How to build an AI marketing strategy for a small business.

01What is an AI marketing strategy?

An AI marketing strategy is a written plan for which parts of your marketing AI systems run, which decisions stay with a person, and how you will know it is working. For a small business it fits on one page and starts from positioning and customer questions, with the choice of tools coming last.

An AI marketing strategy is a written plan that says what the business claims and to whom, which marketing jobs AI systems run, which decisions stay with a named person, and which numbers decide what continues.

That definition is narrower than the pages that rank for this search, most of which are vendor lists of tools. Tools matter, but a list of them is a shopping list. The strategy is the set of decisions that tells those tools what to say, who checks it and when to stop. It is the planning layer of AI marketing transformation, and the 90-day transformation roadmap is how it gets carried out.

The reader here is the owner or marketing lead of a business with 5 to 50 staff in New Zealand or Australia, probably already using AI for something and wanting it to add up to more.

02Why does a small business need an AI marketing strategy now?

A small business needs one now because AI use has spread faster than planning. Most New Zealand small businesses already use AI, but most are teaching themselves by trial and error, so the work it produces has no shared claim, no review rule and no measure. A one-page strategy turns scattered use into a system.

The New Zealand picture is clear. Xero's May 2026 research found 61% of Kiwi SMBs proactively using AI, with 79% of those users learning through self-guided experimentation¹. Australian figures vary with who is asked: MYOB's April 2026 survey of more than 1,000 Australian SMEs found 40% using AI², while the ABS reported around 11% of small and micro businesses using AI in 2024-25³, from a survey of about 7,000 businesses.

Written plans lag well behind use. No recent survey measures how many small businesses have a documented marketing strategy, but data on New Zealand organisations generally points the same way: Datacom's 2026 State of AI Index found 62% of New Zealand organisations have some form of AI strategy, but only 16% have a clearly defined standalone one⁴. Having a strategy is also different from it working. The Content Marketing Institute's 2026 research found 97% of B2B marketers have a content strategy, yet only 12% call it highly effective⁵.

03Step 1: start from positioning and the customer questions

Start with two inputs the AI cannot invent: a one-page positioning statement and a list of the questions customers actually ask. Positioning tells every AI system what the business claims and can prove. The questions tell it what to write about. Both come from the business's own quotes, emails, calls and reviews, gathered in a week.

The positioning page answers five things: who the business serves, what it does for them, what it claims, the proof behind each claim, and what it will not say. Take a 12-person plumbing firm in Hamilton (an illustration, like every business on this page). Its claim might be fixed-price hot water cylinder replacements with a two-hour arrival window. The proof is its quote history and reviews. That page becomes the source file every later AI job reads.

The customer questions are the second half. Collect twenty from enquiry emails, phone notes and reviews, in the customer's own words. They drive search pages, content, outbound emails and ad copy, and they matter more as buyers ask AI engines instead of scrolling results: SparkToro measured 68.01% of US Google searches ending without a click in January to April 2026, up from 60.45% in 2024⁶. A page that answers a real question has a chance of being quoted in the answer.

04Step 2: which marketing jobs should you hand to AI first?

Hand AI the jobs that repeat every week and are easy for a person to check, so a weak draft does little harm. For most small businesses that means search and AI search, content or review replies, once positioning exists. Jobs that reach a named person or spend money come later, when review habits are established.

Content is where most businesses start anyway. The National AI Centre's tracker for December 2025 to February 2026 found content generation and data analytics were the leading uses, each by 54% of Australian SME adopters⁷. That is a sensible first job as long as the positioning file comes before it. The table sorts the ten marketing jobs by when to hand them over, and the ten worked AI marketing examples show each one running.

Pick two from the first wave and stop there. Time is the usual limit: in Xero's May 2026 research, 55% of the Pragmatist group of New Zealand small businesses named time as their biggest barrier to AI¹. Two review loops fit into a working week. Six do not.

05Step 3: what review rules should the strategy set?

The review rules say what a named person must approve before AI work reaches a customer. Write them into the strategy, with an owner for each, because review is the step that slips first on a busy week. The usual list covers claims and prices, anything addressed to a person, customer stories, personal information, spending, and the decision to stop.

Many businesses are still light on this step. The National AI Centre found about half of current Australian SME AI users check AI outputs before they affect customers, and around 65% of non-adopters cited distrust of AI decision-making or a strong preference to keep human control⁷. Written rules answer that concern directly: they keep a human in the loop by design.

