< Home 

Designing world’s largest online platform for trading flowers and plants.

World’s largest market for trading of flowers and plants hired McKinsey to help it transition from traditional brick-and-mortar company to a digital leader. The challenge at hand required re-organization of its business model, operations, revenue streams and of course - a series of best-in-class digital properties. As a lead designer on the project, I had to figure out what those digital properties would have to be.



Research

Except the fact that my university took us on a field trip to see the client for “Operations Management” class (seeing the building where 80% of world’s flowers and plants are sold every day is impressive enough), I had no other prior knowledge of the doman. Especially all of it’s operational, logistical, financial and legal nuiances. And now, with 2 other designers, we had to figure it all out in a matter of couple of months! We felt completely lost, but as designers this wasn’t our first time feeling this way ;) But where do we start? Such market was abundant with complexities. Regulations and the way the business operated seemed different for EU and non-EU countries. There were growers, buyers, intermediaries and logistical partners. There were flowers, plants, potted plants and at least a dozen of ERP systems at all stages of the supply chain. Surprisingly, very little was standardized in this industry. The good news was that we had the power to change that.

Being mindful of potentially trying to boil the ocean, we knew we had to focus our efforts on a single, relatively straight-forward line of business, and then build up use-cases from there. In a workshop with company’s senior management, incl. the CEO and CFO, a decision was made to initially focus on local growers, selling potted plants (plants that you see in a supermarket, sold in a fancy pot, sometimes with some ornaments in it). Being located within an hour drive from growers and buyers of such products, it was easy to access them for interviews. So after having interviewed numerous interal “knowledge experts” to prepare ourselves, we went into the field. It was a mind-boggling experience to visit the greenhouses and to see football fields of colorful flowers growing everywhere, with sophisticated computers controlling hydration, temperature and light exposure.

The growers were happy to cooperate and help shape the system together with us. Working actively together with users is a always a good idea, but you really can’t do without it, if you’re operating in a completely new domain like this. We combined interviews with shadowing, to see exactly how current systems are used, who talks to whom and what artefacts are there. What we found was a lot of paperwork, random Excel sheets and personell hired just to manage the complexity of the system. After interviewing intermediaries and buyers, we saw exactly the same situation on the other side of the aisle. The system was broken, but it was the way business has been conducted for years, usually by older men who didn’t want to change. Their children (flower growing is a very family-based business) were much more open, so it’s usually them that we worked with (they were all in their 20s-40s).

After the research we found that the most difficult part of the transaction, the one that no current system accomodated (and was hence the source of all those Excel sheets), was communicating the structure of a potted plant. For example, I’m selling you:

1x trolley with
4x shelves (shelves mattered for construction of the trolley) of
10x boxes, each containing
16x units, each containing
1x pot (material: clay; colors: could be red, white or brown)
1x plant (of such growth stage; such quality; not blooming)
3x ornamental decoractions

It was very difficult for growers to know if they had enough internal supplies to accomodate orders of such immense complexity, as well as to budget what such orders would cost. For sellers, placing such orders was equally difficult and there were a lot of mistakes in orders.

Moreover, there was no central online marketplace for such orders. It had to be built from scratch, keeping in mind everyone (including 3rd parties like logistics providers) would have to be able to access it through APIs of their own respective ERP systems.

To conclude this stage, the team has created detailed Personas and User Journey maps.

Ideation

To get everyone on board, we’ve set up a series of ideation workshops to examine what the new solution could look like for growers, buyers and 3rd parties. We even invited a few representatives of each group to join us for the workshops. The team came up with a number of solutions, ranging from greenhouse drones counting the inventory, to HD 3D cameras to capture every detail of the sold plant. To “measure the temperature” of the new ideas, we’ve created conceptual prototypes of each, trying to visualize the main aspects of each idea. This phase was coinciding with one of the industry’s main events and we asked the business representatives to test the concepts at the conference. At first, everyone was very uncomfortable, but we explained that this was the pain of becoming an organization that’s not afraid to talk to its users and to get them involved early. After the team came back, they were shining with excitement and entousiasm - ready to share everything they’ve heard. A list of ideas became shorter.

At this stage, we were working with a group of Product Owners, actively discussing the broad strokes of the new marketplace, trying to cross-reference grower / buyer ideas to determine which features should go into the MVP and which we would implement later. After a few hours of such discussions, our whiteboards looked like madman’s scribbles, but surprisingly everyone was still on the same page. It was slow going, but we were making progress.

The hardest part was trying to figure out how the potplant “configurator” would work. How does one create such a complex arrangement of plants, pots, trolleys, boxes, shelves and ornaments? For this we looked into video games! You know how you can edit armour and other things for RPG game characters by managing their “slots”? It was a combination of that and isomorphic IKEA manual. Cool enough, but after testing we found that it wouldn’t scale beyond a certain point. So we tried another metaphor - playing cards. Plants, pots, ornaments - would each have their own “card” and anyone can easily create combinations by placing as many relevant cards together as they wanted. This metaphor worked and everyone from growers to buyers seemed to understand it.

Implementation

The team was ready to start sketching. Our hypothesis was that the solution would be part web portal, part - API. The web portal would be used to construct and visualize potplant “combinations”. The APIs would be used for spreading information to industry’s core ERPs and for placing orders. Although those ERPs were archaic, there was no way to go around them in the short term.

For simplicity and speed, visually the team selected Google’s Material Design UI. All the react components for it were already available and it enabled us to rapidly launch prototypes without spending too much time on the front-end.

Although the web part of the interface was almost ready, the dev team was still trying to understand how to structure the APIs. To learn, a decision was made to quietly launch with just a few selected growers, with whom the APIs would be configured manually.

As all the eyes of the industry were on us, this quiet launch helped us a lot by removing the unnecessary pressure to “shine” and shifted the mindset towards learning and experimenting.

By this time, the work felt like a regular job. Every day we would have stand-ups (I was the Scrum-Master) and every week we’d do reviews and Retros. My biggest challenge at this point was getting the team to think of features in terms of “experiments” rather than something monumental. This is a stage at which McKinsey usually hands over work back to the client or other companies.

As the McKinsey team was preparing to hand-over the project to a digital agency, by the end, I was spending a lot of time talking to various agencies to make sure that they had the right UX processes, people and expertise to take over this project from us. It was crucial that the work wouldn’t be considered as a “project” but rather as a long-term “collaboration” between the client and the agency, including building digital capabilities within the client. For this reason, the client has chosen Digitas LBi as a partner for future collaboration. Here’s their side of the story.

