AI Marketing
Campaigns produced and scheduled without a marketing team.
What is AI marketing?
AI marketing is a system that drafts campaign content in your own voice, groups your customers by what they have actually bought or clicked, schedules the sending, and reports which messages produced results so the next round is better informed.
Marketing in most small businesses happens in bursts. A flurry of activity, then nothing for two months because everybody got busy with the work the marketing brought in.
This produces the material on a rhythm. Content drafted in your tone, customers grouped by real behaviour rather than guesswork, and everything queued ready to send.
You approve before anything goes out. The point is not that a machine takes over your marketing, but that the work is prepared and waiting rather than never started.
Why we keep an approval step
Duara sets up marketing automation from Nairobi for businesses that have a customer list they never message and no time to change that.
We build in an approval step and recommend keeping it. Clients often want to remove it once the drafts look consistent, and sometimes that is reasonable, but a marketing system sending unreviewed messages to your entire list is one bad draft away from a problem you cannot recall.
Who this is for
- You have a customer list you never message
- Marketing happens whenever somebody finds time
- You want different messages for different customers
- Nobody owns marketing day to day
Industries we serve
- Retail and eCommerce
- Hospitality and travel
- Fitness and wellness
- Education providers
- Property and estate agency
- Professional services
Problems this solves
These are the conditions that make marketing automation worth buying rather than another tool nobody opens.
- A list that never gets used. Contact details accumulate for years and nothing is ever sent to them, which is unrealised revenue sitting in a database.
- Marketing in bursts. Activity happens when somebody has a free afternoon, so there is no rhythm and no compounding effect.
- One message to everybody. The same email goes to every customer regardless of what they bought, which trains people to ignore you.
- No idea what worked. Campaigns go out and nothing is measured, so the next one repeats whatever the last one did.
What we deliver
Content drafted on a schedule in your voice, customers segmented by actual behaviour, sending queued for your approval, and reporting that tells you which messages to repeat.
Draft campaign content in your own tone of voice
Customers grouped by what they bought or clicked
Scheduled sending across email and messaging
Reporting on what worked, feeding the next round
An approval step before anything reaches customers
Suppression rules so nobody is over-contacted
Features
- Drafts on a rhythm. Content is prepared to a schedule so marketing stops depending on somebody having a free afternoon.
- Behavioural segmentation. Groups are built from purchase and engagement history rather than from assumptions about who wants what.
- Your voice, learned from your material. Drafts are built from what you have already written, so they read like you rather than like a template.
- Approval before sending. Nothing reaches customers unreviewed by default, and we recommend keeping it that way.
- Contact frequency limits. Suppression rules prevent the same customer receiving three campaigns in a week, which is how lists get destroyed.
Benefits
- A dormant list starts earning. Contacts you already own become revenue without any additional acquisition spend.
- Marketing becomes continuous. The compounding effect of regular contact is what produces results, and irregular bursts never achieve it.
- Relevance improves open rates. Messages matched to what somebody actually bought get read, which the generic version does not.
- Each round is better than the last. Reporting closes the loop, so what worked gets repeated and what did not gets dropped.
Technologies we use
- Behavioural segmentation. Groups derived from transaction and engagement data in your own systems.
- Email and messaging delivery. Sending through established providers with proper authentication so messages reach inboxes.
- Tone modelling from your material. Drafts built from your existing copy so the voice is consistent with everything else you publish.
- Suppression and frequency rules. Hard limits on how often any individual can be contacted, applied across all campaigns.
- Attribution reporting. Which messages produced opens, clicks and orders, tracked back to the campaign that caused them.
The segmentation matters more than the writing. Well-written content sent to the wrong group performs worse than plain content sent to the right one.
How the process works, step by step
Marketing automation is usually producing its first scheduled campaigns within three to four weeks.
-
Step 1
Free list review
We look at what contact data you hold and whether it is usable, which is frequently the deciding factor.
-
Step 2
Segment design
We build the groups from real purchase and engagement behaviour rather than from assumed customer types.
-
Step 3
Voice and template setup
We establish the tone from your existing material and build the templates campaigns will use.
