I’m looking at this from the lens of my own org, which I think strikes a decent balance. As a brief intro, I work in an industry where a day of downtime can cost our customers millions, and a bug can also cost millions due to regulatory fines.

Estimates are bottom up, priorities are top down, and daily reports are largely opportunities for short form discussion about changes to requirements, delays, etc. We spend about 15 min/day, M-Th, on stand-up per team. We have six teams spread across Asia, NA, and Europe, so face to face communication is difficult.

With that out of the way, I have some comments, which I hope will lead to good conversation.

Build projects around motivated individuals

I disagree with this, but perhaps I’m misunderstanding. Projects should be built around customer needs, and individuals should be hired/assigned as needed to fill those needs. Individuals should be given leeway to work on projects they’re interested in, but only within the priorities set based on customer needs.

The focus should remain on trusting your employees, but there needs to be frequent communication so problems can be caught and addressed early.

The most efficient and effective method… is face-to-face conversation

Again, I disagree, the most efficient is text, and the most effective depends on the information being conveyed. Stand-up works well face-to-face, but technical review can usually be handled through text, with face-to-face being an outlier.

My rule of thumb is if there’s more than 3 back and forth messages, we probably need a face-to-face discussion. I can handle several text-based conversations simultaneously, while I can only handle one face-to-face convo at once, so most things should be text first with face-to-face being a fallback.

if your organization is driven by processes and procedures

This is a squishy concept since too much or too little process leads to problems. I think the amount of process should be directly linked to the risks associated to violating that process.

For example, here are some processes we have in place:

• deployments to production require approval from key stakeholders - product team, project manager, support team, OPs, and the relevant team lead
• larger projects require approval from architect, product team, and relevant team lead before work begins
• larger code changes (e.g. introduce new core tech, large scale refactor, etc) require informal consensus from other team leads, which includes a period for those teams to review the proposed changes

Each of those is based on risks directly tied to customer needs. A bad deployment can break customer use cases, cause support overhead, etc. Projects starting without approvals can miss requirements, overrun estimates, and delay other important projects. Large code changes without proper communication can delay other projects who have to manage conflicts with the new change.

So the focus should be on keeping processes that provide value and retiring those that don’t. For example, building confidence in automated testing can help ease deployment anxiety and thus loosen deployment processes.

agile tends to just be a hollow buzzword

Agreed.

Also, agile itself isn’t ideal for all projects. We find value, but we’ve made some changes to fit our customers expectations. We release about once/month, and more frequent releases are generally not wanted by our customers.

So we have each team release to prod every 2-3 months, usually combined with another team. Agile prefers more frequent releases (daily if possible), but we’re not really interested in doing more than two in a month.

Anyway, great post! Hopefully people recognize the BS in their organization and push for change.