Most of what you read about artificial intelligence is either a sales pitch or a scare story. Underneath both, something real has changed for ordinary businesses, and it is worth being clear about what.
The useful question is not whether AI counts as a revolution. It is which parts of your marketing, your website and your working week are actually affected, and what a sensible response looks like. The honest answer is that a few things have genuinely shifted, several have not moved at all, and the businesses getting value out of it are the ones treating it as a tool rather than a strategy.
Key takeaways
- AI has changed how people search and how fast content can be produced. It has not changed what makes a customer trust you.
- Expect fewer clicks from search, and judge search on enquiries rather than visits.
- Publishing more is no longer an advantage. First-hand knowledge is.
- The reliable gains are internal: drafts, summaries, support replies, admin.
- Anything going out under your name needs a named person to check it first.
In this article
What has actually changed
Three things, really. The cost of producing passable text, images and code has fallen close to zero. Search results now often answer the question on the page instead of sending someone to a website. And these tools have quietly arrived inside software you already pay for, so your team is probably using them whether or not anyone decided to.
What has not changed is why a customer picks you. They still want to understand what you sell, believe you can do it, see a fair price and reach someone who answers. No amount of automation moves those. That is the line to keep in mind whenever someone tells you a tool will transform the business: transform which part, exactly, and how would you know?
Search sends fewer clicks
Search engines increasingly put a written answer at the top of the results, and plenty of people read it and stop there. The traffic most likely to disappear is the broad, curious kind: definitions, general how-to questions, background reading. The traffic least likely to disappear is someone comparing suppliers or looking for a price, because those people still want to see who they would be dealing with.
The practical consequence is a measurement one. If you judge your website on visits, this looks like a collapse. If you judge it on enquiries and sales, it often looks flat. Pick the measure that reflects the business before you react to the chart, which is the same argument we made in marketing reporting that helps you decide what to do next.
Example: 3,000 visits down to 1,900, and the same number of enquiries
Before rewriting anything, find out which pages lost the visits. If it is a general article that never produced an enquiry, nothing has gone wrong. If it is your main service page, that is a different problem and deserves a proper look.
There is one thing worth doing either way. The pages that matter commercially, your service, product, pricing and contact pages, should state the facts plainly: what you do, who for, where, roughly what it costs, how to get in touch. Clear, specific pages read well to a person skimming and are easier for any summarising system to quote correctly. Vague marketing language helps neither.
Content is cheap, credibility is not
Anyone can now publish forty competent articles a month. That is precisely why doing so is no longer an advantage. When the supply of readable, generic writing becomes unlimited, its value drops to about zero, and the only things left that a competitor cannot copy in an afternoon are the ones that come from actually doing the work.
Those are your own numbers, your own projects, photographs of real jobs, the objection a customer raised last week, a price you are willing to publish, a named person with an opinion. Use a tool for the first draft or the tidying up if it helps. The substance has to come out of the business, because that is the part nobody else has.
The same logic applies to service pages. Answering the questions customers genuinely ask beats producing more pages, as we set out in how to write service pages that answer customer questions.
Where AI earns its keep
Mostly inside the business, out of sight of customers. First drafts of things you would write anyway. Turning a long document into a summary someone will read. Drafting support replies from your own help notes, for a human to send. Writing meta descriptions for three hundred product pages. Transcribing a client call. Translating a first pass for a proofreader. Sorting a messy spreadsheet.
None of that is exciting, and all of it is real time back. Choose one repetitive task that eats a few hours every week, time it honestly before and after, and keep the tool only if the saving shows up. If it does not, stop using it and say so. A pilot that ends in a clear no is a perfectly good outcome, and much cheaper than a subscription nobody wants to admit was a mistake.
What it still gets wrong
It is confidently wrong. Invented prices, invented policies, statistics with no source, references to studies that do not exist. It does not know what happened in your business last week, what you actually charge or what you promised a particular customer. Left alone, it also writes in a tone that sounds like everybody else, which is a strange thing to pay for when you are trying to stand out.
Then there is what you put into it. Anything typed into a public tool may be stored and may be used to improve the service, so customer records, contracts, staff details and unreleased plans should not go in unless you have an agreement that says otherwise. Two rules cover most of the risk: a named person checks anything published under the company name, and everyone knows in writing what may and may not be pasted into these tools. Half a page is enough.
A sensible ninety days
In the first month, write the half-page policy and pick two internal tasks to test, with the time they take now written down. In the second, leave the tools alone and fix the pages that earn money, making sure they answer the obvious questions in plain words. In the third, look at enquiries rather than visits over the whole quarter, compare it with the same quarter last year, and decide what to keep.
That is an unglamorous plan, and that is rather the point. The revolution, if it is one, mostly shows up as small amounts of time returned and a slow change in how people find you. Businesses that keep answering real questions honestly, in their own voice, with a person accountable for what goes out, will be fine. The ones that switch to publishing volume because it suddenly became free will find out that everybody else did too.
Where to start
- Measurement: report enquiries and sales from search, not visits alone.
- Key pages: plain facts on what you do, who for, where and roughly what it costs.
- Content: publish what only you know, and less of everything else.
- Tools: one repetitive internal task, timed before and after.
- Rules: a named checker, and a written list of what must not be pasted in.