Lakshmi Nair is a founder based in Dubai, UAE. She is building Oula and GUTSY. Before founding, she scaled Deliveroo and Talabat across 8 markets with €100M+ annual budgets.

Oula finds the best version of what you love across every store in the UAE, and tells you if the price is right. It learns your taste, searches 20+ stores at once, and gives an honest read on every price. It has been approved for the App Store and is launching soon on iOS, as a Safari extension and at getoula.com.

GUTSY is a gut-friendly vegan protein brand founded by Lakshmi Nair in Dubai: hydrolysed pea and rice protein with Actazin® kiwi extract, no gums and zero proprietary blends. It was born from 47 kitchen experiments and has been on sale in Dubai since 1 November 2026 at alwaysgutsy.com.

Deliveroo · 2019Business StrategyGlobal standard · A/B tested

Turning Bad Orders into Loyalty

+12%MONTHLY ORDER FREQUENCY

GlobalROLLOUT ACROSS DELIVEROO

A/BRANDOMISED TEST

Problem: Orders that went wrong were treated as a cost to minimise with a one-off refund, and those customers quietly churned.
What I did: Replaced the refund with free deliveries and vouchers spread across the next 5 orders, and tested it against the standard response.
My role: Designed the programme and ran the randomised A/B test end to end, without analyst or data science support.
Result: Monthly order frequency up 12% against the control group. Adopted as a global standard across Deliveroo.

THE SITUATION

Every food delivery business has the same dirty secret: some orders go wrong. The food arrives cold. The bag gets damaged. The rider is 40 minutes late on a Friday night when someone was really hungry.

The standard industry response was to apologise, maybe refund the order, and move on. Treat it as a cost centre. Minimise the damage and hope the customer forgets.

We saw it differently. A customer who had a bad experience wasn't just a complaint to close. They were a moment. Handle it badly and they churn quietly, telling three friends on the way out. Handle it well and you might just make them more loyal than they were before it happened.

THE IDEA

Instead of a flat refund, we gave affected customers free deliveries and vouchers spread across their next 5 orders. Not a one-time sorry. A reason to come back five times.

The logic: a refund closes the loop. Five order credits reopens it. You're not compensating for a bad experience, you're buying back the habit.

I designed the program and ran the AB test independently, no data science team, no analyst support. A randomised customer group received the compensation package after a poor delivery experience. The control group received the standard response. I tracked order frequency, retention, and basket size across both groups through SQL pulls and Excel modelling, measuring whether the investment in compensation was actually changing behaviour.

Customer has a bad deliveryCold food, damaged bag, a very late rider.

Randomly assigned

ControlThe standard response: an apology and a one-off refund.

VariantFree deliveries and vouchers spread across the next 5 orders.

MeasuredOrder frequency, retention and basket size, both groups.+12% monthly order frequency

Test design, simplified

WHAT HAPPENED

It was immediate. The compensated group came back faster and ordered more often: monthly order frequency rose 12% against the control group, and retention was higher too. The program wasn't just recovering churned customers. It was producing better customers than we had before the bad experience.

The results travelled up the organisation faster than expected. Within months, the program was adopted as a global standard across Deliveroo worldwide.

WHAT IT TAUGHT ME

The best interventions don't just fix the problem. They use the problem as raw material for something better.

A bad delivery experience, handled correctly, turns out to be one of the cheapest loyalty tools a delivery business has.

Back to work Next · Not Everyone Needs 30% Off