We need two to tango, with Mat Raczyński.

This one's properly wholesome. Bhav brings on an old Gusto teammate of his, Mat Raczyński, who's now running a small data team as the Head of Data @ Paired (relationship care app). Along the way we get into being an IC vs. leading a team, three different flavours of data leadership, a proper rabbit hole on MMPs and attribution, borrowing Duolingo's Markov chain playbook to find Paired's real north star metrics, and the delightfully cursed problem of attributing a subscription when your product genuinely needs two people to convert. No conclusions were reached, as per usual. That's still the point.

Links

Form: Confess, rant, or simply grass someone up (anonymously ofc) using this form!

LinkedIn: Mat Raczyński @ Paired

LinkedIn: Daniel Perry-Reed @ Kickflip Analytics

LinkedIn: Bhavik Patel @ Huel

Podcast: Comment, subscribe and listen to other episodes over on analyticsunscripted.com

Chapters

00:00 — Cold open: no spoilers. Bhav teases the elegant sample-size trick he and Mat cracked years ago — the story stays locked until the end.

02:02 — How Bhav "recruited" Mat. The egg-conference origin story, gatecrashing meetups, and Bhav's talent for loitering near people's laptops.

03:09 — Mat's origin story. Physics degree, Busuu, Babylon, three years at Gusto with Bhav, and now leading a (small) team at Paired.

05:21 — IC life vs. leading a team. Juggling hands-on analytics work with mentoring when your whole team is two people.

07:57 — Can you even "lead" in a service industry? Dan poses the big one: what leadership looks like in a function that only exists to power other people's decisions.

09:47 — Three data archetypes (and yet another pyramid). Data provider vs. decision-systems builder vs. commercial partner, plus Mat's CEO's usefulness pyramid and Dan's narration/commentary/insight framework, all colliding at once.

15:35 — Down the attribution rabbit hole. Chasing what Meta's actually doing with the budget, and a crash course in MMPs (Mobile Measurement Partners) for the app-only crowd.

18:12 — Markov chains and Paired's real north star. Borrowing Duolingo's playbook to find the one lever — staying on a streak — that actually moves the metric that matters.

22:33 — "We need two to tango." Attribution gets brutal when the app needs two people to convert, and every platform is built for a user ID, not a couple.

28:39 — Keeping metrics simple, and what keeps Mat learning. A closing note on building a product people say saved their marriage.

Tags

analytics · data analytics · data leadership · marketing attribution · MMP · Markov chains · business intelligence · data analyst · analytics podcast · AI · startups