Natura &Co · Direct selling at scale · Senior product designer · 2021 to 2022
Whoever is at the checkout is working
A direct selling platform used by more than 1.2 million sales consultants. The person opening the checkout is not buying for herself. She is restocking the business she runs, with a large order, many line items and margin on her mind, almost always on a phone and almost always outside business hours.
At that scale, one checkout step that stalls becomes lost revenue every week. This case is about how I went after that, and also about what I do and do not have to prove what happened next.
- Role
- Senior product designer, allocated by Triggo Labs inside the company's team. Redesign of the B2B platform and of the checkout flow, with CRO and mobile first design.
- Context
- Direct selling with more than 1.2 million consultants, short campaign cycles, orders with dozens of items and revenue that lands every week.
- Method honesty
- The screens, the funnel dashboards and the test reports stayed with the company. This page carries the framing, the method and the figure the company reported. Where I am missing measurement of my own, the page says so.
From a catalogue platform to a restocking flow
- 01
Context
A B2B direct selling platform, with more than 1.2 million consultants and short campaign cycles.
- 02
Reading the funnel
Looking at the checkout step by step to find where people give up, instead of redesigning the whole platform.
- 03
Mobile first
The order is built on a phone, outside business hours. The design starts there and not on the large screen.
- 04
Short cycle
Continuous discovery with the product owner, engineering and data inside the sprint, with no separate research phase.
- 05
Outcome
The company reported around US$ 2 million per week in additional sales.
The cart of someone buying to resell
A regular retail checkout holds three or four items and one consumption decision. A direct selling checkout holds dozens of lines and a business decision. The consultant is working out what she can turn over this cycle, how much margin is left, what ran out last campaign and what her own customers already asked for.
That changes what counts as friction. In a consumption cart, one extra field delays the purchase. In a restocking cart, one extra field in the middle of a large order makes the person postpone, and postponing means the next cycle. The cost of friction is not that session's sale, it is the whole cycle.
The question stops being how to make buying more pleasant and becomes where the large order stalls. Those are different problems and they ask for different screens.
The second context fact is the device. The order is built on a phone, standing up, at night, between other things, with interruptions in the middle. Mobile first here was not a method preference, it was where the revenue happens. Any step that needs long typing, a long number or a trip to another app to check a detail turns into abandonment.
Go where the funnel shows the leak
On a platform that size the temptation is to redesign everything. A broad redesign is easy to sell internally, takes months and leaves no way to tell what caused what. I went the other way: read the funnel step by step, pick the steps with the highest drop off and change them one at a time, so each change has an author.
The work ran inside the sprint cycle, with the product owner, engineering and data at the same table. Continuous discovery, with no separate research phase that delivers a report and disappears. In a product that bills every week, research that takes a quarter to become a screen misses the very cycle it meant to improve.
A change with no boundary teaches nothing. If five things ship together and the number moves, nobody knows which one paid for it, and the team repeats all five next time out of superstition.
One caveat I make on purpose, because this is where an evaluator pushes. I will not name here which checkout step leaked most, nor reproduce the funnel numbers from memory. The dashboards stayed with the company, and a funnel number rebuilt from memory is exactly the kind of detail that falls apart at the first interview question.
The figure the company reported
The caveat follows the same rule as the rest of this site. The figure of roughly two million dollars per week is what the company measured and communicated in the period, and the report stayed inside. Treat it as what it is, the revenue movement the company recorded after the checkout redesign, and not as measurement I hold in hand and can open.
Checkout conversion, before and after, step by step, is the number that would close this case with data of my own. It is pending, and I would rather leave the hole visible than fill it with an estimate.
Who the person is changes what friction means
The most useful thing in those ten months was to stop treating the checkout as a checkout. The same screen, with the same number of fields, is comfortable for someone buying one lipstick and exhausting for someone building a forty line restocking order. The discomfort does not live in the number of fields, it lives in the distance between what the person is doing and what the interface assumes she is doing.
If I started over, the first thing I would ask for is the breakdown of orders by size. Small orders and large orders are different behaviours inside the same funnel, and averaging the two hides exactly where the money sits.
