Written by Eva Shi, Communications Manager at Chinaccelerator and Producer of the China Startup Pulse Podcast.
A new, brilliant business idea doesn’t come from clueless imagination and perceptions. That’s why you need to test the market and validate your assumptions with a Minimum Viable Product (MVP).
Trouble is, many founders build MVPs in the wrong way by taking three months or even more time to launch over-designed and comprehensive apps or sites that require a lot of engineering hours, most of which end up being thrown away. That’s a waste of time and money.
Most importantly, an MVP is not the initial version of the product you will build upon. You will most likely drop it.
Remember that MVPs are not your products and also not the foundation for your future products.
Nonetheless, you need to start with something that’s not just “minimum” but also “viable” — because moving from problem validation to solution validation requires providing real value to your users.
“Some MVPs will scale but eventually will need to be discarded. That’s just a temporary tool to help you answer not only if you should build it but also define what needs to be built,” says Oscar Ramos, Partner and Managing Director at Chinaccelerator. “Leveraging users’ feedback in a real environment without bias helps with optimization.”
Build, measure, learn
Eric Ries, the author of Lean Startup, describes an MVP as “that version of a new product which allows a team to collect the maximum amount of validated learning about customers with the least effort.”
One of the most important validated learnings for Winzendy Tedja as a founder was to realize that not everything can be done 100% by machine.
He built Yuna & Co, which provides personal fashion recommendations to shoppers in Indonesia. As a tech person, he brought the AI technology to his product from the start. However, only applying the new technology without understanding customers’ needs wasn’t sustainable, even though it got 50,000 app downloads in the first four months.
Tedja’s team reacted fast and restarted everything from the beginning. They started with simple type forms to collect customers’ basic information and needs, including their preferred styles, body information, etc, to set up the customer database. Having human stylists to match products for customers, Yuna & Co. manually dealt with all the orders until they saw it necessary to deploy machine learning technology for building and matching massive customer profiling efficiently.
His major lesson: empathy is needed to understand customers and deliver human support through products and services.
As Tedja explained, “Our mission turns out to be very human-centric and value-driven. The number one in our values is customer-first.”
Anything can be your MVP. Experiment like these founders did
There is no expectation for perfect code or design to build a minimum viable product. Instead, anything can be your MVP, as long as you can validate your assumptions in the simplest and most efficient way.
Coffee Exchange, a Shanghai-based global marketplace that facilitates “direct trade” coffee purchasing, started with a basic WeChat mini-program — a sort of web app that lives inside the popular messaging app. That allows fast development in two or three weeks and easy upgrades every day without bothering users with app updates. It also doesn’t set a high bar for the founders’ technical ability since there are lots of mini-program frameworks that people have already created.
WeChat was also useful as it naturally and easily brings user traffic to a business as one of the largest social networking apps in the world. Every single day, over 300 million people use Wechat mini-programs.
If you are a founder from a non-coding background, should you pay a ton to hire an engineer to build your MVP? Well, not necessarily.
Firstly, there are many social platforms you can use. Giladiskon, a social buying club where 1.5 million Indonesian members get exclusive deals and discounts from restaurants and retailers, started its business with Instagram. By sharing discounts on its Instagram account, Giladiskon has grown its followers to 1.7 million.
“Our followers tag their friends under our posts and people can often see our content on their Explore page. We also receive messages every day saying, “Thank you for the information — I can save some money,” says founder Fandy Santoso.
A simple tool such as Google Forms can also be your MVP material. By simply adding a few questions and avoiding too many elements that require people to write, William Villalobos, the founder of TheFutures.io, got his clients to deliver the design requests easily and quickly. After validating the assumption by the fact that people kept using his MVP, Villalobos later built the technology himself and moved everything to a website. TheFutures.io is a UK-based global design services platform.
Sometimes you even don’t need to test with a large network of people to validate the idea — like David Nicholls did. Rather than building a website for lots of influencers to use, he manually tested the idea by working with just one social media celebrity in Thailand, which brought in over 60,000 customers. That’s how he started Australia-based Flashfomo, which designs and manufactures influencers’ self-branded products.
Pivot fast when you find out it doesn’t work
The failure of an MVP is possible and it can be ascribed to many reasons. However, there is a one-size-fits-all answer: not being customer-centric and fully understanding your target users/customers/clients.
Don’t worry. Talk to them, get the data, and restart. You didn’t expend too much effort to build your first MVP, and it also won’t take you too long to start a new one.
Timothy Yu, creator of Hong Kong-based education startup Snapask, started his own business as a tutor distributing teaching videos on a Facebook page — but he soon realized no-one was paying for that. Luckily, he found out that many students were sending questions through private messages.
And this is quite easy to understand in Asian culture, where many people are not comfortable asking questions publicly. By discovering this, Yu started Snapask, where 3 million students ask questions to their tutors.
Keep up the data-driven experimenting
The key to validate the values you can provide is to run small and rapid tests every day or every week. Having an experimental mindset is a basic requirement for today’s founders and entrepreneurs.
As Giladiskon’s Fandy Santoso adds, “The faster we can validate assumptions, the faster we can learn and improve.”
His startup experiments with everything on Instagram, such as hashtags, timing, repetition, frequency, content, brands, and lots more elements. “For example, we actually go to KFC to take a photo of fried chicken rather than using only a simple banner, which created significantly higher results.” Through experiments, Giladiskon has sailed past a competitor who started earlier with more followers.
“Out of 100 experiments that we do, there will be probably only five to 10 that can really give us great results. What really matters to us is if we are willing to commit to experimenting as a habit consistently,” he tells me.
Even with a complicated business model, you still need to start with simplicity
“If you’re introducing a new business concept and it’s complicated, adoption will be very low,” says Coffee Exchange’s Lewis Harding.
Start with a simple MVP to test only the one riskiest assumption with the least effort, even though the final product you’d like to build will probably be complex.
As a two-sided marketplace, Coffee Exchange had to test both the sell side (the coffee producer) and the buy side (roasters). Instead of adding all the features to validate both sides at the beginning, Harding focused only on the roasters to see if they can gain sales traction from his service.
He initially kept it very simple, only listing and pricing the coffee, so that roasters could easily book orders. Coffee Exchange also “kept the product much like other products that people used to use,” Harding says, because familiarity led to higher adoption.
This article is originally published by TechNode Global on May 25th, 2020