The term still grosses me out, but alternatives have failed to catch on 😅
When prototyping new features or feature changes, dogfooding gives us a vocal internal audience to learn from.
This usually starts as a small set of opt-in users. If an idea has legs, we proceed to a "Friends Of" release – dozens of high-context colleagues for feedback
At this stage, teams are socializing and building conviction in the feature with internal stakeholders and execs. If we are able to get enough confidence at this stage, we proceed to a "Tiny Speck" release (a wall-to-wall internal rollout named after our original company)
We post these into our #released-internal channel, along with a link to a #feedback-feature channel where internal feedback is routed.
Feedback at this stage is typically very high volume! Our colleagues aren't shy about expressing their love, loathing, nitpicks + edge cases
As features are refined, we typically rollout to a customer pilot group for feedback, which is incorporated before final General Availability release
In the meantime our internal users are on to the next set of upcoming features. Living in the future and building what's missing!
Often, these internal features are incubated for extended periods. Slack Huddles and Clips are two examples of recent additions to the product that went through heavy revisions based on internal feedback and product-market fit before release.
Rolling new features out internally (across roles) often gives us a great sense of what changes are going to be exciting or hard for our customers. Internal feedback offers lead time so we can address issues or prepare change management plans for our customers.
It's also worth noting that we constantly kill feature ideas! Both at the nascent stage and after full internal releases. Sometimes they just turn out to be bad, or the timing isn't right. Often they're interesting but just aren't sticky enough to be worth pursuing
Also worth noting that internal feedback is not restricted to new features. We are also constantly providing feedback on and refining core features, and the interaction between existing and new features. The goal is continuous improvement to make Slack a little better every day.
Practically, our release process helps ensure we're constantly dogfooding both the latest features and the latest code. On desktop/web, our instance is assigned to a special Dogfood server tier which is always running the latest build at the leading edge of the deploy pipeline.
On mobile, our teams offer prototype (one-off), Alpha, Dogfood, and Production builds. Most internal users live on the dogfood build, which is updated nightly.
Overall, this approach has a few valuable outcomes:
- we maintain a culture of continuous feedback and improvement
- we can rapidly pivot based on internal and customer feedback
- we practice "good ideas can come from anywhere" and don't restrict feedback to execs/feature owners
A few things we look out for:
- internal feedback can overfit to power-user use cases. We are very good at using Slack, and don't always match the expectations of a more general audience
Customer pilots and tiered releases help counterbalance these effects.
- internal feedback can cause churn if there is not strong local conviction about direction
You can't please everyone. Ultimately a clearly articulated perspective from the team, or a strategic decision from an exec can resolve differences and avoid getting stuck in a loop.
It's sunny where I live ☀️ A perfect day for a rollup of Slack pro-tips. I've worked on @SlackHQ since the first prototypes, and always forget that most people don't know about all the "secret menu" items
These are randomly organized based on the order in which I remembered them
Type a comma in the quickswitcher to see all your most common/recent Group DMs
Open the quickswitcher to the "mode" you want:
• cmd+f for find in current channel
• cmd+g for archive search
• cmd+k for conversation switching ("Jump to")
We started the day with a connected users record of 9.8K. Almost ten thousand humans! Using the little thing we made! We made about $5K from paid seats.
It was a good day.
They say the cells in the human body replace themselves every 7 years (this isn't true but it's poetic). It feels like that at Slack: the ongoing evolution and reinvention of the code, the company, and the product has replaced much of what we started with, though the DNA remains.
The dent we set out to make in how people spend their time at work is apparent. Slack and similar products have largely replaced email for a huge number of people. People's expectations of the quality and care that goes into work software have risen.