Listening to our early adopter community
How can we spark creativity by helping you to make better sense of ideas you've already collected?
Through countless interviews (we estimate we've had over 392!), our early adopter community made it crystal clear: colour is crucial to the workflow, yet often missing in tools they use. They want—perhaps even need—a way of working with colour that is intuitive, delightful and inspiring. Funded by a Future Fashion Factory grant, we embarked on an R&D project with Professor Stephen Westland and Dr Qianqian Pan of Leeds University. Our vision: make colour organisation, search and navigation better for the creative process.
Our immediate technical objective was to create a colour detection algorithm in Visualist that is accurate. "Accuracy" in this case means getting as close as possible to what a human would select as a colour palette. In particular, we needed to detect colours for the types of images that are most likely saved by our users. We streamlined our dataset to focus on 6 types of imagery:
- Full body
Creating a more "accurate" colour detection tool would give us a foundation on which to build further features to execute our vision. We wanted to enable users to search with precision, help users to 'look' if they don't want to 'search', and let users see the links between everything in their library.
Building with your community isn't just user interviews and prototype testing
Our quest to develop our proprietary colour detection algorithm thus began...
... one that is accurate
... one that mimics the human eye (or brain)
... one that helps our community save time when creating
Our community of colour connoisseurs—designers and stylists, working in fashion, interiors, graphics—all have an eye for colour. We wondered: how could we utilise their collective colour knowledge (and intuition!) to help us to build, and improve, our colour detection algorithm?
The solution, conceived by our product designer, came up serendipitously during a Zoom conversation with our Leeds University collaborators. Enter Colour Swatch Fridays, an ongoing activity in which our community members sent us their colour palettes every week, based on images we've selected. We'd then compare these 'human swatches' against the 'AI swatches', and periodically tweak our algorithm to make it AHAP (as human as possible). Here's the catch: there was no fancy data collection site. Everything happened within Instagram.
Here’s our guide on finding ways to involve your community whilst building your product.
The dos and don'ts of involving your users
Asking your community for ongoing input isn't as daunting as you might think. If your early adopters know why something is fundamentally important to your product, they'll help. If the task appeals to their natural interests they'll be more likely to do it.
1. Meet them where they are
Where do your early adopters hang out? Are they on Slack, Reddit, Twitter or Instagram? For our community, Instagram was (is!) their natural habitat. Many are creative directors, interior designers, and personal stylists who are on Instagram to both enhance their personal branding and look for inspiration. So, we slid into their DMs. Every Friday for 13 weeks, we would send them a new message with their latest colour swatching tasks.
Think about where your users are and how you can meet them there. Choose the path of least resistance, always. Just because Slack is the "go-to" for many SaaS early communities doesn't mean it's the right place for you.
Tip: it helps to have already nurtured a 1:1 relationship. Our community knew us as people, not just a company—which brings us to the next point.
2. Introduce intimacy through voice
It's not just what you say, but how you say it. What that meant: it was important to be clear and concise, but it was even more important that our users knew whom they were doing this with. We were intentional early on about getting our community to feel like they knew us individually. That meant regular phone calls and DMs, which sparked familiarity, camaraderie, and also a sense that they were let in on what was happening behind the scenes.
We could have sent an email with clear instructions on what steps to take—but that wasn't us. For our instructions, I made a screen recording of myself demonstrating the colour swatch activity in Instagram whilst explaining what I was doing in a voiceover (even if I absolutely cringed when hearing it back). Colour Swatch Fridays were taking place when much of the world was in lockdown and introducing voice allowed me to not only make the instructions easy to understand, but also to feel more connected with our beta users. As a bonus, sending it as a voice note meant that we had to very concisely convey our instructions. This forces you to really distil down what the 'job to be done' is for the user, and not overwhelm them with the minutiae.
If you’re a brand that's building in public, don’t miss out on opportunities, big or small, that can strengthen your brand's personality or add a new dimension to who you are and what you're building. Find ways to communicate that reinforce your brand voice, be it product updates (check out Head of Instagram, Adam Mosseri), behind-the-scenes content or an activity for your beta users.
3. Be consistent
Consistency creates rhythm. Rhythm establishes discipline, for us, and anticipation, by our users.
We sent our task out on Fridays and collected the results on Mondays. This worked for us because the responses could trickle in over the weekend, and users could colour swatch at whenever was convenient to them. We passed the results on to our Future Fashion Factory partners on Monday mornings and discussed the results on Tuesdays. Pick a schedule and stick to it. Let your community know what to expect. Give it a name like we did as it holds you accountable to stick to your schedule. Plus, Colour Swatch Fridays had a better ring to it than 'Weekly Colour Swatching'.
Consistency builds trust and shows respect for your community members' time.
4. ... but still provide novelty
We didn't send the same style of photo each week. We had 6 types of imagery we were testing, and so we mixed things up each time. Our community is one of aesthetes. Why not use the opportunity to delight them too?
We challenged them. Instead of just creating a colour palette, how about ranking them in terms of importance? We also intentionally chose some photos that we thought might be tricky to create a colour palette of five colours.
We surprised our community by using photos from their community in the activity.
