Straight answer: AI changes the speed and cost of marketing execution, not the quality of your thinking, so it makes a strong strategy stronger and a weak one collapse faster than it used to. It rewires targeting, content production and customer service, but the positioning, the offer and the voice still have to come from a human who understands the customer. If you skip that part, you get a business that publishes more and converts less.
What gets rewired (and what doesn’t)
I’ve spent the last two years rebuilding my own consultancy in public, testing every AI tool that lands in my inbox on my own business before I’ll recommend it to a client. Here’s the honest split. AI has changed three things hard: how fast content gets produced, how precisely ads get targeted, and how quickly customer questions get answered. It has changed almost nothing about the fourth thing, which is knowing what to say and to whom.
That distinction matters because most of the advice online conflates the two. People write “AI has transformed marketing” as if the transformation is strategic. It isn’t. It’s operational. A tool that writes 40 Instagram captions in ten minutes hasn’t told you anything about whether your audience wants Instagram captions in the first place.
The experiment that changed how I think about this
Last year I ran a six-week test on my own LinkedIn account. Half the posts I wrote myself, start to finish, the way I’ve done for over a decade. The other half I drafted with ChatGPT from a two-line brief, then lightly edited. Same topics, roughly matched for length, posted at the same times.
The AI-assisted posts took me about a fifth of the time to produce. But they got, on average, 30% less engagement per post than the ones I wrote myself, and noticeably fewer comments with actual questions in them, which for me is the metric that turns into consulting enquiries. The AI drafts read fine. They just sounded like everyone else’s LinkedIn posts, because they were trained on everyone else’s LinkedIn posts.
What I took from that isn’t “don’t use AI for content.” I still use it constantly, for first drafts, for restructuring a messy paragraph, for turning a long blog post into a short email. What I took from it is that the parts of my content that work, the specific stories, the blunt opinions, the numbers from my own business, are precisely the parts AI can’t generate for me, because it wasn’t there. If your content strategy leans on AI for the bits that make you distinctive, you’ll produce more content and get worse results, and you’ll spend months confused about why.
Where AI earns its place: targeting and spend
The area where AI has quietly done the most damage and the most good is media buying. Meta’s Advantage+ and Google’s Performance Max campaigns now make thousands of micro-decisions about who sees your ad and when, decisions that a human media buyer used to make by hand across dozens of ad sets. For small businesses this is a genuine gift: a client of mine, a B2B training company with a monthly ad budget of around £2,000, saw cost per lead drop from roughly £38 to £24 within eight weeks of moving from manually built audiences to Advantage+ campaigns, simply because the algorithm could test combinations a human never had time to try.
But that same automation removes a layer of control that used to force marketers to understand their audience deeply before they could target it. You used to have to know your customer well enough to build the audience by hand. Now you can skip that step entirely and let the machine find people who click, which isn’t the same as finding people who buy and stay. This is the same logic behind why brands with distinct positioning still win even inside automated systems. Tesco’s marketing strategy works because decades of Clubcard data let them personalise offers with real precision, not because the algorithm is clever in the abstract. The data was good before the AI was.
The uncomfortable bit most people skip past
Here’s the part that doesn’t make it into most posts on this topic: AI doesn’t level the playing field, it widens the gap between businesses that already knew their customer and businesses that didn’t. If your strategy before AI was “post regularly and hope,” AI just lets you post regularly and hope at ten times the volume, which burns through your audience’s attention faster and teaches the algorithm that your content isn’t worth showing. I’ve watched this happen to at least three small business owners I’ve advised this year: they adopted AI content tools expecting a lift, and instead saw engagement flatten within two months because they were producing generic content faster, not better content at all.
The businesses where AI is a genuine multiplier are the ones that already had a clear point of view. Dove’s marketing strategy around Real Beauty didn’t come from a machine and never could have, because it required a human insight about how women felt seeing themselves in advertising. Any AI tool layered on top of that insight, for producing variations, testing headlines, localising copy for different markets, makes that strategy scale further. Layer the same tools on top of no insight at all and you just get faster mediocrity.
A four-step way to build it into your strategy
When I sit down with a client to work out where AI belongs in their marketing, I don’t start with tools. I start with an audit that takes about half a day. Here’s the version you can run yourself:
- Step one: map your current bottleneck. Is it that you don’t produce enough content, that your targeting is too broad, that customer questions go unanswered for days, or that you don’t know your customer well enough to say anything specific? Write down the actual bottleneck, not the one that sounds impressive.
