The honest framing
AI makes producing marketing material fast and cheap.
That is both the opportunity and the risk. Cheap production tempts more output, and more output is usually worse marketing — more emails, more posts, more content nobody asked for.
The businesses that benefit use it to make fewer things better, not more things faster.
Here is where it genuinely helps.
1. Drafting
The obvious one, and a real saving.
What it does well:
- Subject lines and headlines, in volume, to test against each other
- First drafts when the blank page is the obstacle
- Adapting one message for different audiences
- Shortening something that is too long
- Social posts from a longer piece
What still needs you:
- Whether the offer makes sense
- Whether the claims are true
- Whether it sounds like your company
- Whether anyone actually wants this
Read every word before it goes out. Generated copy is fluent, and fluent is not the same as accurate — particularly around prices, dates and availability.
2. Segmentation
The most underrated use, and the one with the biggest advantage if your marketing runs on your business system.
A standalone email tool knows what you uploaded and how people responded to previous sends.
Your ERP knows what people bought.
That means a segment can be defined by real facts:
- Customers who bought a specific product and have not reordered
- Customers whose order value dropped compared to last year
- Buyers of one product who never bought its natural companion
- Customers with no order in six months
Narrow beats broad, consistently. Two hundred people who bought a specific product will outperform two thousand vaguely relevant contacts on every measure, including unsubscribes.
FIGURE 1: WHERE AI HELPS IN MARKETING, MOST USEFUL FIRST
Segmentation
- Building audiences from what people actually bought. The biggest advantage.
Drafting
- Subject lines, first drafts, variations. A real time saving.
Analysis
- Which campaigns produced orders, not just opens.
Personalisation
- Right message to right person — as long as it stays relevant, not creepy.
3. Analysis
What it does. Answers questions about campaign performance without building a report.
The useful version: connecting campaigns to orders, not just opens and clicks. If your marketing and your sales data share a database, you can see the revenue that followed a campaign.
Report that. Open rates are soft and inflated by privacy features. Revenue is real, and most standalone email tools cannot show it to you at all.
4. Personalisation
What it does. Different content for different people, based on their history.
Where it works: recommending a genuine complement, a reorder reminder for something consumable, content relevant to what they bought.
Where it fails: when it becomes obvious. A customer who mentions something once and then sees it referenced everywhere finds that unsettling rather than helpful.
A useful test: would you say this to them on the phone? “I noticed you bought a printer, you might need cartridges” is fine. “I noticed you have been looking at this three times this week” is not.
Where it damages you
Four ways, and all are avoidable.
Frequency
The most common. Producing an email costs nothing now, so more get sent. More sending produces unsubscribes and spam complaints, which reduce deliverability for everyone on your list — including the people who wanted to hear from you.
Test before any send: if this email did not arrive, would anyone notice? If not, do not send it.
Generic content
AI produces competent, conventional output. If everyone in your industry uses it the same way, everyone sounds the same.
Your differentiation is what it cannot generate — your specific experience, your actual customer stories, your point of view.
Invented claims
It will produce specifications, certifications and guarantees that sound right. In marketing material this is not just embarrassing — it can be a legal problem.
Keep a list of things it must never claim and check every piece against it.
Consent
Automation makes it easy to add people to lists. A support enquiry is not a newsletter subscription. A business card is not consent.
Only email people who agreed. Include an unsubscribe link and honour it immediately. This is a legal requirement in most countries and a deliverability requirement everywhere.
FIGURE 2: MARKETING THAT IMPROVES AND MARKETING THAT DECAYS
Improving
- Fewer, better-targeted sends
- Every draft read before it goes
- Segments built from purchase history
- Measured against orders
Decaying
- More sends because they are cheap to make
- Generated copy sent unread
- Everyone on one broad list
- Measured on open rate alone
Deliverability comes first
Worth stating plainly: there is no point improving your copy if the emails are not being delivered.
Authenticate your domain. SPF, DKIM and DMARC. Without these, a growing share of your mail never arrives.
Never buy a list. The fastest way to destroy a sending reputation, and illegal in many places.
Clean your bounces. Repeatedly mailing dead addresses tells providers you are careless.
Warm up gradually. If you have never sent bulk email from your domain, do not start with ten thousand messages.
What to measure
Revenue attributed to campaigns. The number that matters, and the one you can only get if marketing and sales share a database.
Click rate. More reliable than opens, because a click is a real action.
Unsubscribes per campaign. A spike means frequency or relevance is wrong. Treat it as a warning, not a statistic.
Delivered rate. Below 95% means list quality problems.
FIGURE 3: A SENSIBLE ORDER TO START
Fix deliverability
- Domain authentication first. Nothing else matters if mail is not arriving.
Build one narrow segment
- From real order history. Compare it against a broad send.
Keep frequency the same
- Improve relevance, not volume. This is the discipline that matters.
A sensible plan
- Fix domain authentication before anything else
- Clean your contact list and confirm consent
- Use AI for subject lines first — lowest risk, easiest to test
- Build one narrow segment from real purchase history
- Measure it against a broad send
- Keep frequency where it is
- Report on revenue, not opens
The short version
AI makes marketing faster to produce and better targeted. If your campaigns run on your business data, the targeting is the bigger advantage.
The risk is not quality — it is volume. Cheap production tempts more sending, and frequency is what costs you a list.
Fix deliverability, segment narrowly, read every draft, keep the frequency, and measure revenue.
Want campaigns built from what your customers actually bought?
Get in touch. We will check your deliverability and consent position first, then build segments from real order history.
