Where We Started
The founders of Rachis Wine Assessment have a lifetime of wine experience behind them. They also have a yearning to change the industry with the reach of apps and the web into lifestyle pursuits. If they could reach both experts and novices in the same experience, they could bridge traditionally distant wine lovers and gain great new insights on wine tastes. Where other products avoid the traditional aspects of wine tasting, we saw how leading people deeper into their sensory experience could create not only better memories, but also a better understanding of one’s own tastes.
What We Did
Working with the founders’ experience, we translated the market need and initial vision into an HTML-based app prototype to explore critical questions. The most important question was whether we could capture the depth of traditional wine evaluation in a multi-platform digital experience that was accessible and personal?
In the product build, we designed and engineered new key features: a unique visual rating system, personal taste profiles, recommendations, a unique keyword browsing system, automatic composition of short written reviews, and a sensory-rich interface never before seen in wine tasting. With ongoing testing with over 100 wine tasters, we developed for iPhone and iPad, tied together through robust web services that enable tasters to record tasting, get recommendations, and share with friends wherever they go.
Our concept and prototype work not only clarified the initial idea, but helped secure nearly $1 million in funding to develop the whole product. After several months of intense development, Quini launched in 2013 and has seen a steady accumulation and retention of customers. As the product meets its audience, its voice and influence grow, starting a journey to open the world of wine to a new generation of tasters.
Concept & Strategy
Validation, ideation, and roadmap planning for new ideas.
Creative design and production for apps & websites.
- iOS 7
- Amazon S3
- Amazon SES
- New Relic
- Skeleton CSS
Coming up with an entirely new way to recommend wines was particularly challenging. Recommendation engines, of course, are grown more than they’re built. They need data, reflection and time to get right. Just ask Netflix, or Amazon, or anyone else who has spent years getting it right.
But we needed something to start, and took a two-pronged approach with the top-rated wines from the whole community, and personalized recommendations based on what people have already tasted with Quini. The former uses a Bayesian algorithm in conjunction with a couple of extra rules to safeguard against review-spamming. The latter is much more elaborate, using a vector-space model that describes a person’s tasting profile with the same points that we use to profile wines. This allows us to match-make people and wines in a way that can produce results in early days, and still be grown and tuned over time.
We learned a lot from our biggest project yet, but some lessons stand out and are worth sharing.
It’s a lot easier to get going with surface-level prototyping in iOS than you might expect. Going offroading from stock interface elements is harder, but you can make really visible and testable progress in just a couple days.
But, it takes real time and effort to clean up prototype code and to hook everything in. The speed of development on the interface layer is only apparent, and iOS can still be a very involved space to make your ideas work.
Oh and one other thing: colour multiply and Gaussian blurs are really hard to do on iOS without dragging down performance on an iPad 3. Getting these right took way more time than expected.
Little Things Will Surprise You
There are many stories about how almost incidental features can win a customer’s heart. In mid-summer we were playing with ways to encourage some kind of written opinion from people in what is a predominantly tap and drag experience. In addition to full blown notes, we introduced a one-line review to encourage a tweet-length opinion. Arian and E Brake were experimenting with an auto-fill mechanism out of curiosity, as most of our review points involve some kind of qualitative language.
In testing, auto-fill became an instant hit. People loved seeing their earlier actions translated into a natural-sounding mini-review. We spent some time cleaning up the templates and smoothing out the language, and hope to expand on this feature in the future. It’s not something that will impress typing teachers, but by taking away one more point of resistence we hope to win reviews from many more people.
Informal Testing Opens People Up
Our testing sessions were disguised as wine tasting sessions that happened to involve an app. At least once a week for two and a half months straight, we met new and interested people and got great direct and indirect feedback on every aspect of the product. We were less formal than usual in our testing sessions, but what we traded in rigour we gained with insight into how people think, talk about, and enjoy wine.
We especially tuned into the anxieties that people have around wine, namely about looking less than cultured because they lack the words and experience to demonstrate taste. Knowing that, we put a lot of care into making wine reviews look informed and confident, even if the person behind doing the review doesn’t always feel that way. Our hope is that over time, using the app will not only help less experienced tasters notice more in their wine, but also feel more confident in how they talk about it.
Lastly, we learned that everyone - absolutely everyone - has some kind of special memory involving wine. A first trip abroad, a wedding, recollections of bonding with parents; the stories were endless. We started asking this question at almost every tasting, and what we got back kept us in tune with what we were building for: a love of wine.
Artistic Touches Make All the Difference
Even with the most rudimentary code, we would strive to add beauty to anything that people outside our team would see. A survey that we added to test versions of the software helped us collected tester feedback, but also gave us a chance to include beautiful illustrations, like this one done by Bruce Alcock. These helped set the tone of the project and kept us in touch with the sensibility we were going for.
We can do this for you
Maybe this sparks some ideas of your own. Learning about your options and next best steps is as easy as dropping us a line: