Dear Gib: How do you handle CEO "pet projects?" Especially when they are clearly bad ideas?
Short answer: Find the balance between obstinance & indulgence, and learn to debate, decide & do.
Note: I answer one question nearly every day. Please click here to ask and upvote questions.
Tough question. On the one hand, I encourage product leaders to “behold the idea and not the source”—don’t give the idea extra weight just because it comes from the CEO. On the other hand, being a CEO is a tough job. He/she deserves your respect and likely has better than average judgment about people, products, and the business.
After decades of both success and failure, this is how I handle the dilemma:
Ask lots of questions to ensure you understand the CEO’s intent.
Engage in a debate, using existing data, qualitative, surveys, and AB test results.
Agree on a plan to get to the answer, which sometimes means executing the idea to see if it works.
In general, handling disagreements in an organization comes down to learning to debate, decide, and do. A bad outcome is open, festering disagreement. An even worse outcome is agreeing on a plan and then trying to block its execution.
I’ve made many mistakes in this area. In some cases, I was overly obstinate, which drags things out. Other times, I was too indulgent of the CEO. Below, I recount some stories of how I handled CEO debates to give you more insight into navigating this situation yourself.
1999: I was obstinate, and there was no debate or learning.
In 1999, I was the VP Product at FamilyWonder, an e-commerce startup. The founder had a high need for accuracy and control and felt our website lacked polish. He wanted to take four weeks off from planned projects to clean up the site. I didn’t think these cosmetic issues mattered, so I focused instead on executing our plan. I had my way—until I took a two-week vacation, and the CEO re-prioritized everything while I was gone. The result for the team was chaos. Looking back on this, the CEO and I should have found a better way to work out our style differences.
2005: Netflix movie-finding tools: Overly indulgent? Not sure.
When I joined Netflix in 2005, I found myself running a startup team of product managers who warned me about Reed Hastings’ pet project. Reed had hypothesized that unique movie-finding tools would improve monthly retention. (Example: an on-screen “Movie Genie” asks a question, then recommends a movie). The product team believed it was a bad idea, based on lukewarm responses in focus groups and usability sessions. They hoped I’d kill the project.
By then, I had learned to have more productive conversations with CEOs. I spoke to Reed, who was dubious about the value of focus groups and usability. We wouldn’t make a decision based on qualitative. We agreed to get the movie-finding tools into A/B testing as quickly as possible.
The movie-finding tools failed. But along the way, we were able to convince Reed of the value of other data sources: qualitative (focus groups and usability), existing data, and surveys. Some of our best ideas came from listening to customers, and our intuition was sharpened by having the voice of customers in our heads. The product team also worked to get projects into AB tests faster— sometimes skipping qualitative to make faster go/no-go decisions. However, two years later, I chuckled when I found Reed behind the two-way mirror watching usability for our new TV-based designs.
2008: No “Settings” or “Help” on our first set-top box: we found out Reed’s intent and put it into action.
When we developed our first TV-based system, Reed insisted that there be no “Help” or “Settings” menu items, which felt crazy to me. If a customer gave their device to another member, how would members reset the box? But through debate, it was clear that Reed wanted an experience that was so simple, customers wouldn’t need help. He also dismissed the setting issues as “edge cases.” We’d have to find another way to handle this.
Our first set-top box interface had no help or settings, but when hundreds of customers called Customer Support asking for cheat codes to help them solve their “edge cases,” it was evident that the next version of our TV-based interface would need both help and settings.
At the same time, however, we looked at our website experience through a “breathlessly simple” lens and realized there were many opportunities to improve the service. Eventually, we even proved that a simpler experience improved retention. Reed “lost” on Help/Settings, but his intent—to create a simpler experience—took hold throughout the product organization and in our product.
2008: The launch of a Netflix-Ready Device store on our site: an easy compromise.
We launched our service onto multiple TV-based devices and needed to build a store on our website so members could learn more about new “Netflix-Ready” game systems, Blu-ray players, and TV set-top boxes. Reed and I disagreed on the store's design, but we quickly had a “D’oh!” moment: why not launch both experiences in an A/B test to see who was right? Honestly, I can’t remember whose version won, but we did get an answer quickly without turning the organization on its head.
So, here’s what I’ve learned about CEO Pet Projects:
Treat the CEO’s ideas with respect, but dare to challenge him/her if you disagree.
As with other individuals in the company, learn to debate, decide, and do.
Get to the “Why?” behind the pet project. If you can’t find data or build an argument to kill the project, agree on the next steps and how/when you will know which of you is right.
Win or lose, catalog the learning to determine whose judgment you can trust on which issues.
Don’t let a long, drawn-out debate create chaos in the organization. Agree on a plan and support it. If you find yourself executing a project you don’t believe in, don’t disparage the CEO. Do your very best to make it a success.
No matter the outcome, focus on learning to build insight across the organization, making decision-making faster and easier in the future.
I hope you found this helpful.
PS. Click here to give feedback on this essay.
PPS. This question had 52 upvotes: