How to avoid the trap of MVP by setting a long-term vision
Too often we as designers are forced to accept the compromise of releasing a new product feature that has barely made it past the minimum viable product stage. While it’s a great achievement to get to a place where your MVP is successful, an MVP by its very definition will be basic, lack delight, and will likely be built on code that’s not finished to a standard suitable for long-term use.
In this article, I’m going to talk through how we avoided stopping at an MVP by setting a long term vision that would motivate and inspire our team. Then I’ll talk a little about our approach in solving the customer problem: ‘Helping customers find the food they want, more quickly and easily’ and enlighten you with key research findings we made along the way.
Why dish search?
We know that choice is important to our customers, but as we increase the breadth of restaurants and dishes on Just Eat, we may also be making it harder for people to find what they are looking for.
At the heart of the Just Eat experience is our ability to provide a hungry customer with the food they most feel like eating, as effortlessly as possible. So we have given customers the ability to search for a specific dish — without visiting every menu to find the perfect Souvlaki.
Let’s take a step back…
The search feature has always been present within the Just Eat products, it just had some limitations. It allowed customers to search only for cuisines or restaurants, and their spelling had to be accurate otherwise their search would not produce any results. For several years this may have been fine, but users’ expectations were changing. More and more Just Eat users were asking for control over the things they could find within the thousands of restaurants and options that are available on our products.
“Make searching for a certain dish easier — I don’t want to be scrolling through every option on Just Eat” — Just Eat customer
It was obvious we needed to give them more control. We needed to go beyond just surfacing dishes that directly matched a search query. By understanding the defining attributes of every product on our platform, such as how spicy it is or whether it contains chicken, we could exceed users’ expectations by making it more intelligent.
Phase 1, Setting a vision for the team
“If you are working on something exciting that you really care about, you don’t have to be pushed. The vision pulls you,” — Steve Jobs
Setting a vision for your product is useless if the people involved in creating it have different interpretations or lack motivation. A great way to create a shared understanding is through running a collaborative workshop. I’ll touch on this shortly but first, let’s talk about minimum viable product tests and a common pitfall.
Marty Cagan, author of ‘Inspired: How to Create Tech Products Customers Love’ defines a minimum viable product as the smallest possible product that has three critical characteristics: people choose to use it or buy it; people can figure out how to use it, and we can deliver it when we need it with the resources available — also known as valuable, usable and feasible. Eric Rees, creator of the lean startup methodology also popularised the term MVP test, which is basically the same definition, but instead creating the smallest possible experiment to test a specific hypothesis. It’s the first step towards achieving your long-term vision. Dish search was our MVP test.
The problem is that large tech companies often never move beyond that MVP test, yes resource issues could be one reason for not continuing but most often it’s because the experiment was successful, product managers and engineering teams think the job is done and they just move onto the next job.
As a designer, it’s our responsibility to change that thinking by always looking forward to the future, setting a vision of what could be. In 12–24 months from now, what will our users be saying to others about the experience and how it’s solving their problem?
Running a collaborative workshop
The first step was to get the team together and work on a vision statement that could be hung on the wall to be seen at all times by the team and others around them.
“Just Eat knows everything about my tastes. The experience is so simple I can find the specific things that I crave fast by what’s important to me and build a basket without ever visiting a menu” — Our teams vision statement
A vision statement should be short, easy to communicate and understand. Setting such a clear statement creates a shared understanding amongst the team allowing everyone to add value creating a sense of empowerment.
Creating a vision statement also allows you to think about the things that could stop you from reaching the vision. In our case, this could be tech limitations, restaurant supply issues, business objectives, etc. Running this activity with the team also allows key stakeholders to take note and set plans earlier in the process limiting the chances of derailing it further down the line.
Once the vision is set the fun part begins: gathering ideas of how the feature would look and work. I like to follow the Google Ventures version for this activity. I like how after the team has sketched and annotated their ideas they then hang them up on the walls around them and silently take votes on which they feel has the best potential. Doing this allows everyone to have a say and avoid the ’loudest person in the room’ syndrome.
Ideas gathered, it’s time to turn them into higher fidelity prototypes and test them with Just Eat users to make sure the flow makes sense and is usable.
Understanding users’ expectations
At the start of every project we begin talking to our users about it, and continue to do so throughout the design process. We gain feedback that tells us if something’s working or needs improvement, which is all part of being customer-centred and baking research into the Just Eat DNA. The process gives us vital feedback around users’ expectations and feedback to the data team.
An example of this was that after a user performed a search for a specific dish, let’s say chicken biryani, we opted to recommend to them another dish that was very similar to their initial search. Our assumption was that giving users another option that was very similar to their search term would give them a greater appetite for food discovery. Feedback from users suggested that this did in fact give them a way to discover new dishes. But what they were really looking for were suggestions that complemented their initial search. In the case of a chicken biryani, this might be garlic rice or naan bread.
“I like the idea but would prefer to see things that complement my search, maybe a side or starter” — Just Eat customer
Key research insights included:
- Tempting users to try something new — Seeing dishes that were similar to the one a user wanted would allow them to try something new or even buy both? e.g. something vegetarian to go with a King prawn dish.
- Compatibility — Seeing complementary items that pair really well with a certain dish adds more value for users.
- Keeping an open mind — Alternative suggestions help keep users’ minds open to what else is available on the menu.
- Avoid being repetitive with meals — Suggestions could be useful in helping users to vary their diet.
Planning our vision through story mapping
We needed to be sure our design was feasible and sensibly approach an MVP test. We ran a workshop with the whole team of data engineers, product managers, UX designers and various stakeholders to determine what the MVP test would look like. Using story mapping we agreed what was going into the first phase and what was in a backlog for further iterations.
Creating a sense of realism for testing
Before talking to our users we needed to decide how we were to prototype the flow. We needed to make this as real an experience as possible due to the nature of results provided by the algorithm. The engineering team decided to quickly build it using a Chrome extension, which was the first time this method had been used so there was lots of learning to take away from it. Using the extension meant we could make and see changes very quickly, reducing development overhead whilst we were determining the value of the idea. We all agreed that for the MVP test this sounded like a sensible approach with the potential for big benefits.
Finding value at scale
After the qualitative phase of our research with Just Eat users we were in a pretty good position. Test participants seemed to find the usability of the feature straightforward and we took away some key insights to feed the algorithm. Now it was time to test this at scale.
Part of our process with every new feature at Just Eat is to run A/B experiments, putting a variant of the proposed new design against what’s currently live on the platform. So how did the design perform? Well, results from the experiment suggested our new intelligent dish search enabled users to find the taste they craved faster than ever. 🥳
It also provided our team with vital data knowledge. Knowing what users search for gives us the ability to provide incredibly more accurate results and recommendations.
Usually, for some teams this would be as far as they would go. “We’ve proved this feature is viable we have the data to prove it, let’s just move onto the next thing we can always come back and visit it at a later date”. Yea right!
We wanted to move past an MVP test and see our new feature bloom. We had set our vision statement — now we wanted to achieve it, so we set out to continue down the yellow brick road.
Phase 2, Introducing a predictive search
During our idea creation in the first phase, one feature enhancement stood out: the ability to predict what users are searching for. The team felt this was a perfect enhancement to our feature as it would add value and was feasible. Now we had to make it usable.
Designing what seems to be a simple feature resulted in a lot of key insights for us:
- Get to the things you want quicker — The predictive text appearing in the search results made the process feel quicker for users, as they didn’t have to type in the full name of the dish. Particularly if they didn’t remember the full name.
- A touch of inspiration — It was seen as a bit of inspiration “It’s inspiring to see what else is available as you type”.
- A sense of Inclusivity — It made it easier for users who are less tech-savvy as they didn’t need to type as much, as well as users who might suffer from disabilities or impairments (e.g. dyslexia). The way this feature guides users is seen to be very useful.
- A spelling aid — It was also useful for users when trying to spell certain dish types e.g. “spaggetti bolagaise” or “spaghetti bolognese”… 🤔
We even established different scenarios the feature would be useful for:
- Mood — If users were in the mood for something specific (e.g. Souvlaki)
- Time of day — If users wanted something specific at different times of the day (e.g. breakfast, lunch, dinner)
- Findability — If users had a dish in mind but didn’t know which restaurants offered it
- Price comparison — If users wanted to find the ‘cheapest’ of a particular dish in their area
Building on existing key principles
Along with aligning the design to our Just Eat design system and building on the key insights from our research sessions, we decided to also take on board a few key existing principles found across several great articles on designing for predictive search:
- Baymard Institute, 13 Design Patterns for Autocomplete Suggestions (27% Get it Wrong)
- UX Magazine — Designing Search: As-You-Type Suggestions
- Nielsen Norman group Site Search Suggestions
These included:
- The use of scroll bars should never be an option. Keeping results to a limited set avoided this issue and also reduced choice paralysis. We followed cognitive psychologist George A. Miller’s research for this: The Magical Number Seven, Plus or Minus Two
- Highlighting and bolding the differences in the results that are returned. This allows users to easily pick out suggestions that best match their search query. E.g. Chicken Korma.
- Reducing visual noise while searching. Adding things like suggestions while performing a search can distract users. Keeping this simple helps users maintain focus.
- Use labelling to ease scannability. Organising labels by type adds more context and clear expectations e.g. Chicken Korma in dishes.
- Allow for readability. Using appropriately sized fonts, touch areas and spacing allows for a better overall experience.
The impact of predictive search
After running further experiments we learnt that our new predictive enhancement added even more value for our users. Now we guide users to the things they want even quicker, adding a touch of inspiration when they search and not having to worry about spelling a certain dish type 🙌
Onwards and upwards
The vision the team set at the very beginning of the project helped to maintain our focus throughout all the work involved. This vision allowed us to deliver a successful feature that solved a clear customer problem: ‘Helping customers find the food they want, more quickly and easily’ and avoid stopping at an MVP test.
As UX designers, we have a responsibility to always be the voice of the customer, to always improve the features within our products, and not stopping at the first step towards achieving your long-term vision. There can always be improvements made and these improvements will make a difference in how a customer uses your product.
We continue to set new visions, learning through user feedback, iterating to improve the experience and now and again adding a few easter eggs along the way…
Easter eggs I hear you say? The next time you order takeaway on Just Eat try searching for it with an emoji 🐔🍕🍔
Originally published at https://tech.just-eat.com.
