A company I previously worked at, DataRobot, is a case study of succumbing to cargo culting. They had a great story around AutoML during the ML heydays, which got them in the door of many large customers. The problem was that after selling on the promise of value—leaving execs drooling over potential use cases—many customers would review implementation at renewal, realize no models had gone to production, and then churn.
This was not the case 100% of the time. DataRobot did have happy customers—like those who purchased full end-to-end solutions upfront or simply needed better model development software. But it was a fundamental problem that was never going to be solved with more software alone.
In my opinion, the fork in the road was that they charged for services. This one decision had many downstream effects. First, it penalized customers for seeking DataRobot's help in embedding its software into their production environment. Second, it loudly signaled that we were a software company, not an outcomes company. Third, it prevented DataRobot from getting the necessary feedback from end users to innovate on the product.
DataRobot saw services and customer hand-holding as a negative thing when, in reality, doing so for free would have been the key to unlimited customer access and the best possible feedback loop for product development.
I believe a major source of pressure was a board largely composed of investors restless to create the next Salesforce or Snowflake. At a critical moment—just as DataRobot was finally finding success in delivering solutions and outcomes to the government—they replaced the company’s technical founder with a COO lawyer and a group of GTM execs who thought a better GTM motion was the only thing stopping them from lucrative IPO.
Now that software is becoming commoditized, it’s clearer than ever that selling outcomes was the real product all along. Software can certainly be part of the outcome, but never the outcome by itself.
Shyam, I so enjoyed this writing. The Costco screen shot was historical (hysterical). As beautifully as this is written, Palantir will need to guard against being the next screen shot. The reason I love the tools Palantir offers is because they not only show in real time how fat and dysfunctional a bureaucracy is, but the tools also offer the ability to act. The key word is offer. I have a treadmill at my gym, a marvelous tool. It has a dynamic interface that allows me many options to measure and track my progress. I pay a hefty fee to maintain my membership. There are many people at the gym working to achieve their goals. What more can I ask for? Motivation. My bureaucracy decides to use the treadmill as a place to set my smartphone and my water bottle. It is a perfectly suitable shelf. I believe Palantir needs an Army of motivational coaches. I call these people Palantir Prophets. They need to be laser focused on the correct and continued use of the Palantir Treadmill. The Mission of Palantir is to save Western Civilization. AIP is only the first 80%. Correct and continuous use of the tools is the other 80%. I know you have a partnership with AlixPartners to help with this challenge. Be mindful that they are the type of organization in charge of helping Costco.
A company I previously worked at, DataRobot, is a case study of succumbing to cargo culting. They had a great story around AutoML during the ML heydays, which got them in the door of many large customers. The problem was that after selling on the promise of value—leaving execs drooling over potential use cases—many customers would review implementation at renewal, realize no models had gone to production, and then churn.
This was not the case 100% of the time. DataRobot did have happy customers—like those who purchased full end-to-end solutions upfront or simply needed better model development software. But it was a fundamental problem that was never going to be solved with more software alone.
In my opinion, the fork in the road was that they charged for services. This one decision had many downstream effects. First, it penalized customers for seeking DataRobot's help in embedding its software into their production environment. Second, it loudly signaled that we were a software company, not an outcomes company. Third, it prevented DataRobot from getting the necessary feedback from end users to innovate on the product.
DataRobot saw services and customer hand-holding as a negative thing when, in reality, doing so for free would have been the key to unlimited customer access and the best possible feedback loop for product development.
I believe a major source of pressure was a board largely composed of investors restless to create the next Salesforce or Snowflake. At a critical moment—just as DataRobot was finally finding success in delivering solutions and outcomes to the government—they replaced the company’s technical founder with a COO lawyer and a group of GTM execs who thought a better GTM motion was the only thing stopping them from lucrative IPO.
My ex-colleague Thomas Dinsmore elaborates on this further here: (https://www.linkedin.com/pulse/turmoil-datarobot-thomas-w-dinsmore/)
Now that software is becoming commoditized, it’s clearer than ever that selling outcomes was the real product all along. Software can certainly be part of the outcome, but never the outcome by itself.
Shyam, I so enjoyed this writing. The Costco screen shot was historical (hysterical). As beautifully as this is written, Palantir will need to guard against being the next screen shot. The reason I love the tools Palantir offers is because they not only show in real time how fat and dysfunctional a bureaucracy is, but the tools also offer the ability to act. The key word is offer. I have a treadmill at my gym, a marvelous tool. It has a dynamic interface that allows me many options to measure and track my progress. I pay a hefty fee to maintain my membership. There are many people at the gym working to achieve their goals. What more can I ask for? Motivation. My bureaucracy decides to use the treadmill as a place to set my smartphone and my water bottle. It is a perfectly suitable shelf. I believe Palantir needs an Army of motivational coaches. I call these people Palantir Prophets. They need to be laser focused on the correct and continued use of the Palantir Treadmill. The Mission of Palantir is to save Western Civilization. AIP is only the first 80%. Correct and continuous use of the tools is the other 80%. I know you have a partnership with AlixPartners to help with this challenge. Be mindful that they are the type of organization in charge of helping Costco.