AI strategy for 2027: the planning questions to answer this quarter
Easy Insight Team ·
An AI strategy for a small business is a one-page answer to six questions: which two problems AI should solve in 2027, whether to buy, embed or build, what you will spend, what staff may use and with which data, who owns it, and how you will know it worked. Answer them before 2027 budgets close.
Why plan AI in the last quarter of 2026?
Because next year's money gets committed now, and two AI decisions have dates attached.
- Licences. Microsoft 365 Copilot Business lists at £16.10 per user per month, paid yearly and excluding VAT. Microsoft's UK page offers a first-year price of £13.80 to existing customers with an eligible Business plan, on annual plans bought between 1 July and 31 December 2026 (Microsoft UK Copilot Business pricing, as of October 2026). The monthly-billed price is £19.32. If you are going to buy seats, the cheaper year has a deadline.
- Billing. From 1 December 2026, new Copilot Business licences bought through a Cloud Solution Provider (CSP) partner come with usage-based billing switched on by default. Our 16 October news roundup explains what changes and what to check with your supplier.
Neither is a reason to buy. Both are reasons to decide on purpose rather than by renewal date.
Where are UK businesses with AI right now?
Using it widely, using it thinly, and mostly without a plan.
The ONS reports that around 35% of UK businesses with 10 or more employees used at least one AI technology in June 2026, up from about 12% in September 2023. The average adopter uses about 1.6 AI technologies, and only 11% of businesses say more than half their workforce has had AI-related training (ONS, AI in UK businesses 2023 to 2026, published July 2026).
The government's AI Adoption Research from DSIT, which surveyed 3,500 businesses with five or more staff between February and May 2025 (last updated February 2026), shows why a plan matters more than a tool:
- Asked what prevents, or had prevented, them adopting AI, 71% of businesses said they had not identified a use for it and 60% cited limited AI skills.
- Very few businesses had internal guidelines or policies on using AI responsibly.
- Many of the businesses interviewed had no formal metrics linking AI to revenue or productivity.
Each of those is one of the six questions below. A strategy is not a vision statement; it is the answers to the questions that stop most small businesses getting value.
Which two problems should AI solve in 2027?
Pick two, not ten. Our view is that a small business can run one AI change properly at a time, with a second lined up, and that a longer list is a sign nobody has chosen.
A good candidate is a task that happens many times a week, involves text or documents, has a right answer you can check, and costs you measurable hours today. Our post on AI ROI for small businesses lists five patterns that tend to pass that test. If you are not sure the business is ready for any of them, score it first with the AI readiness assessment: readiness tells you whether you can start, strategy tells you where.
Buy, embed or build?
For most of the two problems, the answer is buy or embed. The DSIT research found that businesses planning AI investment over the next 12 months mostly expected to spend on off-the-shelf AI applications (65%) and on AI built into tools they already use (59%), with bespoke AI at 22%.
That matches the cost logic. An assistant inside software you already pay for has no build cost; a bespoke tool earns its place only when the off-the-shelf version cannot reach your data or follow your rules. Our AI chatbot development cost guide shows where that line usually falls, and AI development is what we build when it is crossed.
What will you spend, and what must it return?
Write down two numbers per problem: the full cost for 2027 and the saving that would make it worth it. Full cost means licences, set-up, outside help and the time your own people spend checking output.
| Line in the 2027 budget (illustrative) | Example | Where the figure comes from |
|---|---|---|
| Per-seat AI assistant | Copilot Business, £16.10 per user per month list, £13.80 first-year promo | Microsoft UK pricing, as of October 2026, excluding VAT |
| Outside help to choose and set up | Fixed-fee review or short engagement | AI consulting cost for UK SMEs |
| Bespoke build (only if needed) | Assistant grounded in your own documents | AI chatbot development cost |
| Your team's time | Checking output, training, running the pilot | Your own estimate: count it as a cost |
The return side is simpler than it looks. At list price a Copilot Business seat breaks even if each person saves under an hour a month, which our ROI post works through. The hard part is making sure every licensed person does.
