Email marketing strategy: the complete guide

Why email remains the highest-ROI digital channel

Email marketing delivers a higher return on investment than any other digital marketing channel: on average $36 for every $1 spent, according to Litmus. It does this because it combines owned audience reach (no algorithm dependency), direct conversion paths, and measurable attribution in a single channel.

The challenge is that building an email programme that realises this potential requires more infrastructure and operational discipline than most organisations invest. A list of 50,000 contacts means nothing if deliverability is poor, segmentation is nonexistent, and the only emails sent are promotional blasts.

List building: the right way

The quality of your email list determines the ceiling of your programme performance. A list of 10,000 engaged, permission-based subscribers who actually want your content will outperform a purchased list of 100,000 contacts every time, both in terms of engagement metrics and, critically, deliverability reputation.

Effective list building now requires multiple acquisition touchpoints: website signup forms with clear value propositions, content upgrades that exchange genuinely useful resources for email addresses, event registrations, and (for B2B) carefully sequenced LinkedIn outreach that moves conversations to email.

Segmentation: the difference between broadcast and conversation

The single largest lever in email marketing performance is segmentation. Sending the same email to your entire list treats a prospect who has never heard of you and a customer who has purchased three times as if they were the same person, with predictably poor results for both.

Magic Pulpo recommends a minimum segmentation structure of four engagement tiers (highly engaged, moderately engaged, disengaged, inactive), three lifecycle stages (prospect, active customer, lapsed customer), and any product or interest-based segments relevant to your business. This 12-cell matrix enables personalisation that meaningfully improves performance without requiring engineering resources.

Automation: the sequences that matter most

Automation sequences, properly constructed, do the work of a sales team at scale. The three highest-value sequences for most businesses are the welcome sequence (the 5-7 emails sent to new subscribers in the first two weeks), the post-purchase sequence (onboarding and upsell for new customers), and the re-engagement sequence (win-back programme for disengaged subscribers before removing them from the list).

The welcome sequence deserves particular attention because new subscribers are at peak engagement: GetResponse's 2024 benchmarks put the average welcome email open rate at 83.63%, against 40.08% for newsletters (GetResponse). Apple's Mail Privacy Protection inflates both numbers, but the gap is what matters. A well-constructed welcome sequence sets expectations, delivers value, and begins qualifying intent within the first week of the relationship.

What AI genuinely improves in email

What has changed significantly in the last two years is how email marketing is produced, optimised, and personalised. AI-powered tools have materially altered the economics of email content production, the sophistication of segmentation available to mid-market teams, and the speed at which subject line, content, and timing optimisation can occur.

AI is most valuable in email marketing in three areas. Subject line generation and testing: AI tools can generate dozens of subject line variants from a single brief, and A/B testing infrastructure can identify winners at statistically significant sample sizes faster than previous manual processes. Content personalisation at scale: AI-powered personalisation can now serve genuinely different content to different audience segments within a single email send, based on behavioural data, purchase history, and engagement patterns, which was previously only available to enterprise teams with substantial engineering resource. Predictive send-time optimisation: machine learning models trained on audience engagement data can now predict the optimal send time for each individual subscriber rather than the optimal time for the list as a whole, with measurable impact on open rates.

What AI cannot replace

AI cannot replace the strategic judgment that determines what an email programme is trying to accomplish, which audience segments matter most, and what relationship the brand is trying to build with its subscribers. These decisions require understanding of the brand, the customer, and the commercial context that AI tools, which optimise for engagement metrics, cannot reliably infer. An AI-optimised subject line on an email with the wrong message for the wrong segment at the wrong stage of the funnel will generate higher open rates and worse business outcomes.

AI also introduces quality risks that require active human oversight. AI-generated email copy tends towards generic, competent prose that lacks the distinctive voice and specific detail that builds genuine subscriber relationships over time. The solution is to use AI-assisted production as a tool that accelerates human-directed content strategy, while strategic and creative judgment stays with people.

Deliverability: the invisible constraint

Deliverability (whether your emails actually reach the inbox) is the constraint that makes all other email optimisation meaningless if it fails. Poor deliverability stems from four primary causes: sending to unengaged lists (poor sender reputation), technical configuration errors (SPF, DKIM, DMARC misalignment), content patterns that trigger spam filters, and insufficient list hygiene.

Magic Pulpo manages email programmes for Berlitz across multiple markets using Salesforce Marketing Cloud, where deliverability is managed as an ongoing operational discipline, never as a set-and-forget configuration.

Salesforce Marketing Cloud ships its AI features under the Einstein name, and every one of them depends on the data underneath.

The most common failure pattern in enterprise email is a lack of clean, connected data. AI capability is rarely what is missing. Einstein send-time optimisation requires reliable engagement data history. Einstein content recommendations require a coherent content taxonomy. Predictive audience segmentation only works with purchase and behavioural data that is current, accurate, and properly linked across systems. Before investing in AI-powered email features, organisations should audit the quality of the underlying data infrastructure that powers them.

To discuss building or improving your email programme, speak to the Magic Pulpo team.