A consultancy proposal can be persuasive without being easy to evaluate. It may promise transformation, display a polished methodology and mention the latest AI tools, yet leave the client unsure what will be different at the end of the assignment. For a small business, that uncertainty matters because management attention and staff time are part of the cost, even when the fee is clearly stated.
Buying useful advice starts with a decision the business needs to make or a process it needs to improve. This guide describes how to turn that need into a brief, compare relevant evidence and agree a workable handover. The examples are fictional and do not rank the providers listed in the directory. They are intended to help a client ask better questions, not to imply that any particular consultant holds a credential, offers a service or guarantees an outcome that has not been verified.
Use a standard as context, not as a substitute for checking
ISO 20700 provides guidelines for delivering management consultancy services. Its public overview establishes the subject and scope of the guidance; it does not establish that a provider is certified, endorsed or suitable for a particular assignment. [ISO 20700:2017: Guidelines for management consultancy services] The client still needs to examine the proposed work and the evidence relevant to it.
The current AI conversation adds another reason to be specific. OECD research describes uneven integration of AI into SME operations despite increasing use of readily available tools. That is a reason to ask how a proposal will affect the client’s workflow, rather than treating a tool name as evidence of capability. [OECD: Empowering SMEs in the age of AI — 2026 D4SME Survey] A consultant may use modern software responsibly, but the value of the assignment should still be explained through the problem, method, deliverables and decisions that matter to the business.
Write the problem in terms the team can check
Begin with the observed issue rather than the solution you hope to buy. A business might say that customer enquiries are not consistently followed up, that order changes are hard to trace or that training is not being applied. Describe where the issue appears and what information is available.
For a fictional service company, “we need a CRM” may be an early assumption rather than the actual problem. A review could reveal that staff do not agree who owns a new enquiry. Installing a system without resolving that responsibility may reproduce the same uncertainty in a different interface. A useful brief would explain the current enquiry process and ask the adviser to identify and test a practical improvement. That leaves room for a proportionate solution instead of committing the client to a tool before the underlying workflow has been examined.
Define what is inside and outside the assignment
A small project can expand quickly when the parties do not agree its boundaries. Describe the process, team or business unit included. Explain which activities are excluded and which dependencies belong to the client. An adviser cannot complete a meaningful record review if the necessary records are unavailable or inaccessible throughout the assignment.
For the service company, the scope might cover enquiry receipt through appointment confirmation for one type of service. It might exclude changes to pricing or the purchase of new software. The brief can still allow the consultant to identify related issues, but it should explain how additional work would be proposed and approved. This distinction protects both parties. The client receives a defined assignment, while the consultant has a fair route for raising a finding that requires a separate decision rather than silently absorbing an unlimited range of tasks.
Compare methods against the problem
Ask each provider how they would investigate the issue and why the proposed steps are appropriate. A useful answer connects methods to evidence: a record sample can show where enquiries stop progressing, while a staff discussion can explore how responsibilities are understood. Neither method should be presented as a guarantee that the first explanation will be correct.
Check how the consultant will distinguish observation, interpretation and recommendation. If interviews are included, ask how participant confidentiality will be handled. If market research is proposed, ask about the sample and its limitations. The client does not need to become a research specialist to request these explanations. It needs enough clarity to judge whether the work will answer the question in the brief. A detailed list of activities can still be unsuitable if none of them produces evidence relevant to the actual decision.
Ask for relevant experience without demanding private files
Examples of previous work can help establish relevance, but confidentiality matters. A provider should not need to disclose another client’s sensitive records to demonstrate competence. Ask for a redacted example, a discussion of the method or an authorized reference where appropriate.
Consider what the example actually shows. A large event photograph demonstrates that an event occurred, not that a process improved afterwards. A certificate should be checked for its issuer, scope and current relevance rather than treated as a general promise of effectiveness. The client can also ask who will perform the work, which responsibilities belong to senior and junior team members and what happens if the named person becomes unavailable. Those questions connect the proposal with the people who will deliver it. They are more useful than assuming that an organization’s general reputation automatically applies to every assignment.
Make deliverables usable after the presentation
A final presentation can explain findings, but it may not be the main working output. For the service company, useful deliverables could include a reviewed enquiry map, an agreed ownership rule, a short follow-up checklist and an example record that staff can maintain. These are suggested outputs for the hypothetical case, not a universal consultancy package.
Specify the format, intended users and handover expectations. If a tool is involved, clarify account ownership, permissions and the ability to export business information. If a template is provided, agree whether the client can adapt it. The output should be understandable without the consultant standing beside it. A beautiful document that cannot be maintained by the team may create continuing dependence rather than capability. Ask the provider to show how the deliverables will be tested with the people expected to use them.
Agree review points before the final deadline
Waiting until the final report to discover a disagreement about the problem is expensive for both sides. Set review points where the client can confirm the emerging understanding and raise concerns about scope or evidence. Keep decisions and changes in a shared record.
An early checkpoint might review the current-process map. A later one could examine the proposed improvement and the plan for a limited test. The review should not require the client to approve an unsupported conclusion merely to maintain the schedule. Nor should it give the client a reason to keep redefining the assignment without recognizing the effect on delivery. A clear change process lets the parties discuss new findings honestly. It also creates a record of why the final assignment differs from the original brief if a legitimate change becomes necessary.
Treat AI use as an information and accountability question
If the consultant uses AI, ask which tasks it supports and what information is submitted to the service. A proposal should not assume permission to upload customer records, employee information or confidential documents. Agree the approved uses and the review responsibility for outputs.
The client can distinguish drafting support from analysis that affects a recommendation. An automatically summarized interview still needs checking against the source and the consent arrangements. A generated market figure needs a verifiable reference. The consultant remains responsible for the professional work delivered, including identifying uncertainty. These questions do not reject AI tools. They make the use of those tools consistent with the assignment’s information requirements. A provider who can explain the boundary between automated assistance and accountable judgement gives the client a clearer basis for evaluating the method.
Measure the change the assignment can reasonably influence
Agree a small number of observations connected to the work. For an enquiry-follow-up project, that could include whether each enquiry has an owner, whether promised follow-up is recorded and how unresolved cases are reviewed. A short assignment may not be able to establish an effect on overall business growth.
Keep the evaluation proportional. Record the starting condition, the test period and other changes that might affect the result. The ILO’s trainer guidance is a useful reminder that needs analysis, delivery and follow-up belong to a connected learning process; an assignment involving training should not stop at attendance. [ILO: Start and Improve Your Business generic trainer’s guide] For consultancy more broadly, the same management question is helpful: what can the team demonstrate after the work that it could not demonstrate before? Answering that is more meaningful than counting report pages or celebrating the number of recommendations produced.
Finish with ownership and the next review
Before closing the assignment, identify who will maintain each new process or record. Review open questions and distinguish completed outputs from actions the client still needs to take. Ensure that access, files and account responsibilities are handed over through the agreed route.
A good final conversation may include what did not work, which assumptions remain uncertain and which next step is justified by the evidence. That is more useful than a claim that every business challenge has been solved. In the AI era, polished material is easier to produce, which makes a clear brief and a usable handover even more important. Buy an assignment you can explain through the work performed, the evidence reviewed and the capability left with the team. The presentation should help communicate that result, not stand in for it.
