Product Market Misfit

And the still elusive Data Executive

Benn posed the question of whether the accepted model of running and building data teams was really just for the first few archetypal strong data teams… will it ever cross the chasm? Will it be as ubiquitous as teams like Marketing and Finance are in every organisation?

Do data teams have product-market fit?

Reading his post inspired me to write this one, which is partly a response and also on a tangent from it… is this data philosophy correspondence in the ether? 🤓 Katie » David » Benn » David… who will respond next?

We can see the problems with the ideas taken from these data hero companies. It’s why a bunch of other companies started hiring data scientists when they had no data for them to work with, resulting in recovering data scientists.

Data was branded the new oil, with companies searching for engineers with which to drill it… the tech stack was nearly as complicated as a real oil rig, resulting in Hadoop refugees.

To this day, many companies still struggle and fail at data engineering, which is why so much money has been poured into solving this space. It peaked with an open-source ELT tool being made a unicorn, after < 2 years of existence and < $1m ARR! It’s why so many of us who used to be analysts, myself included, have delved into data and analytics engineering for the last few years - some of us founding those new companies.

Why did this happen?

Execs moving around and taking the tool they had at the big tech company (roughly a pneumatic sledgehammer) and applying it at their new companies, which had some picture hooks to bang in.

Photo by Adam Sherez on Unsplash

Copycats, in the few existing senior data leaders outside of big tech, deciding that this must be the way to go. “Let’s build the data lake”, which soon turned into data toxic waste dumps.

People bandying about terms like AI and Data Mesh etc, which aren’t really well understood by most people outside of data, let alone those inside, and offering up promises which are yet to be fulfilled.

People being frustrated with the pre Big Data stack tooling, which required a huge amount of effort just to get basic numbers out without the systems falling over. I used to work at a company probably worth tens of billions now, that spent 1 week every month doing month-end reporting on MS SQL Server. (Even after my best effort at optimisations on our end, I only managed to get that down to 3 days).

Losing any sight and all appreciation of what impact data could realistically have!

‘My investment in data is bigger than yours’ exec mentality.

All of these things resulted in a disconnect between value driven by data and investment in it. ROI declined and our stock price dropped like some maniac was our CEO.

Self-serve - was it too general? It should always have been an 80/20 hybrid model, with a lot of product-led sales involved.

Multi-year data lake projects left stakeholders feeling like they paid early for a car and waited two years to be given an engine, steering wheel and the lid of a trunk. That same multi-billion dollar company I mentioned above made this mistake - in the meantime, I was there leading a pricing team, being given financial targets to hit: we hit them year after year, we got to set commercial strategies and policy to help us hit these and help our division hit our goals. Even though we were also a data team by another name, we were able to be nimble and move at least an order of magnitude faster to meet stakeholder needs than the central team focused on the data lake project. I’ve written about this before, so I won’t cover it too much.

Data in the top, money out the bottom

…we don’t yet have the influence within our companies that we know we could.

I didn’t have to force my way into impactful settings here. Benn’s post made me remember that this was probably the last time I felt consistently, year after year, commercially impactful in data and it only fell apart because of political tectonic plates moving, with my team being rolled around on top… my stakeholders didn’t want me to leave. Incidentally, my team wasn’t even branded a data team, it was a pricing team - it just so happened that the core discipline empowering us was data.

That’s not to say that everything I’ve done since hasn’t been valuable - it has been at times, but not consistently so. Why?

I think it’s because it wasn’t as focused. We had a relatively small but rich data model that we used in my pricing team, to allow us to understand the past, forecast future expectations and scenario model accordingly. We didn’t worry about adding new data sources all the time. We knew exactly what we needed and we focused on just these things and applying our best rigour.

Many data teams out there don’t have this: what their stakeholders are trying to do is somewhat nebulous to them, so they try to boil the sea, getting every possibly useful piece of data they need into their data lakes/warehouses and then having to test and model it. They spend most of their time trying to deal with tens of pipelines and hundreds of dbt models, but very little on getting deep into their business. They don’t spend enough time focused on specific business problems to solve.

I must admit to falling into this trap on occasion, and this is one of the problems of taking a step too far in applying engineering principles to data. It works well for data and analytics engineering to a greater level, but it doesn’t work so well for analytics.

Analytics from a purist point of view (stripping away the data and analytics engineering elements, that analysts often have to do) isn’t that much like software engineering at all 😲. At times it’s like journalism, at times it’s like triage and, yes, it’s like management consulting sometimes, too.

The big tech companies we’ve borrowed data management principles from have largely had data and ML engineering problems and solved them in this way. It’s highly likely their “product” is not suitable for other companies.

