Managing Data Products in Hurricane-like Headwinds

Every day I open the news, I see another layoff. In the last few days that has been OpenDoor, Stripe, Chime, not to mention hiring freezes across the board. It is no secret that tech is sailing into headwinds, but it seems more likely those headwinds are hurricane force and they will be sustained. If you’re like me, that creates a lot of unease, not just about what it means for 2023, but what it means for investment in things like data product. Now, more than ever, we’re going to be faced with the question “is this really the highest priority?” We might have gotten away with squeezing this work into the margins a year ago, but that will no longer be good enough as we sail into this storm.

So let’s answer the question - “is this really the highest priority?” Having had this conversation many times, the instinct seems to be to justify its importance. “This data product allows us to do this” or “customers really love this feature” or “we couldn’t run the business the same way without it”. The problem is that framing misses the mark. “Is this really the highest priority” means being explicitly clear about the opportunity costs of an effort. Assuming fixed (or slightly reduced) headcount/time to work on something, its important to explain why this is worth it relative to everything else we could be doing. Here are some options/approaches you might take here.

1. Switch to keep the lights on, and even then, at a flicker. There are probably a lot of data products you’ve built that are useful to a specific group of people that can be kept on without a great deal of attention. If you are certain that this is the case, proposing KTLO (as much as I don’t love the phrase) is a way to keep something alive without needing to explicitly put time on a roadmap for it, or if it must be there, perhaps < 5% of time. The tricky thing here is that KTLO around data products often is bigger than we’d like to admit. Upstream data changes, data quality issues, modeling problems all create big chunks of time investment.

2. Deprecate the product with a plan and a timeframe. It is really hard to drop something you’ve devoted your time, energy and pride into. However, if there is not an explicit benefit to keeping the product working, or the cost is simply too high, deprecation is probably the right answer. To be clear, deprecation does not mean “just stop paying attention to it”. This means investing the time to wind it down and make business partners and others who rely on the data/model/dashboard/metric aware of why/when you plan to do so.

3. Explain why development of this data product is critical to the existence of the business and invest heavily in it. I know that “existence of the business” sounds dramatic without context. But if you go “too light” in justification, investment in the data product will just sound something like “this is important”. Important isn’t enough. Is this product (e.g. a model, a recommendation engine) going to help keep the business around for the next 5-10 years? Then say so. It isn’t the time to be conservative. There are some products we’ll go all in on, and others for which it is time to wind down.

4. If the data product is a key asset for a downstream team, think about a tax for keeping it working. Even if you think the right thing to do is to wind down a data product, perhaps another team/person in the business depends on it and refuses to go along with your perspective. That’s understandable! Just don’t let this be cost-less for that team. This puts a burden/requirement on your team in terms of headcount/opportunity cost. You should work to be explicit about how their team can compensate you for it. Perhaps it comes in the form of headcount, on-call support, specific dev expertise. There are many options. Just don’t be the team that absorbs all of the responsibility and all of the costs - at some point that becomes untenable.

Listen, I understand that these decisions suck. No matter what happens, we are making hard tradeoffs. My real hope is that by thinking through your options, you can measure opportunity costs, speak to what it takes to keep things moving forward, and make hard decisions to wind things down when/if it is necessary.

For all of those experiencing the brutality of the job market or pain in your current role, I just want to say I am sorry. If I can be of help in any way, feel free to reach out and I’ll do my best to connect you to opportunities/people or any other action I can take.

---