A Map for Where AI Pays Back First in Your Business

Most SMEs we meet do not have an AI problem. They have a sequencing problem. They have read the same articles, attended the same webinars, and emerged with a vague conviction that they should "do something with AI" — without a way to decide what first.

Picking the wrong first project is expensive in a way that does not show up on the invoice. It sours the team on the next ten ideas. It teaches the founder that AI is "not for us." The map below is the one we draw on a whiteboard in the first session of most SME engagements.

The eight functions to scan

Almost every SME, regardless of industry, can be broken into these functions. Walk through each one and ask: what is the most repetitive, language-heavy, judgement-light task here?

• Sales — proposals, follow-ups, qualification, lead summaries.
• Marketing — copy variants, channel-specific rewrites, briefs.
• Customer service — first replies, FAQ deflection, ticket triage.
• Operations — SOP drafting, process documentation, scheduling logic.
• Finance — categorisation, variance commentary, simple reconciliations.
• HR & recruiting — JD drafting, candidate screening notes, onboarding packs.
• Product or service delivery — first drafts of deliverables, summaries, QA checklists.
• Knowledge management — internal Q&A, search across past work, onboarding the next hire.

The impact-vs-ease lens

For each candidate task, score two things on a simple high / medium / low scale.

• Impact — if this task got 70% faster or 70% better, would the business notice this quarter?
• Ease — can a non-technical person on the team set this up in a week with off-the-shelf tools?

Start with high-impact, high-ease. Resist the high-impact, low-ease temptation — that is the project that consumes a quarter and ships nothing. Avoid the low-impact, high-ease ones too; they are demos, not leverage.

The first AI project's job is not to be impressive. Its job is to be finished, used, and visibly useful — so the team agrees to do the next one.

The most common mistake

Founders almost always reach for sales or marketing first because that is where the upside feels biggest. In our experience the highest-ease, fastest-payback project usually sits in operations or knowledge management — the boring middle of the business — because the work is repetitive, the inputs are stable, and there is no customer in the loop to surprise you.

Build a small win there first. Use it to fund the more ambitious project later, with a team that now believes the tool actually does what it says.

Closing

You do not need an AI strategy. You need a single shipped project that the team uses on Tuesday and would miss on Wednesday. Find that one. The strategy will write itself afterwards.