AI Business Case for UK SMEs

In short

A board-ready AI business case in plain language: what we’d build, what it costs, what it returns, what could go wrong, who signs off what. Wingenious produces this as the Feasibility Study deliverable: £3,950, 2–3 weeks.

What’s in the document

1. Executive summary: one page, board-readable
2. Problem statement: what business outcome are we improving?
3. Proposed AI capability: what we’d build, in plain language
4. Investment + ROI: costs, returns, payback (see AI ROI calculation)
5. Built-vs-buy: defensible recommendation (see built-vs-buy analysis)
6. Vendor shortlist: top 3 with pros/cons (see vendor shortlisting)
7. Implementation plan: sequence, timeline, owners
8. Risk register: what could go wrong, with mitigations
9. Governance: who approves what, audit trail design
10. Recommendation: proceed / pause / kill, with reasoning

Why this exists

Most SME AI investment decisions get made on vendor sales materials. The Feasibility Study replaces that with a defensible internal document. Your board can actually scrutinise the proposal rather than the pitch deck.

What gets missed when an SME skips this stage

The pattern repeats often enough to be worth naming. An SME identifies an AI opportunity, gets enthusiastic about it, watches two vendor demos, picks the more polished one, signs a 24-month contract and a procurement order, and starts the build. Six months in, the team realises the use case needed twice as much integration work as anticipated, the data was nowhere near as clean as the demo implied, the vendor is charging substantially more than the headline number once volume kicked in, and the projected ROI was based on assumptions that nobody actually challenged.

The business case stage prevents this. It is not bureaucracy. It is the cheapest way to surface the failure modes before they cost real money. Two or three weeks of structured analysis at £3,950 reliably saves the SME from spending £30,000 to £80,000 on the wrong build. The Feasibility Study pays for itself many times over even when its conclusion is “proceed as planned”, because the proceeding is now defensible rather than impulsive.

The other thing that gets missed is the sequencing question. AI investments rarely sit alone. The use case picked for the first build affects what is realistic for the second and third. Building a customer-facing AI before the governance posture is in place is a different risk shape from building it after. The business case surfaces sequencing trade-offs that the leadership team would otherwise not consider until they are already mid-build.

What goes into each section, in detail

The document is structured for a leadership team to read in one sitting and to scrutinise in detail if they want to.

Executive summary

One page. The proposed investment, the expected outcome, the headline cost, the headline return, the recommended decision, the risks the board should know about. Designed to be read by a non-executive director on a train. Decision-ready without reading further.

Problem statement

The business outcome the investment is improving, named in commercial terms. Not “we want to do AI” but “we want to reduce our cost-per-acquired-customer by 15 percent” or “we want to recover 12 hours per week per support agent”. The business case is anchored to a measurable outcome, not to an aspiration.

Proposed AI capability

What gets built or bought, in plain language. The shape of the workflow. The tools that touch it. The data that flows through it. The human touchpoints. The boundaries of what is in scope and what is explicitly not in scope.

Investment and ROI

Three-year total cost of ownership: build cost, ongoing licence cost, ongoing operational cost, internal time. Against it: expected return, calculated using your specific volume and salary figures, with sensitivity ranges showing the answer at conservative, base and optimistic assumptions. Payback period in months. NPV at a sensible discount rate. The numbers are calibrated to UK SME reality, not enterprise benchmarks.

Built-versus-buy recommendation

The defensible answer to whether this use case is a vendor product, a custom build, or a hybrid. Reasoning attached. See built-vs-buy analysis for the framework.

Vendor shortlist

If the answer is buy or hybrid, three named vendors with pros, cons, pricing at your volume, contract risk, data residency and reference customers. See vendor shortlisting for the framework.

Implementation plan

What happens in what order. Who owns each step. Where Wingenious sits (delivery, oversight, or out of the way entirely). What dependencies need resolving before the build starts. What dependencies can run in parallel.

Risk register

The realistic things that could go wrong, sorted by likelihood and impact. Each one with a mitigation. The risks that get included are not the ones consultants normally cite (“technology may not work as expected”) but the specific ones for this build (“supplier data integration may need bespoke parsing because the supplier still ships invoices as scanned PDFs”).

Governance

Who approves what. Audit trail design. Human-in-the-loop specification per workflow output. Compliance posture against UK GDPR, ICO guidance, the Data (Use and Access) Act 2025, and the EU AI Act where it applies. See AI governance models for the framework.