1. Claims and prices are approved by the owner. AI drafts options and a person picks. The Commerce Commission lists Fair Trading Act maximum fines of $200,000 for an individual and $600,000 for a business, per offence⁸, whoever drafted the words.
2. Anything addressed to a named person, such as outbound email, review replies and quote follow-ups, is read before it sends. Cold email to New Zealand addresses must also follow the Department of Internal Affairs' three steps of consent, identifying the sender and honouring unsubscribes within five working days⁹.
3. Customer stories and testimonials are confirmed with the customer before publication.
4. Personal information stays out of public AI tools unless the business has checked the risk. New Zealand's Privacy Commissioner expects a privacy impact assessment before AI tools are used and advises that, if in doubt, businesses should not use them on personal information¹⁰ (guidance from 2023, still the current position).
5. Any change that spends money, such as ad budgets or new campaigns, needs the owner's yes.
6. The decision to stop a channel stays human. Systems are built to continue.

06Step 4: how should you measure an AI marketing strategy?

Measure two numbers for the whole strategy and one signal per AI job. The two numbers are usually qualified enquiries and cost per enquiry. Each job adds a working signal, such as pages published or reviews answered within a day. Read the numbers weekly on one screen and make one funding decision a month.

Cost per enquiry is the number that lets a small business compare channels fairly, because it includes the time and tools as well as any ad spend. Track the source of every enquiry from the first week, even if that means asking on the phone and writing it in a spreadsheet. If tracking is not set up in the first month, the third month has nothing to decide with.

The job signals tell you whether the system is running before the enquiry numbers move. Search and AI search: how many of your twenty customer questions name the business when asked of ChatGPT, Claude, Perplexity and Google. Content: guides published and enquiries that started on one. Reviews: share answered within a day. If a signal stalls for two weeks, the review loop has broken, and fixing that comes before adding anything new.

07How much time and budget should the strategy plan for?

Plan for a few hours of review a week, set in fixed slots, plus whatever the chosen jobs cost in tools or help. Small businesses already work with tight marketing hours, so the strategy should name the hours and the person, and choose the budget method on purpose instead of by habit.

The hours are tight to begin with. LocaliQ's survey of 730+ small business owners and marketers, mostly in North America, found 53% spend 1 to 10 hours a week on marketing, and 31% of businesses with 10 or fewer staff have no full-time marketing employee¹¹. An AI marketing strategy moves those hours from producing to reviewing, which is why it can cover more jobs in the same week.

For money, business.govt.nz lists four ways to set a marketing budget: spend to achieve objectives, a percentage of current or forecast sales, matching competitors, or what feels affordable¹². It recommends no fixed percentage. For an AI strategy, spending to achieve objectives fits best, because each job has a measurable signal to judge it against. The guide to AI marketing costs in Australia and New Zealand gives price ranges.

08The one-page AI marketing strategy template

The template below is the whole strategy on one page. Fill in the right-hand column for your own business, using the plumbing firm example as a guide to the level of detail. If a row cannot be filled in yet, that row is the first piece of work, ahead of any tool purchase.

09A 90-day sequence for your first AI marketing strategy

Run the strategy in three 30-day blocks: write the inputs, install the first two jobs, then add a third and turn the report into a monthly decision. Ninety days is three monthly cycles, enough to see whether the system survives a busy month, which is when most self-run marketing stops.

1. Days 1 to 30: write the positioning page, collect the twenty customer questions, set up enquiry tracking and record a baseline of what search and AI engines say about the business today. Nothing new is published yet.
2. Days 31 to 60: install the two first-wave jobs. Ship weekly, review in the fixed slots, and read the one-screen report every Monday.
3. Days 61 to 90: add one job from the second or third wave, such as outbound once the list is clean. At day 90, rewrite the strategy page with what customers actually responded to, and choose the next job from the numbers.

The AI marketing transformation guide sets out the same 90 days department by department. For the jobs themselves, the guide to content marketing for small business in NZ and the guide to outbound sales automation go deeper.

10Where Flow AI fits

Flow AI is an AI marketing transformation practice: the strategy, the systems and the agents that run a marketing function, engineered end to end. The written scope is a written version of steps one and two for your business: the claim, where marketing is losing, and the first system to install.

To check readiness on your own first, the free AI-readiness benchmark takes no email, and the free agent templates are open to anyone building the first jobs themselves.

11The short version

An AI marketing strategy for a small business is one page: what you claim, the questions customers ask, two jobs handed to AI first, the rules for what a person approves, two numbers and a weekly hours budget. Tools come after the page is written.

Run it for 90 days before judging it, then rewrite it with what customers actually responded to. The guide to AI marketing for small business in NZ covers the wider picture.

To get the first version of the page written for your business, talk to Amber.