Whilst consistency creates rhythm, it can also breed boredom or nonchalance. When an activity spans a longer time frame, think about how to maintain interest with little surprises—or challenges—every now and then.
5. Reward your community
Our early adopters were investing their time into our product. We needed to show our appreciation.
The major reward for our early adopters was in the future: a colour detection algorithm they could use seamlessly in their workflows. We weren't able to provide this straight away, as it took us six months to develop, test, iterate and launch. So, we decided to reward them by giving them unique merch designed by our team. We sent bundles of socks, pins and stickers that appealed to their creative identity.
6. Keep them in the loop
When building in and with the public, you want to be able to share the insights you're gathering in a way that's informative and fun.
We brainstormed ways to regularly share the results with our early adopters. We created mini-quizzes on Instagram stories and got them to guess which colour palettes were created by our AI and which ones were created by them (a Turing test, in technical parlance).
Think about ways you can show the results of your community's efforts. Share it with them at regular intervals. You may be surprised by what further dialogue it creates.
Tip: This is when good organisation helps. We stored all the results in our Google Drive, so we could easily pull the results for sharing.
Designing community activities is no different to building a product. It’s important to test your assumptions and throw out anything that doesn't work.
We experimented with the time of day we sent out our message, the number of swatches we were getting our users to do, and even how we worded the instructions. The latter wasn't as crucial once our community got the hang of the task, but was particularly helpful for the new cohorts of 'colour swatchers' we roped in each week. In the beginning, our instructions were simple (or so we thought): create a colour palette from an image. We then tested a different set of instructions with another group: pick the dominant colours.
Continuously assess your process to find the most efficient way to build alongside your community, and remember not to get too attached to a particular way of doing things.
8. Keep it warm
Don’t treat building with your community activities as stand-alone events. Instead, think of it as part of a much bigger tapestry, interwoven with smaller and shorter types of engagement as well.
The nice thing about a weekly activity is that you’re slowly getting to know each member of the community. Each activity shared comes with a quick catch up message. You soon learn which weeks they were feeling productive, and which weeks simply sucked. You learnt which gigs they were going to and what their second or third jobs were like. An ongoing activity is a great jumping off-platform, so don't let these connections go cold. Whatever activity you design, remember to continue to check in and keep them onside.
9. Make it easy (and harmless) to say no
This sounds counter-intuitive, but quickly identify who's not interested (or not available) and let them opt out.
As the activity ran 13 weeks, we also collected data that could help us to see trends in one user's colour swatching activity across the weeks. But we also made it clear internally that this was a bonus, not a requirement. This meant that our fundamental research target of comparing human swatches vs AI swatches was never in jeopardy, even if we had dropouts along the way, as long as maintained our baseline target of the number of human swatches.
When involving the community in building, always be prepared that commitment levels can vary across the length of the experiment. It's thus important to mitigate any impact on your development objectives.
From a community perspective, figure out why someone's not interested, and weigh up the efficacy of re-engaging them in another activity. Was it the scope of the task or the timing? Or were they simply just not that into you?
10. Find ways for it to live on
What did your community like about the activity or process? Is there a thread that you can unravel that shows you new ways to keep them onboard with product building? Three things worked fundamentally well for us:
- Quick: Thanks to Instagram, users could finish the task in minutes, not hours.
- Fun: There was a light, gamified element to it.
- Stimulating: But only just. It was thought-provoking, but not mentally taxing.
We parlayed these philosophies into our Instagram stories: polls and quizzes that appeal to the creative, that have them pause—just for a moment—to think about how they approach visuals.
What we'd do differently next time
Here are the things we'd do differently for our next community-building product activity:
- Aim bigger. It was our first time involving the community, and so we started small: we only engaged with a handpicked group whom we thought were most likely to participate, and we tempered the request to a minimal amount. On hindsight, there was a lot of excitement from our users in being involved in Colour Swatch Fridays. We could have involved much more users, and (maybe!) asked them to do more watches each Friday.
- Bring it to Reddit. We would have looked to Reddit to find our kindred community of colour lovers and introduced them to our mission. Who knows if we could have made a bigger impact, with more data collected and more early adopters? But we live and learn. Check out our newly launched subreddit here where we continue exploring the world of colour.
This was a surprisingly rewarding experiment for us. We started out with the objective of gaining data to build our colour detection algorithm. And we ended the process with a highly-engaged community who not only enjoy the rewards of what was built, but also experienced the inside track of co-building with us.
We received so much buy-in from our community during the 13 weeks.
When we told them that Colour Swatch Fridays were ending, it was a bittersweet moment.
But we were also delighted when our community began experiencing the auto-generated colour palettes on Visualist:
Here’s how you can start brainstorming ways to involve your community in unexpected avenues:
- What are your early adopters made up of: are they developers, designers, writers, photographers or fitness fanatics?
- What activities appeal to their natural interests?
- What data would enrich your product, making it even more efficient?
Once you’ve designed an activity, the rest is simple: meet them where they are, remove all barriers, reward them generously and treat it like a series of experiments to see what works. Good luck!
To learn more about our Future Fashion Factory project, read Cherie's interview here.