Picture a Dubai retail group preparing its next promotion. The agency has Arabic and English creative ready. CRM holds store sign-ups, online subscribers and an older event list. The campaign dashboard reports successful delivery, but customer service hears that the offer never arrived. The obvious response is another subject line or another platform. Email deliverability needs a more disciplined diagnosis: what happened between the approved audience, the sending system and the recipient's mailbox?
I would ask for one recent campaign and its evidence before approving either purchase. Separate four events: the platform queued a message, the receiving service accepted it, the message reached an inbox, and the customer acted. A report that proves one event should not be used to claim all four. Otherwise, the team can spend weeks optimising creative around a delivery failure.
Email deliverability starts with the actual sender
Make a short inventory of everything sending on behalf of the brand: campaign software, CRM sequences, ecommerce receipts, loyalty messages and agency tools. Record the visible sender address, sending provider, message purpose and accountable owner. Include abandoned experiments that still run scheduled messages. Start with the systems the business actually uses, not the supplier list procurement remembers.
Then collect a received test message from each active route. Ask the technical owner to explain the authentication results in its headers. SPF checks the sending infrastructure against the domain's authorisation; DKIM provides a domain signature; DMARC checks alignment with the visible sender domain and publishes a policy. A green configuration screen is useful, but the received message shows what that route actually produced.
Google's sender guidelines require SPF or DKIM for all senders to personal Gmail accounts. Its bulk-sender requirements include SPF, DKIM, DMARC and alignment, with additional requirements for marketing and subscribed messages. The guidance also covers transport encryption, DNS and message formatting. Have the owner check the complete applicable requirements, rather than treating one authentication result as an inbox guarantee.
Diagnose the failure in five passes
1. Separate rejection from placement
Export the campaign's delivery events by receiving domain. Distinguish permanent rejection, temporary deferral and acceptance. Keep the original response code and explanation. Ask whether the platform's delivered figure means remote-server acceptance; do not quietly rename it inbox placement in the management report.
Use controlled test mailboxes to inspect where messages land, while recognising that a few test accounts cannot represent every subscriber. Compare provider feedback with customer reports and campaign behaviour. If one mailbox provider rejects a route, investigate that route. If accepted messages reach spam, investigate reputation, audience and content together. The same resend is not a diagnosis.
2. Give promotional and operational mail separate treatment
An order receipt and a discount sequence have different business consequences. Map their sending routes and volumes separately. If a campaign performs badly, the operations team should be able to see whether receipts and password resets are also affected. A single blended success rate conceals the messages the business cannot afford to lose.
Yahoo's sender guidance recommends separating bulk marketing from transactional and user mail by IP or DKIM domain. It also sets authentication and complaint requirements, and requires bulk senders to honour unsubscribes within two days. Translate that guidance into a plan with your provider. Separation helps organise sending and reputation; it does not excuse unwanted traffic or make a new domain trustworthy automatically.
Do not rush domain or authentication changes immediately before a major promotion. First establish which systems depend on the current configuration and how each change will be tested. The business owner should know which messages could be affected and who can stop the rollout.
3. Inspect how the audience entered the system
Break the audience into meaningful sources: an online newsletter subscription, a store promotion registration, a recent customer relationship or an imported historical list. Record the expectation attached to each. An address existing in CRM is not evidence that its owner expects this particular campaign.
For a Gulf group, preserve language and brand preferences. Someone who subscribed to a specific Arabic newsletter should not suddenly receive every English promotion from every business in the group. Deduplication alone cannot resolve conflicting preferences. The adjacent note on first-party data strategy covers the broader purpose and permission record; here, the immediate question is which audience should enter this sending run.
Quarantine records whose origin or current status the team cannot explain. Examine complaints, permanent bounces and unsubscribes by audience source, not just across the campaign. A large established segment can hide a smaller imported segment that produces most of the damage. Assign the repair to whoever owns collection and list handling.
4. Test the exit from the customer's side
Open an actual promotional message in a test mailbox. Check the visible unsubscribe link and the mailbox-level one-click mechanism separately. Follow the request into the sending platform, CRM and any connected journey. Verify that a scheduled export or later list import cannot silently make the same address eligible again.
RFC 8058 specifies an HTTPS POST mechanism for one-click unsubscribe using email headers, with those headers covered by a valid DKIM signature. A link that opens a preference page is not, by itself, proof that this mechanism works. Ask the provider to demonstrate both the message headers and the resulting suppression.
Make withdrawal understandable in Arabic and English where those languages are offered. Define which subscription stops and how other choices remain visible. Test the next scheduled campaign against that status. The useful evidence is that the address stays excluded from the relevant promotional stream, not merely that a confirmation screen appeared.
5. Resume with an explicit stop rule
After repairing an identified problem, choose a limited, clearly eligible audience and monitor it before expanding. Agree who can pause sending when rejections, deferrals or complaints worsen. Keep provider-specific results separate, and compare similar audience sources and message purposes. Do not use a healthier segment to declare every route repaired.
The weekly review can fit on one page: route, audience source, attempted messages, acceptances, rejection reasons, complaints, unsubscribes and verified commercial actions. Show missing evidence explicitly. A low open rate alone does not isolate the cause, and a higher open rate does not prove that the campaign produced useful business.
Make sending quality an operating responsibility
This belongs inside marketing automation work, because the problem crosses collection, CRM, campaign execution and customer choice. Give one person responsibility for the full sending path, with clear technical and commercial counterparts. Review it whenever a new brand, agency or application starts sending.
Email deliverability improves as a management discipline when every campaign has an explainable audience, a verified sender and a working exit. Before buying another tool, put one message through that complete path. If the team cannot show where it went and why it was sent, increasing volume only makes the uncertainty more expensive.