Email marketing agencies have always sold personalisation. The problem was delivering it at scale. Writing 12 variations of a subject line, building dynamic content blocks for six segments, testing send times across time zones. The work was valuable but grinding.
AI has changed the economics of all of it. The agencies adapting fastest are not just using AI to write emails quicker. They are rethinking what “personalised” means when the constraint is no longer time.
Subject line optimisation
This is the entry point and the easiest win. AI generates dozens of subject line variations in seconds, scored against engagement criteria: length, emotional trigger, clarity, urgency.
What we are seeing: Agencies that test AI-generated subject lines against their own are reporting 12-18% higher open rates on average. The improvement comes from volume. Humans write 3-5 options and pick the best. AI writes 30, and the best of 30 is almost always better than the best of 5.
The practical workflow:
- Brief the AI with the email topic, audience segment, brand voice, and any constraints (character limits, banned words).
- Generate 20-30 options.
- Short-list 4-6 for A/B testing.
- Let the data decide.
Time saved: 15-20 minutes per email. Multiplied across a client sending 3-4 emails per week, that is 3-5 hours per month per client.
Dynamic content generation
This is where AI delivers the most transformative impact. Dynamic content (different copy blocks served to different segments within the same email) used to require writing multiple versions manually. Most agencies either charged heavily for it or simply did not offer it at the depth they knew they should.
AI removes that bottleneck. Give it the base message, the segment definitions, and the tone parameters, and it produces tailored variations for each group.
Example: A retail client with five customer segments (new customers, repeat buyers, lapsed customers, high-value customers, bargain hunters). Instead of writing one generic email, AI generates five versions with tailored product recommendations, different value propositions, and segment-specific calls to action. One brief, five outputs, 10 minutes of editing.
Performance impact: Agencies running segment-specific dynamic content report 25-40% higher click-through rates compared to one-size-fits-all sends. That is the kind of result that keeps clients.
Send time optimisation
Most email platforms now offer some version of AI-powered send time optimisation. Mailchimp, Klaviyo, and HubSpot all have features that analyse individual subscriber behaviour and deliver emails at the time each person is most likely to engage.
The agency angle: This is a feature you should be activating for every client, not selling as a separate service. It requires no additional work once enabled and typically improves open rates by 5-10%. Position it as part of your standard service and let it compound results quietly.
Where it gets interesting: Combining send time optimisation with time-zone-aware content. An email about a flash sale can land in every subscriber’s inbox at 10am local time, with content adjusted for regional preferences. AI handles the scheduling and the variation.
Segmentation improvements
Traditional segmentation relies on demographic data, purchase history, and manual tagging. AI-powered segmentation goes further by identifying behavioural patterns that humans miss.
Predictive segmentation: AI analyses engagement patterns to predict future behaviour. Which subscribers are likely to churn? Which are ready to purchase? Which respond better to educational content versus promotional offers?
Practical application: Feed your client’s subscriber data (engagement metrics, purchase history, email interaction patterns) into an AI tool and ask it to identify distinct behavioural clusters. You will often find segments you did not know existed.
What this unlocks: Instead of broad segments like “opened an email in the last 30 days,” you get precise groups like “subscribers who engage with product content on weekdays but only purchase after weekend promotional emails.” That specificity drives results.
A/B test generation
Most agencies test too little because creating variations takes time. AI eliminates that friction.
Beyond subject lines: Test email structure, CTA placement, copy length, tone, imagery descriptions, and offer framing. AI generates the variations; you set up the test matrix.
The system that works:
- Identify the element to test.
- Generate 4-6 variations with AI.
- Set up the test in the email platform.
- Run for statistical significance.
- Feed the result back to AI for the next round of testing.
Agencies running continuous AI-assisted A/B testing across all elements report 20-30% performance improvements within three months. Not from one breakthrough insight, but from dozens of small, compounding gains. This is the kind of measurable improvement that makes the real ROI of AI easy to demonstrate.
Lifecycle sequence creation
Building a complete email lifecycle (welcome series, nurture sequence, re-engagement flow, post-purchase series, win-back campaign) used to take 2-4 weeks of strategy and copywriting. AI compresses this dramatically.
The workflow:
- Map the lifecycle stages with the client.
- Define the goal, tone, and key messages for each stage.
- Generate the full sequence with AI (subject lines, body copy, CTAs).
- Edit for brand voice, accuracy, and strategy alignment.
- Build and launch.
Time comparison: A 12-email welcome and nurture sequence takes 20-30 hours to write manually. With AI generating first drafts, that drops to 8-12 hours. The quality of the strategic thinking stays the same; the production time collapses.
Deliverability monitoring
AI is increasingly useful for deliverability management, though this remains more tool-dependent than the other areas.
What AI can do: Analyse engagement patterns to identify deliverability risks before they become problems. Flag when open rates drop in specific email providers (Gmail vs Outlook). Suggest list hygiene actions based on engagement decay patterns.
Current tools: Validity (formerly Return Path), GlockApps, and built-in features in Klaviyo and HubSpot all offer AI-assisted deliverability monitoring. For agencies managing multiple clients, centralising deliverability monitoring with AI alerting saves hours of manual checking.
What requires human strategy vs what AI handles
The line is clear.
AI handles: First-draft copy, subject line generation, content variations, send time optimisation, segmentation analysis, test variation creation, performance pattern detection.
Humans handle: Brand voice and tone decisions, strategic sequencing, offer strategy, audience understanding, creative direction, client relationship and context, interpreting results in the context of business goals.
The mistake to avoid: Letting AI drive strategy. AI is exceptional at producing content within strategic parameters. It is poor at setting those parameters. The agencies getting the best results use AI to execute faster while keeping strategic decisions firmly with their strategists. Knowing when not to use AI is part of getting this right.
The agencies that win in email marketing over the next two years will not be the ones writing better emails. They will be the ones delivering true personalisation at a scale that was previously impossible, and pricing accordingly.
This is part of Delivery Notes, a series on implementing AI inside your agency. Subscribe to the newsletter to get new articles weekly.
