Updated: Nov 8
Email Personalisation That Pays: Better Journeys, Measurable ROI
By Trayan Hristov. Originally published 3 November 2026.

The best personalised email makes the next buying decision easier. It helps a first-time shopper choose the right fit, a new customer get value from their purchase, or a returning buyer find something worth coming back for.
That is the personalisation worth building. Not twenty versions of the same discount. Not a first name above a product grid that ignores what someone already bought.
At Thrivelia, we connect customer insight, brand-specific creative and lifecycle execution. The question is not how many segments we can create. It is which customer decision we can improve, what needs to change in the journey, and whether that change earns back the work.
Start with the next useful customer action
Personalisation can change the product, the explanation, the timing, the offer or whether a message should be sent at all. Start with the part that matters to the customer, then choose the simplest reliable way to deliver it.
Before the first purchase: remove a real hesitation
A shopper who selects a wide-fit preference needs available widths, a clear size guide and evidence that the shoe will work for them. Show that information in the welcome journey and continue it on the landing page. A personalised greeting does not compensate for sending them to an unavailable size.
After the first purchase: help the product succeed
A customer who has just bought a garment may need fit reassurance, care instructions and styling ideas before another offer. A device buyer may need setup support. Use reliable order and delivery information, and provide a way to get help. Do not assume an order confirmation means the product has arrived or been used.
Before the next purchase: recommend a credible next step
The next useful purchase might be another colour of a trusted style, a complementary piece, a refill or a compatible accessory. Use purchase history and product relationships to choose the hypothesis. A bestseller is not automatically the right recommendation for every previous buyer.
When the customer moves on: change the message
A second purchase should stop an obsolete first-purchase reminder. A return, service issue or subscription pause may change the next action. Sometimes the most useful personalisation is leaving someone out of a campaign.
Our Progressive Personalisation framework connects these changes across the relationship. The Marketing Hourglass helps identify which stage needs attention first.
Use fewer, stronger decisions instead of endless versions
Imagine a fashion brand with two groups of first-time buyers: customers who entered through a core tee, and customers who bought a statement piece. The useful question is whether a different product story helps each group find its next purchase. The answer does not automatically require two completely separate campaigns every week.
Start with the smallest meaningful difference. That might be a product module, a fit explanation or a short post-purchase branch. Keep the rest of the creative system consistent. A reusable module can preserve the brand and reduce repeated production work.
For each proposed change, write down the customer need, the evidence available, what changes in the experience, the fallback when data is missing, and the outcome that would justify keeping it. If nobody can explain what improves for the customer, the segment is not ready for production.
Keep uncertainty visible. Someone buying a gift may choose a size that is not their own. A historical category purchase is not a permanent preference. Let customers update what they tell you, and keep a useful general route for people whose needs are unknown.
An easier edit is not a new source of customer data
Klaviyo’s 6 November 2026 update adds no-code use of trigger-event data in flow email content. The show/hide documentation limits that event context to the relevant event-triggered flow. It does not retrieve arbitrary past events for every message.
For the illustrative fashion groups above, decide whether the treatment needs the product in one order or a current customer preference. Then inspect that exact source, write the missing-data route and test the destination. A quicker editing interface can reduce a production task; it does not demonstrate that the customer needed another version.
The TopReg case illustrates the historical event-data problem behind product-specific recovery. It is not evidence that the account used this later feature. Keep data readiness, editing cost and the measured value of the treatment separate in the ROI calculation.
What Thrivelia brings beyond segmentation
A segment is a rule. A useful lifecycle programme connects that rule to a product decision, a message, a working customer journey and an owner who checks the result.
Our approach is to diagnose the opportunity, brief creative around the brand and customer, build the necessary routing and exclusions, then use the results to decide what comes next. That can mean adding a journey. It can also mean consolidating several journeys that now do the same job.
The published perfectwhitetee work illustrates that broader approach. Tracked owned-channel revenue grew 94% in H2 2025 versus H2 2024, while tracked flow revenue grew 117%. Separately, reported flow revenue per delivery rose from $1.05 in H2 2025 to $2.13 in H1 2026. The programme developed to approximately 55 live flows before consolidation to approximately 45.
Those are account-level results across the stated periods, not a controlled estimate of personalisation alone. The practical lesson is to build useful coverage and then simplify overlap, rather than make the number of flows the objective. Our flow-efficiency review explains the calculations and comparison periods.
Measure the difference the treatment makes
Compare the personalised approach with a competent standard experience for the same eligible audience. Do not compare valuable repeat buyers receiving personalisation with unrelated new subscribers receiving a generic email. The groups already differ before you send anything.
For a practical test, randomly assign eligible customers to the standard or proposed treatment. Record the assignment and observation window before sending. For a sequence, keep that assignment consistent. Compare outcomes for everyone assigned, not only the people who opened or clicked.
