OKRs and AI

I suppose it was inevitable. With OKRs being one of the top priorities and points of friction for most organizations these days coupled with the broad proliferation of AI tools, features and integrations, the question of aligning this new technological capability with team goals was bound to come up. There’s no denying that AI is changing and will continue to change the way we interact with digital products and services. Unlike other tech fads over the past few years, AI is already changing how we do every part of product development, business, customer service and beyond. It is the new “shiny object” executives are chasing. And, without a doubt, it is being dropped into backlogs from above as the seagulls tend to do on the boardwalks at the beach – swoop and poop! If our purpose is to make our customers successful and our leaders are demanding specific implementation of features, how do we reconcile our OKR goals in the face of this AI-driven demand? 

AI Is a Feature

Let me be clear and direct: AI is an output. It is a feature of your product or service. It is something we “make” that we hope will drive new and better customer behavior. Regardless of executive directive, AI integration into your product offering should first and foremost solve or improve an existing problem of way of working for your users. The interesting part of adding AI to your product is that it’s a technology that allows you to reinvent your user experience completely. While in many cases it can make an existing workflow easier, more efficient or unnecessary, there will certainly be opportunities to completely rethink how your customers work with your product. In all cases though, incremental improvement or total redesign, the customer and their needs must be at the center of your discussions. 

Here’s an example from my daily routine. I use an email product that I love called Spark. It has made my email life easier, more collaborative and convenient. I am a paying customer. Over the past few months the Spark team has added “AI” into the product. 

Here is an example of that AI integration. I am responding to an email that is prepping me for an upcoming podcast interview and has made some requests from me. Step 1 is to hit reply as usual. Then I’m presented with the option to have the AI feature generate a reply. 

Next, it prompts me to add context to the response. In other words, I have to start writing the response myself to “help” the AI know how to respond “automatically.” 

Finally, it generates a response that I now have to read, review, edit and then send. 

The integration of AI in this workflow saves me nothing. I still have to prompt the AI to respond in a specific way, review that it wrote something I agree with, make the proper edits and only then send the email. In this case I would have sent all of the required materials along with this email to save another back and forth. I would also have edited this to sound like me and, well, not a bot. (I never say “Best,”). 

What OKR is this feature helping the Spark team achieve? Less time spent answering emails? That’s not the case for me at least. Higher number of emails disposed of in a similar amount of usage time? Again, not for me. The reality is the Spark team added this because they, like many other companies, see AI, for now, as an arms race and they felt they had to have it to stay competitive when, in reality, I’m guessing it does nothing for their download rates, account setup percentages, usage rates and retention (all OKRs). 

Stability Over Innovation

It can be so tempting to chase the shiny object. Instead, consider the goals you have for your product. Boiled down they are likely to be some variation of Acquisition, Activation, Retention, Revenue and Referral (the infamous Pirate Metrics). What attracts your customers, gets them to use the product and keeps them coming back? In the case of Spark it’s the various collaboration tools, ease of email disposal and keyboard shortcuts, for me at least. What frustrates me about the product is that it is, on occasion, unstable on my MacBook Pro. I’ve complained about this to them more than once yet the problem persists. Innovation is important and critical to the success of your product, but not at the expense of basic usability, performance and stability. Getting the fundamentals right first will help you achieve your core OKRs.

Once the fundamentals are in place, then you identify where and how to innovate and improve the service. There will likely be use cases where AI makes a ton of sense. The question to ask, though, is in the service of what? How will AI make the user experience better, faster, more accurate, more efficient, etc ? Then, once the AI integration is deployed measure those user behaviors (aka key results). If they haven’t changed, find out why. For example, if the Spark team would ask me, I’d tell them the AI feature is literally useless for me. Now I may be an outlier but I’d guess that there are many other folks like me who’d prefer to write their own email responses in their own voice. 

OKRs Make AI Implementations Realistic

If it hasn’t happened already, your boss will come to you and ask you to integrate AI into your product. Your job is to remind them that AI is a feature and to ask, “What problem are we trying to solve with AI?” You should also ask, “How will we know the AI integration is helping the customer experience?” And finally, “What customer experience are we NOT going to do while we work on the implementation, fine tuning and maintenance of the AI feature?”

These are the same questions you’d ask of any feature. If an idea, regardless of how shiny, new or powerful, isn’t going to make your customers more successful, wait. Wait until you have a better sense of where AI can help solve real problems in your product. Ensure that the effort you put in yields a better user experience for your customers and better business results for your company. Otherwise, we’re ignoring the core value of OKRs and filling our product with bloated software that we’ll have to maintain forever.