Shonna

Every non-technical startup founder I meet asks "how much will it cost to build my app?". The answer I’m usually thinking in my head is “a lot”. The hope-filled answer is "not as much as you probably think" thanks to the explosion of No-Code options. I'll explain a little. 🧵

If you've never created something of your own — a business, newsletter, website, app, course, blog, book, anything... You're missing out on the greatest avenue possible for reaching your potential.

Transparent work processes help uncover inefficiencies and optimize workflows. Continuous improvement becomes easier when we can see where we can refine and streamline. bit.ly/3Z60AQv #SoftwareDevelopment #MakeYourWorkVisible

Your product should meet demand, not attempt to create it.

There are only three "planning poker" values that are useful: 1, Too F'ing big, and No F'ing clue. You can even get a card deck for that: estimation.lunarlogic.io

No good manager tells anybody what to do. The good ones describe a domain-level outcome or a business' strategic goal and trust the people doing the work to figure out the rest. This observation also applies to Product Managers/Owners.

I'd rather work with a human than a requirements document. It's not that I'm gregarious, but I'm more likely to get what I need when I need it, have the info that backs it up, and have answers to the questions I actually have rather than the ones they predicted I might.

People keep asking me about how to get into Agile Coaching/Consulting as a profession. I don't want to sound like a gatekeeper here, but it's not an entry-level job. Sorry.

I have seen teams complete hundreds of story points, and the business impact was exactly the same as if they'd taken six-month vacations. Points are dust on the scale. I can live with using them, but please don't talk about them like they mean something.

The hard-metric obsession in the software engineering community is ineffective at best and destructive at worst. The military demonstrated (but maybe not learned) that lesson in the Vietnam War (see McNamara fallacy). Deming was absolutely right when he said, “You can only measure three percent of what matters.” Software engineering is about people, who are inherently complex (in the Cynefin sense). Numbers are of little to no help in that quadrant.

Re user/customer feedback: First, questionnaires don't seem to work very well. They typically ask the wrong questions, are too narrow in scope, and often have an implied bias. An actual conversation is better. Chatting while you watch a user/customer work is ideal, provided that you spend more time listening than talking, and you listen with an open mind instead of trying to manipulate your user to your way of thinking. That last bit is often subconscious and needs active attention. I often hear people try to turn whatever the user's saying into something that makes sense to a programmer but not to a user.

Making work visible is secret sauce for effective software teams. It promotes transparency and collaboration. #SoftwareDevelopment #MakeYourWorkVisible

Curiosity takes you farther than judgment. Instead of saying, "That's wrong," say, "That's interesting." Ask a question. Try to understand. The moment you think you know everything, you're doomed.

The Vegas principle applies to the inner workings of a team: What goes on in the team stays in the team. Transparency is essential in agile organizations. But this does NOT mean that you deliberately expose your team to micromanagement and the like. Don't, for example, supply management with easily abused metrics that you use internally for improvement. The way to get transparency with management is to deploy improved software every day. They can monitor progress just by looking, and they can't micromanage that.

Don't budget work. Don't budget projects. Instead, invest in a team, then assess the work and have that team work on the most valuable thing. Starting with the budget is a failed strategy. Money is a constraint, not a driver. If your customers really are demanding something, then you can't afford not to build it. Saying "We don't have the budget to build the one thing that really matters" is a failed business strategy. Admittedly, startups and the like may not have the funds on hand to build everything you'd like in all its glory, so scale down the scope to the core of the core solution and build that.

Re roadmaps: I strongly believe in strategic roadmaps. In my experience, tactical ones are usually a waste of time. One reason I don't really like the term "roadmap" is that it implies a well-defined route to me. I think more in terms of a sea chart. Set a strategic goal, map the hazards, and move towards the goal it in small increments, avoiding the hazards. You can't control the wind, so the route is flexible. You may even be driven backward sometimes, but the goal gets you turned back around.

Combine Git with TDD "tests" as spec, and most regulatory issues are covered. I have worked with auditors who were happy with the Git+TDD approach, so this is not a theory. They were actually happier with TDD specs than paper ones, in fact, because it was easy to determine that the spec was satisfied simply by running the test.

"Concerned about the possibility of developers running out of tasks." The more I think of this, the more I see wrong. The core problem, I think, is that an organization like that does not have a product focus. When the team mandate is "improve the product," there's an infinite amount of work to do and nobody has to order people around and assign "tasks" for productive work to go on. A concern about "tasks" is an indication of no trust and single-point-of-failure decision making. None of that is good.

Hard WIP limits on a backlog are critical. The backlog is small and of limited size. Nothing goes on unless something comes off. Engineering must be absolutely hard nosed about this. Not only does it improve engineering productivity, but it imposes business discipline on the “stakeholders.”

Can somebody explain to me what a "Delivery Manager" does that can't be done better by a good automated DevOps system and a well-functioning dev team?

Finishing a project "on budget and on time" is not an indication of success. It is often merely the result of gaming an easily gamed system. #BeyondEstimates