Data & AI Strategy: Running a Profit Centre
Part 4: Transforming Data & AI cost centres
This is part 4 of the Data & AI Strategy series. Check out the below posts for the previous entries:
Data functions tend to be cost centres in most companies.
If you are a data leader, you must be familiar with the struggle.
There is a myriad of reasons why this is the case:
Most companies don’t really need big, centralised data functions; if you are small enough, spreadsheets and Excel are all you need. If you have <15 employees, you don’t need bespoke data analysts—just look at your own data.
Similarly, you shouldn’t invest in the modern data stack if you don’t have/need the scale; use spreadsheets-as-databases (just make sure to use rows instead of columns to record observations…apparently the NHS didn’t get the memo).
In most industries, there is no need to hire PhD-turned-data-scientist profiles. Tech is not the core business, but an enabler. That ex-academic will be demotivated before you can say ac hoc EDA for random stakeholder.
In a lot of companies, there is no DS/ML/AI-enabled next step; sure you can have a more accurate churn predictor, but that’s not going to transform your business.
Because of FOMO, data professionals are paid quite handsomely.
The last bit makes matter uniquely worse for Data functions, especially when combined with tech not being the core business: running a support function that costs a lot.
This is not good news. Most functions in a business exist in a support capacity; however few cost as much as Data. When you run a Data & AI function, unless you are an AI company, congratulations! You are running the most expensive support unit.
You might think being a support unit can’t be that bad, after all, the expectations must be tampered accordingly, right?
🤣
Anyways, this is your predicament:
You run a department that is not core to the business; regardless of how many times the CEO says they want to be data-driven, they rather increase the marketing budget by x million before investing in data
Your contributions—moving the needle—heavily depends on cross-functional initiatives; do something for sales and pray they give your team their due credit
Your people cost a lot of money; making the FP&A person drool at the opportunity to cut costs
Because your department is so damn expensive, the leadership conveniently forget points 1 and 2 and expect the world in return
In other words, you are not prioritised, you can at most have indirect business impact, you are expensive, and because you are expensive, you must be really good?!
Chef’s kiss🤌
Tenets of a Profit Centre
This wouldn’t be an engaging blog (is it, though?) if I only whine and offer no solutions.
I take a quite simple approach to this problem, so I have nothing to sell you in terms of elaborate frameworks and such. I think the best way to turn a cost centre into a profitable one is to ask the leadership how you can achieve that.
As in, when you are drafting the next quarter’s OKRs, reach out to CxOs and influential VPs/directors whether there are any low hanging fruits in their domains that you can help with, as well as any long-term initiatives. The former is a quick win that you can deliver in a quarter or so. The latter are multi-quarter projects.
While you collect this data, get together with your team and brainstorm what you think you should prioritise in the next quarter and the next 12 months. Now you have to lists:
An external list from execs/leadership
An internal list from your own team
Both lists contain potential quick wins and some long-term undertakings. The next step is why they pay you—distil these two lists into a 3x3* OKR structure that your team can deliver. Things you want to balance are:
Doing enough visible ROI in a quarter to justify your team’s existence
Invest time and bandwidth into some of the long-term projects that will transform the company
For the first part, literally collate a list based on what the leadership came back to you with, and pick ones that you think are the easiest. Never ignore everything they said because you know better! Especially after explicitly asking them. If you are going to ignore execs and play your own tune, might as well keep your mouth shut.
For the second bit, you have more leeway. These are things like moving away from spreadsheets-as-databases in medium and above sized companies, switching to a new BI tool, deploying a LLM into production (and keeping it there) etc. You still want to align with the execs, so don’t do tech for tech’s sake.
The dance is to keep feeding execs quarterly business impact while working towards truly transformative initiatives. If you dedicate all your resources into big projects, it is likely that you won’t last long to see them done!