Marketing Automation
Sequences that trigger on what people actually did, with consent handled properly.
What is marketing automation?
Marketing automation sends messages triggered by what a customer actually did rather than on a fixed schedule, across email, SMS and WhatsApp, with consent recorded and results measured against a group that received nothing.
The failure in most automation setups is volume without relevance. A sequence fires because seven days passed, not because the customer did anything, and the recipient learns to ignore you.
Triggering on behaviour is a different proposition. Somebody abandoned a cart, downloaded a price list, or has not ordered in ninety days when they used to order monthly. Those are reasons to make contact.
The second thing that separates a working setup from a nuisance is measurement against a control group, because without one you cannot tell whether the sequence caused the sale or the customer was going to buy anyway.
Why we insist on a control group
Duara sets up marketing automation from Nairobi across email, SMS and WhatsApp, which in this market usually matters more than email alone.
We hold back a control group on every sequence and we will argue for it if you would rather not. Without one, every result looks like a success, because customers who were going to buy anyway are counted as conversions and nobody can prove otherwise.
Who this is for
- Sequences fire on a timer rather than behaviour
- Nobody knows which message caused a sale
- Consent records are incomplete
- Lists are sent to indiscriminately
Industries we serve
- E-commerce and retail
- Property developers
- Insurance and financial services
- Education providers
- Travel and hospitality
- Equipment suppliers
Problems this solves
These describe sending that is measured by volume rather than by whether it worked.
- Sequences firing on a timer. A message goes out because seven days passed rather than because the customer did anything, so relevance drops and people disengage.
- No way to prove what worked. Conversions are attributed to whichever message came last, which credits the sequence for sales it did not cause.
- Consent records incomplete. Nobody can say when a contact agreed to be messaged, which is a data protection exposure rather than a marketing one.
- Everybody receives everything. The whole list gets every campaign, which drives unsubscribes and damages deliverability for the messages that matter.
What we deliver
Behaviour triggered sequences across email, SMS and WhatsApp, consent recorded with its source and date, control groups held back on every sequence, and reporting on what actually changed.
Sequences triggered by real customer behaviour
Email, SMS and WhatsApp handled together
Consent recorded with source and date
A control group held back on every sequence
Frequency limits so nobody is over messaged
Reporting on incremental effect, not just opens
Features
- Control groups as standard. A held back group on every sequence, because otherwise customers who would have bought anyway are counted as conversions.
- Behaviour, not calendar. Sequences fire on what somebody did, which is the difference between a relevant message and a nuisance.
- WhatsApp treated seriously. In Kenya it frequently outperforms email, and it is set up within the rules rather than through workarounds.
- Consent recorded with provenance. When and how a contact agreed is stored, since the record is what protects you rather than the intention.
- Frequency capped. A limit across all sequences prevents somebody receiving four messages in a day because three campaigns coincided.
Benefits
- You learn what actually works. Control groups turn reporting from a description of activity into evidence of effect.
- Fewer unsubscribes. Relevant, capped messaging keeps the list healthy, which protects every future campaign.
- Better deliverability. Engaged recipients and clean consent are what keep messages out of spam folders.
- Recovered sales you were losing. Abandoned carts and lapsed customers are the two sequences that most reliably pay for themselves.
Technologies we use
- Behavioural triggers. Firing sequences from real events in your website, CRM or order system.
- Multi channel delivery. Coordinating email, SMS and WhatsApp so channels complement rather than duplicate.
- Consent management. Recording opt in source and date, and honouring withdrawal immediately across all channels.
- Control group allocation. Randomly holding back a portion of each audience so incremental effect can be measured.
- Deliverability setup. Authentication records and list hygiene so messages reach inboxes rather than spam folders.
Deliverability is a technical discipline, not a content one. Authentication records, list hygiene and sending reputation determine whether your message arrives at all.
How the process works, step by step
Marketing automation usually takes four to seven weeks, and it starts with consent rather than with creative.
-
Step 1
Free marketing review
We look at your list, your consent records and what is currently being sent to whom.