-
Step 4
First supervised campaigns
Early sends are reviewed line by line, which is how the drafting improves quickly.
-
Step 5
Ongoing rhythm
Content is drafted to schedule, queued for approval, and reported on after each send.
Quality assurance and how we work
What we commit to on anything that messages your customer list.
- Approval stays on by default. Nothing sends unreviewed unless you explicitly change that, and we advise against changing it.
- Frequency limits are enforced globally. Suppression applies across all campaigns, so no customer receives three messages in a week regardless of which system sent them.
- Unsubscribes are immediate and permanent. Opt-outs take effect at once and are honoured across every future campaign without exception.
- We authenticate your sending domain. Proper email authentication is part of setup, because unauthenticated sending damages deliverability for everything you send.
- We will tell you if the list is unusable. Old, unconsented or poor quality contact data is a reason not to proceed, and we say so at the review stage.
Frequently asked questions about AI marketing
Do we approve before anything sends?
Yes, by default, and we recommend keeping it. A system messaging your whole list unreviewed is one bad draft from a problem you cannot recall.
Will it sound like a robot?
It drafts from your existing material, so the voice matches what you already publish. Anything that reads wrongly is edited at approval.
How does it decide who gets what?
Segments are built from actual purchase and engagement history rather than from assumptions about customer types.
Will customers feel watched?
Segmentation uses their dealings with you, such as orders and renewals. We keep wording natural rather than referencing their behaviour directly.
Can it send on WhatsApp as well as email?
Yes, where you have the appropriate consent and business account. Rules on frequency apply across both.
Our database is years out of date, does that matter?
That is the first thing we review. Old or unconsented data is a reason not to proceed, and we will say so rather than sending to it.
How often will it contact people?
Within the frequency limits you set. Suppression is applied globally so no contact receives three messages in a week.
Do unsubscribes work properly?
Immediately and permanently, honoured across every future campaign without exception.
How is our sender reputation protected?
We authenticate your sending domain as part of setup, because unauthenticated sending harms delivery for all your email, not just campaigns.
How long until the first campaign goes out?
Three to four weeks, most of which is segment design and establishing the tone from your existing material.
Can it write product descriptions too?
That is a separate piece of work, though it uses the same tone modelling. We scope it independently.
How is performance attributed across channels?
Reporting ties opens, clicks and orders back to the campaign that produced them, which is what informs the next round.
Does it need a marketing person to run it?
Somebody approves the queue, which takes minutes rather than hours. It does not need a dedicated marketer.
Is there a way to halt every campaign at once?
Yes, immediately and across all campaigns, which matters if something goes wrong elsewhere in the business.
Will it help with social media?
It can draft content, but scheduling and community management sit outside what we set up here.
What consent do we need?
Recipients must have agreed to hear from you. We check this at the list review because sending without it damages both deliverability and your standing.
Can campaigns be written in Swahili?
Yes, and mixed English and Swahili campaigns often perform better here than either alone.
Can different segments get different tones?
Yes. Corporate and consumer segments frequently need different registers, and the templates support that.
What does marketing automation cost per month?
It depends on list size and send volume. The monthly figure is projected from your own numbers before you commit.
What if performance drops over time?
Reporting shows it early, and the usual cause is frequency rather than content. Tightening suppression normally recovers it.
Glossary of terms
Marketing terms as used in this context.
- Segmentation
- Dividing your customer list into groups so each receives messages relevant to them.
- Suppression
- Rules preventing a contact from being messaged, whether through opt-out or frequency limits.
- Deliverability
- Whether your messages reach inboxes rather than spam folders, which depends heavily on authentication and list quality.
- Open rate
- The proportion of recipients who opened a message, used as a rough measure of relevance.
- Attribution
- Connecting an order or enquiry back to the campaign that produced it.
- Nurture sequence
- A series of messages sent over time to a contact who is not ready to buy yet.
Related services in AI Implementation
Talk to us about AI marketing
Tell us roughly how many contacts you hold and when you last messaged them. Those two answers usually determine whether this is worth doing.
Start your project