We have adapted to some decent halfway houses, like the hub and spoke model, where senior analysts typically are embedded into an area to get that focus. But this hasn’t translated into a solution for the missing analytics executive, either. I still feel that, even with the best product that stakeholders want to buy, the incentives to allow that analytics executive to exist and operate aren’t necessarily there. Just because you find a great product that you want subscribe to, doesn’t mean that you necessarily want to take advice from its producer or share credit with them for anything you achieve by using it. After all, why would they?

Let’s go with the primate/4 year old behaviour often referenced to be at play in boardrooms: based purely on self-interest, it makes more sense to use the great data product and exclude the data leader from the table; why share the credit or the power? One more voice at the table means yours is heard less. I’m not saying all of senior leadership are this self-interested, but I am saying about half are in the “average company” in the real world, meaning that the specific company you work at either has mostly self-interested leadership or mostly not - the middle doesn’t hold out well here.

---

If we had product-market fit—if stakeholders were enthusiastic buyers of the services we offer—would we still be kept out of the influential rooms we want to be in? Would we have to fight to be heard? Would we have to constantly remind people why our work is valuable?

No. Products that have product-market fit are bought, not sold. And as people’s responses to posts like those from Katie and Erik show, most data teams still have to do an awful lot of selling. 

I think both of these possibilities can occur at the same time. I’ve run a hub and spoke data org before, where stakeholders were such “enthusiastic buyers” of the services that they gave up their own budget to have embedded analysts and analytics engineers who would work according to the toolset and method set centrally. You could argue that this was just a queue-jump ticket at the theme park, but I’d argue that’s even more evidence of value.

These same stakeholders also wanted additional support from central resource too. They started to ask for deeper pieces of work to validate strategy at CXO level - they were even willing to submit decent quality Jira tickets, by and large, to get what they wanted. Demand outstripped Supply and Supply wasn’t cheap.

Nonetheless, they didn’t want there to be data representation at the senior level (by this time Data was larger than HR, Finance and Legal combined). Whether this was for a self-interested reason or due to lack of understanding… I will never know.

What if we’re not prophets in the wilderness, but salespeople selling a lemon? What if the problem isn’t that the market doesn’t understand what we’re offering—it’s that they do, and they don’t want it?

I guess what I’m trying to say is that we’re not even having to sell that hard to exponentially increase spend on data tooling, headcount and output (or at least we weren’t till this current economic change)… but we’re trying and failing to sell the idea that our best and most senior people should be helping to run the company alongside their peers.

We still need to be valued for what we bring to the table as data folks, just like Finance, Product and Marketing are. They aren’t purely brought into the room because of their knowledge of their domains, like Capos to the Boss: their own opinions, shaped by their disciplines, are also valued. Our opinions, shaped by our knowledge of the data discipline, are equally valuable.

Just like Finance won’t have information for every possible scenario, but are still heard where they don’t, Data should be, too. In fact, our instincts where there isn’t complete data, or any data at all, can be very valuable. Because often, we’re talking about venturing into new spaces, and hoping to be able to measure our success there afterwards; data folks are great at knowing what’s possible and what needs to be done to achieve this. Us not being present at the table is why few projects ever properly estimate how much data work or resource is needed.

Often, tech leaders are cited as the leaders at the table to represent data, but very often they just don’t understand the discipline properly. They have the mindset of an architect of machines, and are often so pushed to handle the tech part of what they cover that there is no way they could represent data well, too.

CTOs often have nearly half of all headcount report into them. If they have data too, then what is most likely more than half of the company will be represented by one voice at the table…. it just doesn’t work from a representation point of view. Legal, HR and Finance often each have their own representation and they, combined, are can be smaller than Data. Analytics is particularly poorly represented by a CTO, as it’s the furthest away from engineering in Data.

I do still believe that Data needs a seat at the table, that we will be a valuable asset there and a source of competitive advantage. I’ve wrestled with the idea that perhaps we could learn another discipline in order to take their seat… but our discipline is at least as deep and nuanced as anyone else’s there. Maybe we could have more specialised roles like “Pricing” or “RevOps” which could be particularly helpful in our orgs. It’s never a bad idea for us to get alongside our customers and focus harder on solving their problems, and less on things that don’t drive discernible value for them. Hopefully, the last few years focused on Data Engineering will empower this.

At the same time, both learning another discipline to progress or having specialised but obscure titles are a compromise on what I really believe… Maybe a more forceful approach is needed; perhaps we should refuse to take a role where there isn’t Data representation in senior leadership. Let’s get our pointy elbows out.

We’ll have a thread on whether the current style of data teams needs to be rethought