Startup explorations #19 / Experiment No. 2: Kickoff

I’m kicking off the new month by repeating the same experiment as last time with a different set of users (more indie developers and founders). The objectives are to 1) see if we can achieve similar results as the first group, 2) get to know our prospective customers more, and 3) get a feedback loop going and hopefully generate continued interest in the project.

In addition, I’m paying special attention to a few more things this time around:

  1. More adaptability and responsiveness in each day’s challenges, taking into an account what a particular user has done the day prior and adjusting accordingly, hopefully making the experience much more personalized.
  2. Community instead of following: this means emphasizing thoughtful engagement over merely trying to get more followers.
  3. Defining one or more customer personas.
  4. Noting with as much precision as possible how I’m finding conversations to recommend (see my previous post about how I’m making recommendations).

Regarding the fourth point: recommending conversations is by far the hardest and most time-consuming part of the exercise on my end because I have to look at each user’s profile everyday to handpick good opportunities for engagement. If this second experiment yields good results, we’d be at a point where starts to make sense to think about ways to scale up this operation little by little. I think it could be a very fun engineering challenge down the road.

Startup explorations #17 / Wrapping up the first 7-day audience building challenge

I’ve gotten a lot out of doing this experiment over the past week. To reiterate: I set up a Slack workspace and over 7 days, told a few creators what to do each day to encourage them to keep up their Twitter engagement. I recommended accounts to follow and conversations to respond to.

I’m not a Twitter expert by any stretch of the imagination; I just looked at their profiles individually, got a quick sense of what they were about, and gave them daily targets based on what they were already doing or what they were obviously not doing.

Anyway, doing this experiment as a way to test my product idea on a very small scale allowed me to confirm a few of my assumptions early on:

  • That creators indeed go on Twitter specifically for audience/community building.
  • Creators who don’t have high engagement yet but are motivated enough would appreciate being told just what to do everyday.
  • The need for habit-building, target-setting, and consistency over a period of time.

To the third point above, the results speak for themselves. In just seven days, one of my “users,” an app developer, reported significant improvement in tweet impressions, profile visits, mentions, and follows. I was happy to know as well that by the final challenge, he had built enough momentum to write a post on Reddit about the app he’s building.

I noticed similar improvements in two other users, though only about as much as you can expect from the very short timeframe. I plan to continue following their accounts over the next several days to see how they’re doing. A fourth user dropped out of the experiment after the third day, as he decided he wanted to take a more organic approach to Twitter—fair enough, and all part of the process.

Still, to me there’s enough of a green light to try the experiment again with a few tweaks and specific areas that I might want to play around with. In no particular order:

1. Recommending accounts to follow
I learned early on in the experiment that recommending accounts to follow didn’t seem to make that much of an impact, as my users would pretty much just follow whoever they want anyway. Besides, one of my users was more deliberate than usual about curating his timeline, with which I sympathize, and was very selective about his follows. Still, it might be useful to suggest them occasionally when a user doesn’t seem to be growing their network, or getting much engagement from the usual options.

2. Recommending conversations
Corollary to the above, there seems to be a lot more value in curated lists conversations that a particular user might find easy to respond to, based on their interests, recent activity, and so on. Usually these conversations have some kind of prompt or invitation to respond, and are often made by larger accounts. I’ve found this to be an effective way to get exposure, especially with consistency, as lots of other people looking to make connections hang around these big accounts.

Similarly, I’ve also found it easy to get people to just respond to thoughtful questions in general, even from smaller accounts. This doesn’t result in a lot of exposure, but does increase the likelihood of a connection that can become more meaningful over time.

Naturally, I have a long way to go to streamlining this whole process so that the recommendations are more likely to be good fits than not—one can only do so much manually!

3. Progressive difficulty and retaining users
A sharp and linear progression makes sense over a short period of time like 7 days, but what if we were to extend this over months? A whole year? It would probably make more sense for it to have ebbs and flows. In the future I’m thinking of giving the option to opt out on certain days.