-
Step 2
Consent audited first
We establish which contacts you can lawfully message, which sometimes reduces the list and is better learned now.
-
Step 3
Two sequences first
We start with abandoned cart and lapsed customer, since those pay for themselves most reliably.
-
Step 4
Controls and measurement
Held back groups are configured so the first results are evidence rather than impressions.
-
Step 5
Extended on evidence
Further sequences are added based on what the measured results actually showed.
Quality assurance and how we work
The commitments attached to messaging your customers at scale.
- Every sequence keeps a control group. Without one, customers who would have bought anyway are counted as conversions and every result looks like a success.
- We audit consent before sending anything. Which contacts you may lawfully message is established first, even where that reduces the list.
- We will not buy or use scraped lists. Bought lists damage deliverability and expose you legally, and we decline the work rather than doing it quietly.
- Withdrawal is honoured across all channels. Somebody who opts out of SMS does not keep receiving WhatsApp messages, which is a common and avoidable failure.
- We report sequences that did not work. Where the control group performed the same, we say the sequence had no effect rather than quoting open rates.
Frequently asked questions about marketing automation
Why do you insist on a control group?
Because without one, customers who were going to buy anyway get counted as conversions, and every campaign looks successful.
What if we do not want to hold anybody back?
We will argue for it. A small held back group costs a little revenue and is the only way to know whether the sequence works at all.
Does it work with WhatsApp?
Yes, and in Kenya WhatsApp frequently outperforms email. We set it up within the platform rules rather than through workarounds.
Can you use a list we bought?
No. Bought lists damage deliverability and expose you legally, and we decline that work rather than doing it quietly.
What happens to our consent records?
They are audited first, with source and date stored, because the record is what protects you rather than the intention.
What if our consent records are poor?
The usable list may shrink. That is better learned now than after a complaint, and we would rather tell you at the review.
Which sequences should we start with?
Abandoned cart and lapsed customer. Those two pay for themselves most reliably and prove the setup quickly.
How do you stop over messaging?
A frequency cap across all sequences, so three campaigns coinciding does not send somebody four messages in a day.
Does opting out of one channel stop the others?
Yes. Withdrawal is honoured across email, SMS and WhatsApp together, which is a common and avoidable failure elsewhere.
What does deliverability depend on?
Authentication records, list hygiene and sending reputation. It is a technical discipline rather than a content one.
How long before sequences are live?
Four to seven weeks, starting with the consent audit rather than the creative work.
Will you report campaigns that failed?
Yes. Where the control group performed the same, we say the sequence had no effect rather than quoting open rates at you.
Can it integrate with our online shop?
Yes, with common e-commerce platforms, which is what makes cart abandonment sequences possible.
Does it work with our CRM?
Yes, triggering on CRM events and writing engagement back so sales can see what a contact received.
Can we send in Swahili?
Yes, and language can vary by contact where you hold that preference.
How many sequences should we run?
Fewer than most people expect. Two working sequences beat eight that dilute attention and frequency budget.
Do you write the content?
We can, though most clients write their own with our structure. The offer and tone are yours to own.
Is SMS worth it given the cost?
For time sensitive messages, usually. We compare cost per outcome across channels rather than assuming email is cheapest overall.
Who owns the contact data?
You do. It stays in your systems, is never sold or shared, and trains no public AI model.
What does marketing automation cost?
Channel mix and message volume determine it, quoted after the marketing review.
Glossary of terms
Marketing automation terms in plain language.
- Control group
- A randomly held back portion of the audience that receives nothing, used to measure real effect.
- Incremental effect
- The additional result caused by the campaign, as distinct from what would have happened anyway.
- Deliverability
- Whether your messages reach the inbox rather than the spam folder.
- Behavioural trigger
- An action by a customer that causes a sequence to start.
- Frequency cap
- A limit on how many messages one person receives across all sequences.
- Consent record
- The stored evidence of when and how somebody agreed to be contacted.
Related services in Automation
Talk to us about marketing automation
Ask what your last campaign actually caused. If the answer is an open rate, that is the gap this closes.
Start your project