Recommendation

The defensible answer to the proceed-pause-kill question, with reasoning. The Feasibility Study has produced more than a few “pause and address X first” recommendations and a small number of “kill” recommendations. The fee is the same regardless of which answer comes out; the independence is the whole point.

How the work is done

The Feasibility Study runs over two to three weeks. Roughly the shape:

• Week one. Kick-off with leadership. Use case definition refresh. Data and integration scoping. Stakeholder interviews. Volume and cost baseline gathered. Initial vendor scan begun.
• Week two. Built-versus-buy analysis. Vendor deep-dive on the shortlist candidates. ROI modelling with sensitivity. Risk register drafted. Governance posture mapped.
• Week three (where needed). Draft document reviewed with leadership. Refinements. Final document produced and presented in a 60 to 90 minute readout. Q&A. Sign-off on next steps.

The output is delivered in editable form (Google Doc, Word, or whatever shape suits the SME’s board pack). Nothing is locked in Wingenious tools.

What happens when the board approves

The same document feeds straight into a delivery engagement. Three typical next moves.

1. AI Implementation Sprint from £8,000, four weeks. Production workflow delivered, with 30-day stabilisation.
2. Quick Win from £1,500, one to three days. Smaller scopes that do not need the full Sprint shape.
3. Vendor procurement track. Where the answer is buy, the SME proceeds directly with the recommended vendor using the Feasibility Study as the procurement reference document.

About two-thirds of Feasibility clients proceed to delivery within 90 days. The remaining third either pause and address prerequisites, kill the investment, or run the delivery themselves using the document.

What happens when the board does not approve

The Feasibility Study has paid for itself. The SME has avoided a wrong investment that would have cost many times more to discover by building. The document is reusable; many SMEs return to it 12 to 18 months later when conditions have changed and the original use case looks more viable.

How to commission this when there is no use case yet

A common SME shape: the leadership team knows they want to do something with AI but does not yet have a specific use case to commission a business case against. The Feasibility Study is the wrong starting point in that situation. Two better entry points.

The first is the AI Readiness Audit at £2,450, five days. The audit produces the ranked shortlist of candidate use cases; the Feasibility Study then deep-dives the top one. Fifty percent of the audit fee is credited against any subsequent engagement booked inside 60 days, so the combined cost is competitive against starting with the Feasibility Study cold.

The second is an AI strategy workshop, half day or full day, price on application. The workshop produces a draft 90-day plan with named use cases; the Feasibility Study then deep-dives the priority one.

Either entry point produces the specific use case definition that makes a business case worth commissioning.

What the readout meeting actually covers

The 60 to 90 minute readout at the end of the Feasibility Study is structured for a leadership team that may not have been involved in the detailed work.

The first 15 minutes are the executive summary and recommendation. The leadership team gets the conclusion before the detail. If the recommendation is “proceed”, the rest of the meeting is about how. If the recommendation is “pause” or “kill”, the rest of the meeting is about why and what to do instead.

The next 30 minutes work through the substantive sections: the ROI projection with its assumptions, the built-versus-buy reasoning, the vendor shortlist, the risk register and the governance posture. Each section invites challenge from the leadership team; the document is built to be argued with, not absorbed passively.

The final 15 to 30 minutes are decision and next steps. Who is going to do what by when. What additional information would change the picture. What dependencies need resolving before the build can start.

Notes go out within 48 hours capturing decisions, action owners and any open questions.

Why the Feasibility Study price holds at £3,950

The fixed price is deliberate. SMEs benefit from knowing the cost in advance rather than discovering it as the engagement unfolds. The £3,950 covers the two to three weeks of structured work plus the written deliverable plus the readout meeting plus the editable model handover.

Some engagements take less time than the budget allows because the use case is clearer; the SME does not pay less. Some take more because the use case is messier; the SME does not pay more. The fixed price is the discipline that forces Wingenious to scope honestly rather than running open-ended billable hours.

The fee is independent of recommendation. A study that recommends proceeding costs £3,950. A study that recommends pausing costs £3,950. A study that recommends killing the proposed build costs £3,950. The independence is structural: Wingenious has no incentive to push toward proceeding because the fee is the same regardless.

Related capabilities

AI ROI calculation · Vendor shortlisting · Built-vs-buy analysis · AI use case identification

Related

Sectors where the business case matters most: manufacturing, law firms, accountants.