4. What if a user misses a target?
Initially, when this happened, I just restarted with the previous day’s challenge, with a new set of recommendations. Again, this makes sense over a 7-day timeline, but if this is built out and extended, there would have to be a much greater incentive for a user not to miss any daily target.

5. Recommendations or strict orders?
I’m continuing to think of ways to make this experiment more challenging and engaging. Perhaps a fun challenge might be to make users, either on all or some days, follow my directions strictly in order to meet a day’s goals—for example, follow so-and-so users, respond to this or that conversation specifically. But I don’t believe this would be necessary if the recommendations were always as good fits as possible.

6. Finding the right users
I like to think of this whole thing as an audition to find the right users who will benefit the most from what I’m offering, rather than a race to make everybody happy. I noticed the most positive results from my users the more open they were to following my recommendations, and of course, the more they had to offer in the first place—for example, having one main, visible project that they’re working on or trying to promote. As I work with more people I hope to get a much more refined sense of our ideal customers.

7. Will anybody pay for this?
Pretty self-explanatory.

Naturally, all of this is just the tip of the iceberg, and I may have many more thoughts in the coming days. We have a very long way to go from here to actually building anything. But my hope is that by sustaining this experiment and allowing it to evolve, we create a feedback loop that will serve us in the long run and lead us toward a product that will actually achieve our stated goals.

Startup explorations #16 / The 7-day audience building challenge: How I’m making recommendations

In my last post I wrote about testing this ever-evolving idea I’ve been writing about here by actually running a few people through a weeklong Twitter challenge. I’m running all of it in a Slack workspace, which is all running smoothly so far with four participants of various stripes. Naturally, promising actual “audience-building” is a tall order, so we focus instead on things we can actually control: I like to think of it as helping people build a habit of regular community engagement. It is meant for creator types who need an audience but would rather not have to think about it.

The heart of the project, if it turns out to be viable, is a personalized recommendation system that gives you a daily goal or challenge, along with recommending other users and conversations that a user might be into. I’m doing this manually right now by looking at each profile in advance and trying to get a sense of what each one is about as quickly as I can. I look at who they’ve recently followed, what they’ve recently responded to, etc., and adjust my recommendations accordingly. For example, if you already tweet regularly but don’t reply much, I’ll handpick a few ongoing conversations and make you respond to one or another. Otherwise I stick to a default plan.

I’m trying to recommend profiles that have been active recently, are engaging with other users, and especially those who ask questions in public or invite feedback. I’m also trying to get a good mix of profiles that have roughly a similar follower count, and some larger accounts as well who engage with smaller accounts (this is not all that common). To make things more interesting, I also include a little randomness. Obviously, as a human, I can’t do all of the above very efficiently, but there might be something in a system that can respond to what you’re doing and make very good suggestions.

Startup explorations #15 / The 7-day audience building challenge

This week we’re shaking things up a bit by actually implementing the project idea on a small, small scale as a way to validate it. Obviously nothing at all has been built yet—so I’ll be doing everything manually. I will personally walk a handful of real, live users through meeting daily Twitter goals over seven days. The goal is to help these users get a habit started of regularly engaging with people on Twitter, in the hope of growing their own personal communities.

Specifically, here’s what I’ll do for each user:

  1. Give them a daily goal to meet.
  2. Recommend people for them to follow.
  3. Recommend conversations for them to engage in.
  4. Give ideas about what to post.
  5. Send a reminder when they forget or miss a goal.
  6. Track their progress daily.

In addition to the above, I’ll be analyzing their profiles as carefully as I can to guide my recommendations. The main challenge here for me will be keeping track of everything (and in different time zones), and meticulously crafting a decent user experience, but I think by the end of seven days we’ll get a good idea about the validity of the project and how to move forward.

I’ve created a very simple form to collect basic user info, with a link to a Slack workspace from which I’ll be running everything. I’ve no idea what to expect, but this is an exciting next step in the project where I get to actually test the ideas I’ve been brewing and see how they fare against reality. Once again I refer to Derek Sivers for inspiration: you can get started right now by just helping one or a handful of people.