- Step two: match the bottleneck to the AI category that solves it. Content volume problems get solved by generative writing and image tools. Targeting problems get solved by automated ad platforms. Response time problems get solved by AI chat and email triage. A strategy problem doesn’t get solved by AI at all, it gets solved by talking to customers.
- Step three: run one small test before you commit. Pick one channel, one AI tool, four weeks. Track a number that matters (leads, replies, conversion), not a vanity number like posts published. My own LinkedIn test is a version of this.
- Step four: keep the human layer on anything customer-facing that carries your voice. Drafts, yes. Final copy that goes out under your brand’s name, no, not without a real edit from someone who knows what the brand sounds like.
This is roughly the same audit process I use in my AI implementation coaching work, and the businesses that get real results are almost always the ones that do step one honestly instead of jumping straight to buying a tool.
Personalisation is where the change is biggest, and most invisible
If you want to see AI’s effect on strategy at its most developed, look at recommendation engines. Spotify’s marketing strategy is built on the idea that a personalised playlist is itself a marketing asset, Discover Weekly and Wrapped both function as retention tools disguised as product features, and neither could exist without machine learning doing constant, invisible work behind them. The strategic decision, that personalisation should be the product, came first. The AI is the mechanism, not the idea.
Smaller businesses can copy the logic without the budget. A subscription meal kit like Gousto’s marketing strategy uses recipe recommendation data to reduce churn by suggesting meals customers are statistically likely to order again, a much smaller-scale version of the same principle: use data to make the customer feel understood, then let AI handle the scale. An email tool like Klaviyo can do a rough version of this for a shop with 2,000 customers on a list; you don’t need Spotify’s engineering team to personalise a follow-up email sequence based on what someone browsed.
Where I’ve seen it go wrong with real clients
One client, a professional services firm with about 15 staff, brought in an AI writing tool to handle their entire blog output, aiming for two posts a week instead of two a month. Traffic went up within three months, roughly 60% more organic sessions. Enquiries didn’t move at all. When we dug into it, the new posts were broad and generic (“5 Tips For Better Cash Flow”) because that’s what the AI defaulted to without specific input, while the old, slower posts had been built around real client questions and included numbers from actual cases. The traffic increase was hollow, people landing, reading a generic post, leaving. We rebuilt the process so the AI still did the drafting, but every post started from a real client question and a real number, fed in by a human first. Enquiries recovered within two months.
That’s the pattern I’d want anyone reading this to take away over any list of tools: AI strategy work is really an editing discipline. What you feed it and what you refuse to let it decide matters more than which platform you use.
Where to spend the budget
If you’re deciding where AI money goes first, the order I’d recommend, based on where I’ve seen the fastest return across small and mid-sized clients, is: customer service and lead response first (an AI chat or email triage tool can cut response time from hours to minutes, and speed to lead is one of the strongest predictors of conversion in B2B), then ad targeting automation second, then content drafting third. Content gets the most attention online because it’s the most visible, but it’s usually not where the money is.
Costs vary a lot depending on how much of this you want to build yourself versus bring in help for. If you’re weighing up doing this alone against getting someone in to set the systems up, it’s worth reading through what an AI consultant costs before you commit a chunk of your budget to a tool subscription you might not need yet.
The one question worth asking before any of this
Before you touch a single AI tool, ask: if I had unlimited time and no software at all, would I know what to say to my customer and why they should care? If the honest answer is no, fix that first. AI will make the wrong message travel faster, and that’s worse than a slow, wrong message, because it burns the audience’s trust before you’ve even worked out what you should have said. Get the message right by hand, then let AI carry it further than you could carry it alone.
Frequently asked questions
Does AI replace the need for a marketing strategy?
No, it does the opposite: it makes a clear strategy scale faster and makes a vague one fail faster, because AI amplifies whatever input you give it, good or bad.
What part of marketing has AI changed the most in 2026?
Ad targeting and media buying, through automated platforms like Meta Advantage+ and Google Performance Max, followed closely by first-draft content production and customer service response times.
Should a small business use AI for its content strategy?
Yes for drafting and speed, but keep a human editing pass on anything that carries your brand voice; generic AI-only content tends to produce more traffic without more enquiries, based on what I’ve seen across several small business clients.
How do I know if AI is helping my marketing or just adding noise?
Track a business number, not a vanity number: leads, replies, or sales, over a four-week test on one channel, before and after you introduce the tool, the same way I tested AI-written LinkedIn posts against my own writing.
