report”.

If a later commission will compare several providers, check whether the discovery result can be tied to a common question. The brief for choosing a B2B service provider (in Polish) can help prepare that comparison. “Implement our solution” alone is not a basis for assessing other possibilities. It may be a provider's proposal, but it should be named as such.

There is no need to pretend to be indifferent to further work. A company selling discovery may also offer implementation. It should nevertheless separate its findings from a commercial proposal for the next step. The client must be able to see what follows from the problem identified and what reflects the choice of a particular offer.

When an expert takes part in a sales conversation (in Polish), agree their role beforehand. Are they helping establish fit, explaining the scope of discovery or already analysing material? An expert's knowledge does not turn every meeting into a paid assignment. Payment and the subject of the work should be agreed before it starts, not inferred from the fact that the conversation was useful.

The price of one phase should not conceal the next

Compare discovery proposals by the work, inputs and outcome. One may cover a limited sample; another may examine several separate workflows. Their titles alone do not make them equivalent. Choose a scope that fits the decision the company needs to make now, rather than buying the largest report just in case.

If the provider proposes crediting the discovery cost against later work, that condition should be clear before the order. It does not remove the need for the first phase to be useful on its own. The client should still be able to stop after discovery without feeling they bought only a more expensive introduction to an inevitable project.

Also identify the unit in which the service is sold (in Polish). Access to an expert for a set time and preparation of an agreed analysis are different offers. Do not promise a complete assessment of an entire company in a short meeting if the actual service is a consultation on one problem.

At Aeneia Audit, the conversation can begin with a specific AI use, the tools involved and the decision for which discovery is needed. The scope should fit that area. The outcome can help establish priorities or prepare a further trial, without assuming in advance that a new solution must be implemented.

For our distributor, the first step after discovery may be to designate a source for stock availability and an owner for product documentation. If those tasks are enough for the present phase, the client already has a useful result. They can return to a new tool trial with a named problem rather than a general sense of chaos.

Aeneia / Aeneia editorial

Editorial article. Source links are included in the text.

About Aeneia

Brands we have worked with

  • L’Oréal
  • Garnier
  • Netflix
  • Carrefour
  • Warner Bros. Discovery
  • Biedronka
  • HBO Max
  • Wolt
  • Żabka
  • COSRX
  • Embassy of China in Poland
  • Mixa
Aeneia / Initium

From knowledge to a decision

Tell us what you want to put in order. We will agree on where to start.

Contact Aeneia

We partner with and use solutions from:

  • Google
  • OpenAI
  • Microsoft
  • Stripe
  • Anthropic
  • Linear