Retention workshops: five practical guides and six brand experiments
By Trayan Hristov · Founder, Thrivelia · 19 November 2026
If we worked on your retention together, I’d want the buyer, the person answering support tickets and the analyst in the same conversation. Each sees a different part of the decision. A review can tell us why the customer hesitated. It can’t tell us how many others walked away or what it would cost to change the outcome.
Start with the guide that matches a decision your team needs to make. The brand articles each have their own question, source material and proposed experiment. They are independent editorial analyses of public information. The named teams haven’t supplied the proposals or approved the conclusions.
Five questions to work through with your team
Six public brand workshops
TALA: choosing a second pair after the first fit is known. Two reviews disagree on length. A merchandising lead could test whether remembering a kept fit helps someone choose again.
Trinny London: what happens after the shade match? A recommendation, a purchase and a shade the customer actually likes are three different pieces of information.
DUSK: which decision follows a sofa swatch? A scan can resolve fabric choice while room access and delivery are still open.
On: a replacement pair or a different running shoe? The next recommendation changes once a runner tells us what the second pair is for.
Oura: who needs help after a ring is bought as a gift? A gift has a buyer, a wearer and a handoff. A hardware order and a membership renewal need separate clocks.
e.l.f.: can a short application demo earn another order? The product already comes with a specific instruction. The experiment asks when that instruction is worth showing.
What counts as retention for each of these six brands?
Give merchandising, customer care and analytics the same six examples. Ask each person to name the event they would count as a win. I’d resolve any disagreement before asking anyone to write copy or build a segment. Here’s what I’d put on the worksheet.
Workshop | The event worth following | The easy counting mistake |
TALA, fashion | First leggings kept, then a paid second pair kept after the returns window. | A size swap creates a second order record. |
Trinny London, beauty | A buyer confirms a shade, then pays for and keeps a refill or a complementary product. | A free shade correction looks like a repeat sale. |
DUSK, furniture | A swatch project becomes a delivered, kept sofa, then the household starts another room. | Several parcels for one project look like repeat demand. |
On, footwear | A runner keeps a second pair and identifies the job it serves. Follow the third observed purchase. | A gift or a first direct purchase gets labelled as a replacement. |
Oura, wearable membership | A gift recipient activates, then reaches and pays their first self-funded renewal. | A prepaid term applied during setup looks like a renewal. |
e.l.f., beauty | A buyer keeps a second purchase. Record mini to full size, same-formula refill and finish switch separately. | More mini buyers lift the blended repeat rate. |
To forecast the fourth purchase, start with one group of first buyers of the same entry product and follow them through three purchase windows. If 1,000 first buyers produce 200 second buyers, 70 third buyers and 28 fourth buyers over the specified stage windows, the conditional rates are 20%, 35% and 40%. Only 2.8% of the initial group reach four. These are example counts from one cohort. Set the clock again at each kept purchase and allow every included buyer the full follow-up time. Build a first-to-fourth purchase forecast.
The question I’d leave on the table is which of those three steps is both fixable and worth fixing. In fashion, a missing size can make a good creative look weak. In furniture, a failed delivery can wipe out the margin from a successful swatch journey. Let the person who deals with the exception challenge the numbers first.
Bring one assumption you disagree with
For a 30-minute working session I’d spend the first ten minutes listening to support and merchandising describe the same problem. Use the next ten to choose one intervention and the simplest comparison. Spend the final ten on ownership, data quality and what would make you stop. If nobody can agree what would disprove the idea, keep it in research.
Download the editable retention workshop workbook. It includes an evidence log, a problem-to-journey brief, a repeat-order calculator and a sheet for the brand’s response. Keep a working copy in your own shared drive and give each assumption an owner.
If you work at a featured brand, you can correct a public detail, tell us this has already been tested, or suggest a better experiment. Send the article title, the specific passage and any public link to office@thrivelia.com. Tell us whether we may quote and credit you. A private reply stays out of the article unless you agree otherwise.

.png)



Comments