Discover more from Data Operations
The coach will see you now
Elite athletes use coaching to maximize their performance. Shouldn't software provide a similar always-on feature to help you get better? Read: "Everything Starts Out Looking Like a Toy" #164
Hi, I’m Greg 👋! I write weekly product essays, including system “handshakes”, the expectations for workflow, and the jobs to be done for data. What is Data Operations? is a post that grew into Data & Ops, a team to help you with product, data, and operations.
This week’s toy: Janelle Shane writes about AI-generated memes that don’t exist. After reading about these examples, “linoleum harvest season” sounds quite believable. Meme culture works precisely because of the way we pattern-match against meme tokens, so this is a clever way to troll. Edition 164 of this newsletter is here - it’s September 25, 2023.
If you have a comment or are interested in sponsoring, hit reply.
Data Operations (“Everything Starts Out Looking Like a Toy”) is a reader-supported publication. Please consider subscribing.
The Big Idea
A short long-form essay about data things
⚙️ The coach will see you now
When I was younger I wanted to be a professional baseball player. Professional baseball announcer was probably a more realistic goal given my level of baseball skill. Clearly having talent is an important first step, but watching MLB games today reminds me how much coaching is available and expected for players at the highest levels.
The difference between average and elite players is often mental.
We would all like a coach like Janne Mortenssen, but we don’t have access to her in person. What if software could bridge the gap for coaching all kinds of skills, especially the ones you need to skill up quickly?
Coaching as an overlay is a key feature in hardware and software today. Your phone tells you you need to walk more today. Your earbuds inform you of an incoming text message or upcoming meeting. The goal? Get more contextual information that helps you respond in space and time to what’s next.
“Coaching as a service” is an underappreciated part of software to uncover the capabilities of a service at the moment that you need it. When it works, it feels like magic. The difference between average and elite software users can be bridged with coaching.
We are all becoming cyborgs. Just like the audio cue from your Airpods that lets you know that there is an upcoming meeting and you need to switch focus 10 minutes from now, the best software tells you what’s next.
The “motivation layer” of software makes the behavioral difference between people who log in to software and look like active users to those who are using software and are able to become engaged users. There is simply too much to remember in short-term memory, so we depend on other systems to get us there.
What does this mean in practice? It means that the best software teams are going to use the same kinds of nudges (probably powered by machine learning) to recommend the next steps when doing a workflow. Why? Because it’s a lot easier to build a capability for nudging than to build a perfect piece of software.
Building just-in-time competence
You are not going to be able to build a feature that solves every use case for every persona. If you focus well, you’ll build a feature that meets a critical use case for a key persona. But helping users to get to the next step of a flow (or the next best action) might be an even more powerful step than building a perfect feature.
I believe that building just-in-time competence is a critical goal in building software today because it’s a faster way to fulfill user needs than creating a fully realized feature for every persona. Coaching (if you don’t like the word, think about the feeling you get when you receive expert help) unblocks the user without involving another resource.
Learn when to coach the user.
If you think of the coaching motion, it’s a process that:
objectively watches what you are doing
suggests an improvement (hopefully, one you are able to make)
if you can’t make that improvement, helps you understand the knowledge needed to get there and how to build your skill
checks for understanding and cadence
Great coaches help you to get better while understanding that you might need some scaffolding to get to the next level. They also let you know what’s going on and catch you in the moment of doing the process before you move on to the next step.
What could this look like in everyday usage?
When designing a persistent “coach” for software, I’d start by thinking about the things that it doesn’t do.
The coach probably doesn’t have a dedicated icon to invoke it.
The coach probably doesn’t have a tab or a persistent area of the software.
The coach does know about more than one tool you are using and has the ability to ask that other software for information
But once you invoke it, that coach is always listening.
A great coach is available when you need help, makes you better, and has an encyclopedic knowledge of the problem and the solution. That coach also is effective at suggesting ways you can apply that knowledge in practice and to help you measure whether you did that correctly.
Your coach might be an AI Copilot
Think about the last time you asked a co-worker for help. It probably was a quick DM or a mention in a shared Slack channel.
What would this look like if you had an always-on coach? You wouldn’t need to use a private channel (though this could be possible), and you could ask for assistance by writing: @coach, help me out with my logic here. Does this make sense?
Using a GPT bot for this purpose right now feels a little weird. It’s great for answering factual questions or for repeating information that’s been published elsewhere. These bots (ChatGPT among them) are great at parroting information but not as good at following more complicated logic.
If you need a specific playbook for your help because of your job function, you need to give these bots specific instructions and even then you run into the problem of being careful about sharing proprietary or private information.
Providing context through integration
What would a smarter bot do?
In addition to listening to your question and offering features like “summarize this”, “help me with my logic”, and “what should I do next”, a smarter coaching bot would also know:
based on your privileges in the organization, what systems can you access to find information?
based on your job role, what are typical playbooks you might use and what types of information do you need to review?
if you find a novel way to solve a problem, how would you store this and share it with other people in the organization as needed?
I’d love to see a bot like this in my Slack instance. It would be really useful at least several times a day. And it would be more valuable if it spanned applications, implying a framework for talking to other apps, not just living in a walled garden.
Who’s building this today?
What’s the takeaway? The gap between active users and engaging users is coaching them on what they need to do next. This ability to offer just-in-time information is a great case for an always-on bot to be available in your existing conversation tools. The bot needs to respect existing security, compliance, and privacy needs and is the key to getting user success before building features and tech debt that might be unnecessary.
Links for Reading and Sharing
These are links that caught my 👀
1/ Gmail is using you to train AI - It shouldn’t surprise anyone that Google is using Gmail (and your inputs) to train AI. Thought experiment: would they consider creating a tier of Workplace that excludes AI training for an additional fee?
2/ An invisible feature in the New iPhone - The Ultra Wideband Chip is an underappreciated feature of Apple’s new iPhone is the updated Ultra Wideband Chip. It makes it easier to calculate the distance from your phone to anything else. Why is this neat? It opens up a whole new interface: “drawing” with your phone in space.
3/ Shrinkflation in action - Clever site that tracks price changes for CPG products in the UK. If you could get the price data, this could be a really interesting service to compare everyday purchases. (Imagine an API for pricing data, and how you might use it.)
What to do next
Hit reply if you’ve got links to share, data stories, or want to say hello.
The next big thing always starts out being dismissed as a “toy.” - Chris Dixon