Where AI actually helps in email
Four places, and they are not equally valuable.
Drafting. Subject lines, body copy, variations to test. The most obvious use and a real time saver.
Segmentation. Working out which customers a message actually suits, from your live order history.
Send timing. Choosing when to send to each recipient rather than to everyone at once.
Content selection. Choosing which products or articles to show each person.
The second one is the most underrated and, in Odoo, the one with the biggest structural advantage.
Drafting
The straightforward use. You describe what the email is for and get a draft.
What it is genuinely good at:
- Subject line variations to test against each other
- A first draft when the blank page is the obstacle
- Adapting one message for different audiences
- Shortening something that is too long
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 email
The rule: read every word before it sends. Generated copy is fluent, and fluent is not the same as accurate — especially where prices, dates and availability are involved.
FIGURE 1: WHERE AI HELPS IN A CAMPAIGN
Drafting
- Subject lines, first drafts, variations. Fast, and always read before sending.
Segmentation
- Deciding who a message suits, from live order history. The biggest advantage in Odoo.
Timing
- Sending when each person is likely to read, rather than all at once.
Segmentation — the real advantage
This is where running email inside Odoo differs from running it in a separate platform.
A standalone email tool knows what it has been given — an uploaded list, some tags, and how people responded to previous emails.
Odoo knows what people bought.
That means a segment can be defined by real business facts:
- Customers who bought a specific product and have not reordered
- Customers whose order value has dropped compared to last year
- Buyers of one product who have never bought its natural companion
- Customers with no order in six months
- Leads at a particular pipeline stage for more than 30 days
No export, no upload, no list that was accurate three months ago.
Narrow beats broad, consistently. A message to 200 people who bought a specific product will outperform a message to 2,000 vaguely relevant contacts on every measure, including revenue and unsubscribes.
Send timing
Sending each recipient at the hour they usually open, rather than everyone at nine on Tuesday.
Honest assessment: it produces a modest improvement. Worth having, not worth building a strategy around.
The bigger levers are relevance and frequency. If open rates are poor, timing is not why.
The consent line
Automation makes it easy to send more. That is the risk, not a benefit.
Three rules that do not bend:
Only email people who agreed to hear from you. A business card at an event is not a subscription. Neither is a customer having bought something once, in many jurisdictions.
Include an unsubscribe link and honour it immediately. Odoo handles this once configured. Verify that it is.
Keep a record of consent — when and how each contact opted in.
Being careless here does two kinds of damage. There is legal exposure, which varies by country. And there is deliverability damage, which is immediate and affects every future email you send, including the ones people wanted.
FIGURE 2: WHAT PROTECTS YOUR SENDING REPUTATION
Authenticate your domain
- SPF, DKIM and DMARC. Without these, a growing share of your mail is never delivered.
Send only to people who opted in
- Bought lists destroy reputation faster than anything else, and are illegal in many places.
Clean your bounces
- Repeatedly mailing dead addresses tells providers you are careless.
Watch complaint rates
- If people mark you as spam, fix the cause rather than the number.
The frequency problem
The most common way companies damage their own list.
AI makes producing an email fast. Fast production creates the temptation to send more often. More frequent sending produces unsubscribes and spam complaints, which reduce deliverability for everyone on the list — including the people who wanted to hear from you.
A useful test before any send: if this email did not arrive, would anyone notice?
If the honest answer is no, do not send it. Producing an email cheaply is not a reason to send it.
Measuring it properly
Four numbers, and one of them is better than the others.
Delivered. Sent minus bounced. Should be above 95%.
Open rate. Directionally useful. Modern privacy features inflate it, so read it as a trend rather than a figure.
Click rate. More reliable, because a click is a real action.
Revenue. Because Odoo links mailings to your sales data, you can see the orders that followed a campaign.
Report the fourth one. Open rates are soft and most email tools cannot give you the fourth at all. This is the advantage of running campaigns on the same database as your orders.
FIGURE 3: TWO WAYS TO USE AI IN EMAIL
Used well
- Narrow segments from real order history
- Every draft read before sending
- Frequency unchanged, relevance improved
- Measured against revenue
Used badly
- Broad lists because drafting got cheap
- Generated copy sent unread
- More emails because they are easy to make
- Measured on open rate alone
A sensible way to start
1. Fix deliverability first. Domain authentication before anything else. There is no point improving copy that is not being delivered.
2. Clean your list. Remove bounces, confirm consent, merge duplicates.
3. Use AI for subject lines first. Lowest risk, easiest to test, and subject lines are the highest-return thing to improve.
4. Build one narrow segment from order history. Something a standalone tool could not have built. Measure it against a broad send.
5. Keep frequency where it is. Improve relevance, not volume.
6. Report on revenue. Not opens.
What not to automate
Sending without a human read. Customer-facing text, at scale, unreviewed, is how a wrong price or a discontinued product reaches thousands of people at once.
Consent decisions. Whether someone opted in is a fact, not a judgement call for software.
Adding people to lists automatically. Someone submitting a support ticket has not subscribed to your newsletter.
The short version
AI makes email faster to write and better targeted. In Odoo the targeting is the bigger advantage, because your segments can be built from what people actually bought.
The risk is not quality — it is volume. Cheap production tempts more frequent sending, and frequency is what costs you a list.
Fix deliverability, segment narrowly, read every draft, keep the frequency, and report on 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 your real order history.
