Producing more content is not a marketing strategy in itself. Our view is that the value of AI starts before you open a tool: by choosing which problem deserves solving and which outcome would help the business.
For a small business, the attraction is understandable. When time and resources are limited, faster copy, ideas and variations are appealing. But faster production can also amplify a confusing offer, repeat messages nobody needs or create more review work.
Start with a decision, not a tool
Before introducing AI, complete this sentence: “We want to improve this task because it prevents us from achieving this outcome.” The task might be drafting a proposal, organising sales questions or adapting a technical explanation for different readers.
“Publish every day” describes a frequency. “Answer the questions that stop people requesting a quote” describes a problem. The second formulation helps determine whether you need an article, a service page, a sales response or no new content at all.
A hypothetical case: one page before twenty posts
Imagine an office maintenance company. It receives website visits, but potential customers repeatedly ask what the service includes and how quotes are calculated. This is an illustrative example, not an Impulse ROI result.
The company could use AI to produce twenty productivity posts. Our first question would be whether the service page explains the scope, exclusions and steps to buy.
A more useful application could be to organise the questions the team already knows, create an outline and help write a clear explanation. The company would need to validate every condition and remove promises it cannot fulfil. The deliverable would be a better page rather than a content quota.
Business context still requires human judgement
A tool needs instructions, and the company needs the judgement to provide them. Who is the service for? Which objection should it address? What experience can you demonstrate? What do you not offer? Without those answers, you can easily produce competent text that could belong to any competitor.
At Impulse ROI, we suggest a short brief covering the audience, offer, service boundaries, tone and facts that can be used. It should be current and approved by someone who knows the operation.
As a working rule, avoid entering personal data, identifiable conversations or confidential information without first checking the tool’s terms and controls. An anonymised or fictional example is often enough to explore an idea.
A simple process: define, produce, validate and learn
1. Define what needs to change
Choose a focused task and a quality criterion. For a page, a potential customer should understand the scope and next step. For a proposal, the terms should be clear and consistent with what was agreed.
2. Produce a first draft
Use AI to explore structures, identify ambiguity or prepare alternatives. Ask it to flag missing information. A draft should not fill those gaps by inventing prices, clients, testimonials or results.
3. Assign a responsible reviewer
Check facts, sources, tone and promises. Someone must confirm any stated service capability. Any figure needs a traceable origin. Editorial responsibility remains even when a tool produced the first draft.
4. Learn from actual use
Publication ends production and starts evaluation. Observe whether repeated questions decrease, enquiries are a better fit or proposals need fewer corrections. Avoid attributing every change to AI: traffic, seasonality or the offer itself may also contribute.
Measure time saved, including review costs
Compare the whole process when evaluating a test. Include briefing, review, corrections and maintenance. If drafting is faster but review takes longer, the apparent saving may disappear.
You do not need to automate everything. A repetitive, well-defined task can be a good starting point; a sensitive decision or commercial promise needs more control. The aim is to use the team’s attention better, not remove their involvement.
Judgement is a competitive advantage
If several companies can draft content with similar tools, the difference may lie in what they know about their customers and the decisions they make with that knowledge. We advocate using AI to support a strategy, with explicit priorities and human review.
Before asking how much content you can produce, ask what your customer needs to understand to move forward. That answer should guide the next prompt.
A briefing template for an AI test
Before generating content, prepare a brief that someone else can review:
- Problem: the question or task to address.
- Audience: what the reader knows and needs to decide.
- Approved facts: service information and verifiable sources.
- Boundaries: claims, data or promises that must not appear.
- Delivery: format, review owner and acceptance criteria.
For example: “Organise these fictional questions into a service page. Use only the attached terms. Flag missing information and do not invent prices or results.” Judge the brief by whether it helps produce a verifiable answer, rather than by prompt length.
What using AI means for SEO
Google explains that generative AI can help with research and content structure, but producing many pages without adding value may violate its scaled content abuse policies. AI does not replace accuracy, relevance or usefulness. Source: Google’s guidance on AI-generated content.
SEO fundamentals still apply to Google’s AI features: a page must be indexable and eligible to appear with a snippet. No special AI file or exclusive markup is required, and meeting the requirements does not guarantee a mention. Source: AI features and your website.
Evaluate each piece by the decision it helps readers make and the evidence it provides. If the goal is demand generation, connect it to a specific offer and review how to choose between SEO and Google Ads.
Explore our marketing consultancy and what to consider when outsourcing marketing. To identify a useful first application for your business, request your marketing review.