The Mythology of AI and OKRs

OK, that’s a lot of alphabet soup already, but what is this about??

At the beginning of 2023, all the tech futurist pundits were breathlessly telling us that our world would be transformed by Artificial Intelligence in 2023. Months later, Gartner has located AI at the peak of what it calls the “hype cycle.”? In the world of OKRs, nearly every OKR software company is rushing to include AI features to help their customers write OKRs. This on top of previous attempts to provide libraries of OKRs for users to copy and modify.

I argue that this is the result of a misguided tech fantasy. And, it encourages lazy thinking.

Why?

1. OKRs provide the framework for agile implementation of strategy. Being agile requires teams to challenge each other’s assumptions what about outcomes and measures are meaningful. And that will change over time.
2. Strategy is a constant search to define and execute on a unique value proposition, based on a series of choices about products, markets and capabilities.
3. Artificial Intelligence is just that — artificial. It? is a machine generated summary of what a large number of people have previously said about a particular topic, which, in the absence of supporting data, may include machine generated interpolations known as “hallucinations.”

If you follow my logic, how on earth could you convince yourself that machine-generated OKRs will do anything other than put you squarely in the ranks of …. everybody else? By deriving your goals from a pre-existing library of old karma, how do you expect that to help you manifest and deliver on a unique selling proposition?

AI-generated OKRs remind me of the fad for “best practices” we saw when I worked for large consulting firms back in the 90’s. Clients would ask for a list of, say, best practices for managing payroll and benefits, and we would dial up our firm’s repository on Lotus Notes and spit out a list that we could massage into a PowerPoint slide. The long ago 90’s were a different world in business, mostly featuring mature enterprises in stable industries trying to improve margins by becoming more efficient. That’s what we were selling, re-engineering and all that. In the uncertain world we live in now, better margins are necessary, but not sufficient. Efficiency is useful if you’re trying to improve the bottom line in a mature business in a stable market, or to manage cost while you grow. And if you become more efficient the same way as your competitor, no problem. That’s not what you’re competing on.?

So, “keeping up with the Joneses” with AI-generated OKRs might be a useful starting point for improving operations in parts of the business, but I would strongly argue against using them at the top line? level — the place where you generate revenue by offering something no one else is. This is where the hardest work has to take place, translating a unique strategy into unique OKRs. They are powerful because they are? distinctive, not the result of a collective hallucination. What gets measured, gets managed, so OKRs keep a focus on executing this unique strategy.?

At the end of a two-day leadership team workshop I facilitated last week, one of the participants said “It’s so easy, it’s hard.” Everyone laughed and agreed — what that meant was that coming up with a clear, simple OKR — that reflected what was important to this team — took a lot of work.?

OKRs should be hard to write. They have to come out of a challenging conversation about your competitive landscape and strategic priorities. No machine is going to do that for you. They are “easy” because well written OKRs are clear, their intent and implications obvious. Hard, because more often than not a clean, clear OKR is the result of a cloud of ideas about big visions, metrics, and actions that need to be sorted and bucketed into crystal clear Objectives, Key Results, and Tasks.

The best I’ve seen coming from AI so far is to take a cloud of ideas and sort out the grammar - what are the Objectives and measurable Key Results, along with a list of activities you think will get you there. And, truth be told, some folks aren’t very good at distinguishing among these. In those cases, AI can be effective at grammatical sorting.?

But not at generating innovative ideas in the first place. The hard work is the conversation. Among humans. For a senior leadership team embarking on transformative change, this is a process of:

1. Reviewing, debating, and if necessary clarifying and revising strategic choices about where to play, how to win, and what capabilities are critical.
2. Creating a high level North Star that describes what you want the future to look like
3. Identifying “What would have to be true” in order to make progress toward the North Star
4. Breaking that desired progress down into what matters in the next 90-days
5. Writing no more than three cross functional OKRs
6. Running regular check-in conversations to review progress and adjust course as needed
7. Conducting a retrospective at the end of 90 days, rinse and repeat.

Notice that all the steps leading up to the creation of OKRs are entirely dependent on conversation and alignment among a group of humans!?

I’ll admit that I’m on the Luddite end of the spectrum when it comes to new technology. I still wear an analog watch. When I first heard of the internet in 1991, I didn’t see the point. Within a few short years it had changed my life.?

What I am convinced the AI evangelists get completely wrong, though, is a very limited view of what intelligence actually is. If you do a search on “How many kinds of intelligence are there?” you’ll find lists ranging from 4 to 12 types of intelligence. AI emphasizes two kinds: linguistic and mathematical. You know, the stuff we got graded on in school. But there are all kinds of other intelligences, notably emotional, social, spatial, kinesthetic and on and on depending what list you look at.?

Point being, effective leaders tend to be people who have high levels of multiple intelligences. When my clients do OKR check-ins, I encourage them to bring gut feelings, intuition and hunches into the conversation in addition to what the data says. As Larry Bossidy and Ram Charan pointed out years ago in their book Execution, “How well people talk to each other absolutely determines how well the organization will function.”

Many operational measures, the kind that tend to measure efficiencies, do tend to be stable over time. So what somebody else used as a metric might be useful. That’s more relevant to KPIs than for OKRs.?

With OKRs, measurement is an imperfect and ever evolving art, choosing what to pay attention to out of a huge sea of data. What we choose to measure ultimately arises from a holistic sense of the world we’re operating in, focused through a rigorous conversation among humans. And what mattered six months ago may not be so important today, particularly in the growth edges of our organization.?