This is Johanna Rothman's June 2025 Pragmatic Manager newsletter. The Unsubscribe link is at the bottom of this email.
All my clients are dabbling with AI. They're using it to think about more, such as generating more features in a backlog so each person can do more. Or more code with vibe coding. Or even more writing, so a person can do some other work that is “more important.”
These managers are not stupid. However, they're optimizing down, for people's time, instead of up to complete an overarching goal. Worse, they're not seeing the benefit to the customers in the form of releasing better products faster.
Instead of optimizing down, we can focus up, using the team, project, or program as the unit of work. This allows the organization to progress more quickly on its product strategy. In turn, that allows everyone to progress on the corporate strategy.
What do we need to optimize up? Think in “how little” terms:
- How little WIP can we have right now? (That decreases cycle time and increases throughput.)
- How few features do we need in a backlog or a roadmap? (The fewer the features, the more flexibility we have in choosing the next features.)
- How few management decisions do we need right now? (Fewer decisions means faster management feedback loops.)
At a client's request, I asked ChatGPT this question: How can we use AI to help us optimize up for the entire organization?
My Prompts Offered Generic Answers
I asked more questions, but all of my prompts offered what I would call generic answers because I did not have use my client's specific processes. Thank goodness ChatGPT recommended my books. (Insert wry grin here.)
Some of Chat's answers included many topics I've written a lot about, including limiting work in progress, empowered teams, clear and ranked goals, and faster feedback loops. But how can managers learn to optimize up?
Consider these three starting questions you can use to see where your processes prevent people from optimizing up. If you, like my client, love AI, ask your LLM of choice to analyze your processes for these three areas:
- Identify processes that conflict with and between each other. The more conflict, the larger the WIP and the longer the decision loops. That tends to make managers focus down, not optimize up.
- Processes that create conflict between teams, so people don't know how to decide or experiment.
- Processes that create blame instead of encouraging experimentation and real risk management.
As with any data, decide what you can trust your AI tool with. Remember, a human being might be able to assess these various conflicts better than a bot. But these three ideas work regardless of who/what assesses them.
1. Conflicting Processes That Create Delays
Tom just started a new role as the Engineering VP at ACME, a mid-size organization. The PMO wants estimates of projects to decide on the project portfolio. Tom is accustomed to creating options and deciding on the portfolio based on investment decisions. (How much does the organization want to invest before they stop with this effort and move to another?)
Clyde, the head of the PMO, described their process. He always asks teams to estimate how long their work will take. That often takes weeks—a significant delay—to return an answer. Worse, it interrupts all the current work. That's because it's impossible to know how long anything can take until a team starts it. However, Clyde likes this process.
While Tom isn't wedded to his approach, he knows it works. Worse, he's never seen Clyde's process work. Even better, when the senior leaders discuss the portfolio, they describe to each other the value they place on the work.
As with many bureaucratic problems in the organization, this one started for a good reason. Several years ago, a new CEO, Allen, started. He could not understand how such a small organization could have hundreds of projects. Allen hired Clyde to get a handle on all the work, hire the right people to finish the projects, and help him, Allen, then choose what to do next and not to do next. All good reasons.
Now, as the company has evolved, the old problems are no longer the current problems. This is often the case with many processes. The organization has outgrown all the reasons for the process.
Worse, this PMO approach prevented collaboration between Engineering and Product.
2. Separate Goals that Often Create People-Based Conflicts
Jane, the new Product VP at ACME, was surprised when Clyde told her she was not supposed to join them as they discussed the project portfolio. She asked, “How am I supposed to clarify each product's strategy? I need to know what's fixed in our strategy and where we want to experiment and take bigger risks.”
Clyde had never thought of product strategy that way.
That's when Tom said, “If we know our product strategy first, we can more easily choose how little to invest. Even better, if the strategy isn't working, we can decide what to do as an alternative. We want to collaborate with Product to create the best value for our customers.”
That's when Clyde asked, “What about all those status reports I need? How will I know who to blame when things don't go well?”
3. Processes that Create Blame Also Create Delays
Status reports are often a way for people to CYA (Cover Their Tushes) and avoid blame. Very few status reports are worth anything, because they don't discuss the value the project or product offers. Worse, anytime there is even a hint of blame, people stop talking about or taking risks. Instead, they stick with the current bureaucracy and the current stultifying products.
However, the whole idea behind product development is that we have an opportunity to innovate. It does not matter how much planning or risk management we do. As product developers, something will surprise us. At the worst possible time, of course.
The way to avoid surprises is to do real risk management and acknowledge that risks will occur in any work of value. The real issue then, is how do we manage those risks, not paper over them?
I've offered just one perspective on these three problems. I suspect any AI interrogation with more specific prompting might uncover other problems and more perspectives.
Ask AI for Help to Optimize Up for an Overarching Goal
This particular client swears by AI. (I'm not convinced, but he is.) Since he was willing to feed all his processes into his LLM, he saw more opportunities to use overarching goals, reduce people-based conflict, and reduce blame. Those opportunities will make work better and be more effective—for everyone.
You don't need to use AI for this. Instead, you can examine your current processes.
However you choose to investigate, review your system of work with an eye to optimize up for the greater organizational goal. Avoid cascading goals down, because the sum of the parts does not create a whole. Instead, ask, “How can I (and then my team, project, program, etc) work to achieve the product strategy and the corporate strategy?”
All leaders and managers have the ability to change the system to focus up on fewer things. Look for processes that create conflict and blame as a first step. That will make the work better for everyone and allows everyone to be more effective.
Read More:
This newsletter touches on several of my books:
- The Modern Management Made Easy books
- Predicting the Unpredictable: Pragmatic Approaches to Estimating Cost or Schedule
- Manage Your Project Portfolio: Increase Your Capacity and Finish More Projects
- Project Lifecycles: How to Reduce Risks, Release Successful Products, and Increase Agility
If you're not sure which book is right for you, reply and let me know your specific challenge.
Learn With Johanna
I'm integrating the technical review comments for Effective Public Speaking. Then, I'll be able to finish the book and offer my Kickstarter. I'll send you the link so you can decide if you want to back it.
If you are part of the agile community, consider checking out The Agile Network. Also, don’t miss out on discounted membership options. Use Discount Code: ROTHMANPMC33 to get 33% OFF all memberships. I’m speaking several times at the June micro-conference next week.
Links of Interest
New to the newsletter? See previous issues. (I post these newsletters to my YouTube Channel a few days after I send them.)
Here are other links you might find useful:
- Managing Product Development Blog. (Yes, I offer an RSS feed so you can read everything in a newsreader.)
- Create an Adaptable Life Blog to see the weekly question of the week.
- My Books. You can also buy my self-published ebooks and audiobooks on my store.
- My Workshops
- Johanna’s Fiction
Till next time,
Johanna
© 2025 Johanna Rothman
Pragmatic Manager: Vol 22, #6, ISSN: 2164-1196