Keep stock, offer, timing and the surrounding contact policy comparable when you are testing creative relevance. If several elements change together, describe the result as a comparison of the complete treatments, not proof that one particular element caused it.
Klaviyo supports campaign A/B testing. The available winner settings depend on the test configuration. A native winner does not automatically answer a profit question, so review the commercial result alongside the platform report.
A personalised-versus-standard test answers which experience performed better. Testing the value of sending compared with not sending requires a different comparison. Our attribution and incrementality guide explains that distinction.
Calculate retained contribution, then the extra cost
For each group, start with retained sales after discounts and refunds. Subtract the agreed variable order costs, such as product, fulfilment and payment costs. Use the same currency, period and definitions. Divide that contribution by the number of people assigned to the group.
Estimated added contribution = difference in contribution per assigned customer × customers assigned to the personalised treatment.
Net result = estimated added contribution − additional production, quality-checking, approval, data, software and maintenance costs.
Count the extra work required by the new treatment, not the entire agency retainer when both groups would incur it anyway. Record internal time as a capacity cost, and distinguish it from additional cash paid. Separate one-off setup from the ongoing cost of maintaining the treatment.
Worked example: the same framework can expose a loss or a useful opportunity
The following numbers are fictional arithmetic examples, not client results or promises. In each example, 20,000 customers are assigned to each group. Every qualifying purchaser makes one retained order contributing $40 before the additional treatment costs.
Measure | Standard | Personalised: example A | Personalised: example B |
Assigned customers | 20,000 | 20,000 | 20,000 |
Qualifying purchasers | 600 | 640 | 760 |
Purchase rate | 3.0% | 3.2% | 3.8% |
Observed added contribution | Baseline | $1,600 | $6,400 |
Additional treatment cost | Baseline | $1,900 | $1,900 |
Net after additional cost | Baseline | −$300 | $4,500 |
Example A matches the calculator’s starting example: 40 additional purchasers × $40 = $1,600, less $1,900 of extra work and expense, leaving a negative $300 observed result. A higher purchase rate has not paid back the treatment yet.
Example B shows why worthwhile personalisation deserves investment: 160 additional purchasers × $40 = $6,400, less the same $1,900, leaving $4,500. This is a separate fictional scenario, not a second result from the first test.
Under these equal-margin assumptions, at least 48 additional whole purchases would cover the $1,900 cost. That is an economic threshold, not a statistical-significance threshold. Neither scenario establishes that an observed difference is reliable. A real decision also needs sufficient outcome data, uncertainty assessment and a sound test design.
Download the Testing and Incident Workbook and open Personalisation ROI to change the assumptions. It calculates the economics, not statistical significance. Keep setup and recurring costs separate; do not assume a one-time effect repeats indefinitely.
Turn the result into a decision
A useful review ends with one of four actions: keep, improve, retest or stop. Keep a treatment when the commercial evidence and customer experience support it. Improve the implementation when the idea is sound but the route is broken. Retest when the result leaves a material uncertainty. Stop unnecessary work when there is no credible customer benefit or business case.
Check more than orders. Review returns, complaints, unsubscribes, product availability and maintenance effort. Do not reward a message that creates quick purchases by making an inaccurate promise.
For a smaller audience, favour a few meaningful comparisons over many cosmetic variations. An inconclusive result is not proof that personalisation does not work. It may mean the difference is too small to resolve with the available audience and time.
Keep the brand. Make the next decision better.
Personalisation should make a brand feel more attentive, not less recognisable. Product facts, tone, creative direction and approval still matter when AI helps draft variations. The customer should recognise the same brand, now answering a more relevant question.
Thrivelia can work with your existing creative team or take responsibility for the agreed lifecycle delivery. The first step is to identify the customer transition worth improving, the evidence available and who will own the result. We do not need to replace effective internal work to add value.
Book a strategy call to review where your customer journey should become more personal, where it should become simpler, and which changes deserve a test.
Frequently asked questions
Should every campaign have personalised versions?
No. Use a different treatment when reliable context supports a different customer need. A shared message can still be relevant. The goal is a better experience and commercial outcome, not a maximum version count.
What should a brand personalise first?
Start with a meaningful difference in the buying journey: first-time versus returning customers, a verified product preference, the first product purchased, or an appropriate next-order need. Prioritise the change with the clearest customer reason and enough eligible volume to evaluate.
Does a higher click rate prove a better result?
No. Clicks help diagnose the journey. Retained purchases, contribution, experience measures and the cost of operating the treatment determine whether the change deserves to stay.
Can our team keep the design and copy?
Yes, when that is the right operating model. Agree who owns analysis, prioritisation, briefing, implementation, approvals and measurement. A stronger strategy should make good internal creative more useful, not erase it.

.png)
Comments