Does your data science organization charge groups internally for your work? Are you set up like a typical IT development team that requests requirements, produces estimates for the work, and then bills it? I have yet to see a world-class data science organization that functions this way. Not only because the very act of billing creates friction in the system that lowers the speed and creativity of the solutions being built but also because it is a more systemic sign of the culture and mission of the organization.
There are solutions we have built-in hours or days as a data science organization. It would have taken weeks to create an estimate, get approval, move money, charter the project, and execute all of the IT processes I see within many companies. A typical larger-scale project at a major corporation could easily take weeks, if not months, to get approved, and if it wasn’t already contained within the budget cycle, it might need to wait for the next year to get started.
When I discuss this with many colleagues who are struggling within their own companies under this model, they are surprised that there are organizations out there that do not bill. I suggest that they have other support organizations at their companies that do not find billing to be necessary: Finance, HR, Legal, and typically every support organization other than IT. So why would data science follow the IT model versus these other organizations?
While data science has a component of the discipline that is highly technology-focused, leveraging algorithms and computers to do much of what they do, that does not mean it should follow the same processes and norms that are seen within an IT organization. One could argue that nearly all organizations will continue to grow the amount of technology and algorithms used in their business as well.
Negatives of the billing model
What are some of the negatives of the billing model? There are likely too many to list, but here are some of the few I see:
- It creates no value for the shareholder or the business: The internal movement of money is artificial and is all non-value-added work that simply costs money.
- It creates a build-to-specification mentality: If I’m a businessperson in the operation and I am asked to pay $140,000 to have someone internally build a solution for me, I expect that the solution does exactly what I specified. I want the names of those working on the project and a detailed timeline of what will be done. This contradicts the iterative, collaborative problem-solving methods used in data science. We want data scientists to be trusted advisors of the business that help drive transformation, and the billing doesn’t help support that goal.
- What gets done is based on who pays: The best data science problems that will have the most transformative effect on the business may not be where the business unit has allocated the funding, and if there is no funding source, the work will not get done. If you want data scientists to transform your business, you can’t assume you know where the work will take place in advance.
I have heard some suggest that billing is needed to prioritize work; however, this doesn’t appear to be the case for all other departments that do not use this mechanism. Finance teams know when to say ‘no’ to a request and where they will insert themselves. I would argue that if they do not know how to do this, billing is likely not the best way to solve the problem.
Why some say billing is important
So why do some say billing is important? I have hard people say that:
- It is a way to measure value: However, ROI is not measured by how much one estimated a project would cost or what someone paid; it is based on the real cost to deliver it and the real savings/impact it made on the business. I agree that measuring ROI can be important, but this should not be confused with billing.
- We need a way to prioritize the work: Again, if you can’t tell what is important, billing certainly isn’t going to help. You will now be working on the items that departments happen to have money for, but this isn’t necessarily based on the impact it will make.
- It allows the business to competitively bid the work outside: Certainly, the choice to go outside can be executed in any function, from legal, to finance, and HR. The decision of what to outsource is likely best made by the domain expert, not by the operation simply quoting each item. Legal decides when to leverage outside counsel, not the department seeking legal support. Data science shouldn’t be any different.
In the end, if the goal of your company is to have more people using data and analytics to solve problems, billing models simply add friction and reduce the chance of success.
I have not seen other organizations that engage in problem-solving work under a billing model. Six Sigma black belts do not work up estimated costs to bill their work. Engineers who are heading to a manufacturing site to solve an issue don’t typically ask for funding. If anything, I wonder if, one day, IT organizations will change their models to drop the billing. We don’t have a billing model at Alteryx for our Data Science organization, and we don’t have one for IT either. What does your business do? Would you like to drop your billing model?