Which tools may staff use, and with what data?
Assume AI is already in use. Free chatbots and AI features in everyday software reach staff whether or not you approve them, and the DSIT research found very few businesses had guidelines on using it responsibly. Banning it rarely works; deciding what is allowed does.
The policy can be short: the approved tools, the account type staff must use (a business account, not a personal one), and the data that must never go in, such as customer personal data in a tool that has not been checked. Our guide to GDPR and AI for UK SMEs sets out the six checks before any tool touches personal data.
One legal change belongs in the 2027 plan. Since 5 February 2026, under the Data (Use and Access) Act 2025 (Commencement No. 6) Regulations 2026, the UK GDPR rules on solely automated decisions have moved from a near-prohibition to permission with safeguards. If either of your two problems involves AI deciding something about a person, such as a job application or credit terms, plan in human review and take advice first.
Who owns it, and who gets trained?
Name one owner per problem: a person who already runs the process, not whoever is most enthusiastic about AI. They decide the measure, run the pilot and make the keep, fix or stop call.
Then plan training for the people who will use it. The ONS found that around 62% of businesses citing a lack of AI expertise had trained or retrained existing staff, against about 26% of those reporting no barriers. In other words, the businesses that hit the skills wall mostly answered it by training people they already had. Budget a few hours per user, on your own tasks rather than generic demos.
How will you know it worked?
Measure the task before you change it. A week of rough logs is enough: how many, how long, how many errors. Then run the AI version alongside the old way, count the checking time, and decide against the break-even figure you wrote down at the start.
If you are testing something new rather than switching on a feature, run an AI proof of concept with a pass mark agreed in advance. If you are ready to roll out, our 90-day plan for implementing AI in a small business covers the cycle from choosing the process to scaling it.
What goes on the one page?
| Question | A good answer looks like |
|---|---|
| Which two problems? | Named tasks with today's volume and hours |
| Buy, embed or build? | A tool named for each, with the reason |
| Budget and return? | Full 2027 cost and the break-even saving |
| Tools and data? | Approved list, account type, data that never goes in |
| Owner and training? | One named owner per problem; training hours per user |
| Measure? | The baseline, the test period, the keep/fix/stop date |
Review it each quarter. If a project misses its measure, stop it and move the second problem up.
If you would rather work through the six questions with someone who has done it before, that is what our AI strategy service does, and our AI consultancy page covers the rest of what we offer.
Frequently asked questions
What should a small business AI strategy include?
Six answers, on one page: the two problems AI should solve next year, how you will get the tools (buy, embed or build), the budget and what it must return, which tools staff may use with which data, who owns it and who gets trained, and how you will measure whether it worked.
How long should an AI strategy be for an SME?
One page is enough for most small businesses. If it needs more than that, it is usually trying to cover too many uses at once. Write the detail into each project plan instead, and review the page every quarter.
Why plan AI before the end of 2026?
Because budgets and licence renewals for next year get set now. Microsoft's first-year discount on Copilot Business (£13.80 rather than £16.10 per user per month, paid yearly, excluding VAT) applies only to annual plans bought by 31 December 2026, so the licence decision has a deadline this quarter.
Do small businesses need an AI policy?
Yes, even a short one. Staff are likely to use AI tools whether or not you approve any, and the government's AI Adoption Research found very few businesses had internal AI guidelines. A one-page list of approved tools and the data they may never receive covers most of the risk.
Easy Insight is a UK consultancy for AI, web, apps and data — senior specialists only, no juniors.
Next step
Wondering where AI would actually pay in your business?
An AI strategy and readiness review (from £2,500) tells you where AI pays, where it doesn't, and what to leave alone. The price is fixed in writing before we start, and the advice comes from a team that runs AI in its own products.