Data Product Roles Require Specialization: Here's How To Do It

I did a search yesterday for “data product manager” job ads on LinkedIn. The vast majority of them referenced product background and an interest or basic background in data. I read through at least 20 job ads and came away with the following assessment: companies want a generalized data product manager who can work on anything associated with data. Here’s the reality: data products might have some general principles that apply regardless of customer and domain, but to have a truly impactful product, data product managers must have specific domain knowledge about their area.

Now, that doesn’t mean that data product managers must be world experts in their space. They don’t need to be a rockstar in product, data and every other domain known to those who work in data. However, they need to know two things quite well: 1) their customers’ workflow and needs and 2) how to build a product to improve the customers’ experience.

Is it possible to be effective in this space coming in with zero knowledge of either one? Maybe. I’ve seen examples of strong data scientists and engineers who quickly pick up on how to shape a product for particular customers. However, they have a distinct advantage - domain knowledge. That domain knowledge doesn’t come for free. It also does not come quickly.

Let’s use an example that is fairly common across product focused companies - product experimentation. Typically, product experimentation is used to evaluate the impact of product changes on user behavior, captured through a number of metrics that are important to the business domain. On the surface, running an A/B test seems fairly simple. Create a control and an exposure group, evaluate the difference between those groups through a primary metric, and boom, product impact can be evaluated.

Yet it is not that easy. Why? Running a high quality experiment means getting many things right - from metric definition to cohort allocation to logging to dealing with variance issues. A seemingly innocuous ask from a customer such as “I’d like to speed up an experiment” requires a deep level of sophistication and understanding from a product manager. There is no “easy button” to hit. The product design depends on solid scientific foundations and those foundations require specialization.

If specialization matters (and I fundamentally believe it does) that doesn’t solve the problem of how to hire data product managers who have enough of it to create a successful product. I don’t have perfect answers here, but I hope these ideas spur your own thinking about how to 1) identify what specializations you need and 2) create job descriptions and openings that appeal to people with those specializations.

1. Avoid generic ads for data product managers. I know it is tempting to create a job ad that fits nicely within your product management framework at your company. In fact, HR might push hard for it so that pay bands and leveling make sense. They might have templates that you are required to use for product management roles. As much as possible, avoid the generic requirements and say what you need. If you are working on experimentation, write the job so that experimentation experts will pay attention. If you need a metrics expert, say that. Don’t fall back on comfortable, generic language. Be specific.

2. Don’t feel compelled to make a “data product” role a product management role. If you are getting blocked by your company or you feel like you are fitting a square peg in a round hole, so to speak, be open to positioning the role in a different way. Maybe you describe it as a platform role (data platform, experimentation platform, etc). Maybe you create the role for a data scientist with a particular expertise or interest in product. If you don’t feel like you can make a successful hire or role with a product title, then don’t do it. Change it up.

3. Proactively recruit and position the role for highly qualified candidates. I can almost guarantee that in this job market, you won’t magically have a pipeline full of people who meet the specific needs for what you want. You are going to need to find them. Whether it is just you or a recruiting team, start looking early and expect a search to take time. Don’t be afraid to write the role description to fit a specific type of profile. Sure, it might be scary to have a smaller top of funnel volume than normal (e.g. for a product management role). But your search is about quality and finding a specific skill-set, not creating a pipeline of generic people.

4. Ask for advice outside of your organization. Don’t just create a role that fits the parameters that you agree on internally. Go ask people who are experts in a domain area. Pay them if you need to. Their input on whether the role you are describing meets the needs of and bar for candidates in industry will help you a lot. I’ve worked with a number of hiring committees who meet weekly, frustrated by the lack of candidates coming in. Part of the problem is they aren’t evaluating what they could do better. Ask experts. Ask those in industry. People are far more willing to help than you might expect.

5. Have a backup plan and don’t be afraid to pause a search. The job market right now is one of the strangest and most competitive I’ve ever experienced. Many companies are making desperation hires and choices they wouldn’t in other situations. If you cannot find the right person for a role, have a backup plan. Is status quo tenable at least for a while? Be okay with that. Is there someone internally who can do most of the job for a while? Give them an opportunity. I see many companies that define an important role and once its defined, only feel good once they fill it. The number of data product managers with specific expertise is still small. Patience goes a long way here. Even if that is not popular in the hiring committee discussions.

I’ll be the first to admit that no one has all the answers. I have very few of them. What I do have are ideas that hopefully help you think about what data product roles look like at your company and how you can craft a job description and search that helps you identify the right candidates. A final piece of advice - while it might be tempting to look externally, there’s far more internal talent than you imagine. You just need to search for it.