The Accidental Data Career: How Context Separates Good Teams from Great

‍

Most data careers don't start with a grand plan. They start with someone going on vacation and asking you to babysit an Access database on their laptop.

That's how Cathy Tanimura's journey began at StubHub, where she was hired as a partnership manager, not a data analyst. Today, she's a data leader at Summit Partners, author of "SQL for Data Analysis," and has built data teams at Zynga, Okta, and Strava.

On this episode of Crunching Data, Cathy shares what separates good data teams from great ones, why trust matters more than technology, and how AI is repeating the same hype cycle data went through a decade ago.

If you're building a data team or trying to figure out where AI actually fits in your stack, this conversation cuts through the noise with hard-won wisdom from someone who's been doing this for 20+ years.

Context is everything in prompting and in data teams

The biggest differentiator between good and great data teams isn't technical skill. It's context.

"You can work on numbers all day long, but if they don't have that surrounding sort of story and information about how are they generated, how does that data come into the world, why did somebody collect it, what do they think it means, and then on the other side, what are we going to do with it, how is this going to help somebody make a decision or understand what's going on in the world in a way that they then take action to change it."

Long before attribution models and UTMs became standard tooling, Cathy was already doing the work by hand. At StubHub, she unified radio station spreadsheets, affiliate reports, and early Google Ads data into what she describes as “a kind of original first touch attribution model before that was even a thing.” That instinct, to go upstream, connect the dots, and understand what the data actually means, is the same instinct she sees in the best data people today.

Great data people are curious. They ask why the data exists, who collected it, and what decisions will change because of it. Without that context, you're just pushing bits around.

Trust is the foundation everything else is built on

You can have the fanciest tools in the world, but if your stakeholders don't trust your data, you have nothing.

"In data trust is such a foundation that the data has to be accurate, it has to be trustworthy, it has to be vetted. And so that is really the most important thing you have within an organization."

When something breaks (and it will), transparency is your only play. Cathy's approach: explain exactly what happened, why it happened, and how you're preventing it from happening again. Take responsibility. People respect that more than perfection.

Walk a mile in everyone else's shoes

Cathy's career success didn't come from being the best coder or the most technical analyst. It came from empathy.

She talks to marketers about their campaigns, salespeople about their pipelines, even lawyers about GDPR concerns. Understanding what someone actually needs to do their job better (and delivering data in a format that fits their workflow) builds trust and makes you indispensable.

The people who were entry-level when she started? They're running companies now. Help everyone. The karma pays off.

AI is where data was a decade ago: hyped but misunderstood

Companies treat AI the same way they treated "big data" ten years ago: hire someone, expect magic, then ask "okay, now what?"

Cathy's advice: think strategically about where AI actually provides an advantage in speed, quality, or cost. Don't just sprinkle it on everything and hope it works. Find the specific brittle processes (like those 200-line case statements) where an LLM can make things more adaptable.

It won't be magic. It'll be targeted improvements in the right places.

Data quality tooling is underrated and underinvested

Every data person has spent countless hours debugging reports because "the numbers are wrong." Cathy's lived it repeatedly.

Companies will spend hundreds of thousands on salaries to manually maintain data lineage and quality checks, but balk at spending $10K-$100K on tools that automate it.

"I don't know any data person who hasn't spent many, many hours debugging a thing because somebody calls you and says, the numbers in the report are wrong."

Having visibility into the entire chain (from source system to dashboard) saves those painful debugging sessions. It's an area that's historically been hard to justify investment in, but the cost of not investing is enormous.

Pick tools your team knows (or tools worth learning)

When it comes to building a data stack, Cathy takes a pragmatic approach: look at the latest tools, but also consider what your team already knows.

If a new tool delivers significantly more value, invest in training your team. But sometimes the tool that "gets the job done" and people already know is the right choice. We live in a time of plenty. Lots of great options exist. The hard part is choosing what fits your team's needs.

Watch the full episode here.

‍

• The Myth of the 100% Data-Driven Company with Shiv Malhotra (Tanium)
• The Speed Advantage: Why Velocity Outpaces Over-engineering with Ryan Delgado, Ramp
• Context, Curiosity, and Cutting Through the Noise with Qun Wei (Lemonade)

‍