Analytics Unscripted

Episodes

• 
  We need two to tango, with Mat Raczyński. 05.08.2026 32m

  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.LinksForm: Confess, rant, or simply grass someone up (anonymously ofc) using this form!LinkedIn: Mat Raczyński @ PairedLinkedIn: Daniel Perry-Reed @ Kickflip AnalyticsLinkedIn: Bhavik Patel @ HuelPodcast: Comment, subscribe and listen to other episodes over on analyticsunscripted.comChapters00:00 — 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.Tagsanalytics · data analytics · data leadership · marketing attribution · MMP · Markov chains · business intelligence · data analyst · analytics podcast · AI · startups This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.analyticsunscripted.com
• 
  Stick that in your pipe, with Rupali Singhal. 29.07.2026 38m

  This might be the most "unscripted" episode of Analytics Unscripted yet. Bhav WhatsApp'd Rupali Singhal ten minutes before we hit record and asked if she fancied joining us. She said yes, and thank goodness, because Rupali leads marketing analytics for Twilio's international regions and runs their global MMM programme, and she very much knows her stuff.We get into what media mix modelling actually is versus multi-touch attribution, why Sky swore by MMM while a gaming company like King still can't get it centre stage, whether marketers should really be paying an agency 200k a year to build one, and the slightly uncomfortable realisation that most models (MMM included) mostly reward direct response, brand-building be damned. We also go down a B2B attribution rabbit hole, talk about why finance is secretly the most important relationship in all of this, and Dan finally forces Rupali to define the word "analytics." No conclusions were reached. As is our way.LinksForm: Confess, rant, or simply grass someone up (anonymously ofc) using this form!LinkedIn: Rupali Singhal @ TwillioLinkedIn: Daniel Perry-Reed @ Kickflip AnalyticsLinkedIn: Bhavik Patel @ HuelPodcast: Comment, subscribe and listen to other episodes over on analyticsunscripted.comChapters00:00 — Welcome to the least planned guest booking in podcast history.02:59 — Who is Rupali (and what actually is MMM)? Twelve years in marketing analytics, running the global MMM programme at Twilio, and the cleanest "unscripted" answer we've ever had.06:06 — Dan gets blunt: is MMM genuinely useful, or just an academic project clever people build for themselves?07:09 — Sky vs. King: why one business lived and died by MMM and the other still can't get it centre stage.17:35 — Why telling a marketer "we're moving your budget" feels like a personal attack (even when it really isn't).19:33 — Don't pay an agency 200k a year: building MMM in-house, and Dan volunteering to do it for a discount.22:38 — Bhav puts his anti-MMM hat on, and we go looking for the brand-awareness blind spot no model can see.26:08 — B2B attribution chaos: when your "customer journey" is actually five different job titles at the same company.33:12 — Career advice, LinkedIn as a slightly controversial learning tool, and why finance is the relationship that actually gets MMM adopted.36:39 — The last-question tradition: what actually is "analytics"? Rupali's answer, and Bhav's mic-drop for the qual researchers.Tagsanalytics · data analytics · artificial intelligence · business intelligence · data analyst · analytics podcast · AI · marketing analytics · media mix modelling · multi-touch attribution · B2B marketing This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.analyticsunscripted.com
• 
  We love the smell of our own farts, with Juliana Jackson. 22.07.2026 40m

  Bhav and I finally let a guest into the room and it had to be Juliana Jackson - friend of the show, friend of ours and AI strategy and growth director @ Jellyfish. No prep, no questions sent over, no editor to tidy it up afterwards. Exactly how we like it.We get into whether any of us really work in analytics anymore, and why nobody can agree what the word "analytics" even means. Juliana makes the case that folks have moved to Reddit, YouTube and TikTok, that the one thing a brand actually controls is the creative asset (no, not cookies Bhav), and walks us through pre-flight testing, ABCD and decomposing an ad before it ever goes live. Bhav pushes back on whether the algorithms are quietly squishing creativity and making us all regress to the mean, and we land on the only thing everyone agreed on - nobody has time for your dashboard, and we all love the smell of our own farts.LinksForm: Confess, rant, or simply grass someone up (anonymously ofc) using this form!LinkedIn: Juliana Jackson @ JellyfishLinkedIn: Daniel Perry-Reed @ Kickflip AnalyticsLinkedIn: Bhavik Patel @ HuelPodcast: Comment, subscribe and listen to other episodes over on analyticsunscripted.comChapters00:00 — Welcome to the first-ever guest episode. What "unscripted" means when there's a third person in the room, and a promise to solve nothing.02:06 — Enter Juliana. Thirty seconds in and she's already plugging her own podcast. A very entrepreneurial guest.03:19 — "Do any of us really work in data anymore?" Fancy new job titles, and why nobody agrees what "analytics" even means.05:22 — Top-down AI mandates. FOMO, the propaganda machine, and everyone trying very hard to look busy.07:42 — Efficiency vs effectiveness. On bolting agents onto dashboards nobody asked for, and asking "so what?"10:54 — The fish are on Reddit now. Using AI to read the qualitative stuff at scale, instead of chasing people to your website.15:11 — The one thing a brand actually controls. Bhav guesses cookies. It's not cookies — it's the creative asset.20:38 — Pre-flight, explained. ABCD, response surfaces, and decomposing an ad before you spend 100k on it.28:07 — Are algorithms killing creativity? Regressing to the mean, leading with feeling, and a not-a-plug for Huel.37:26 — Nobody cares about your dashboard. The TL;DR era, and why we all love the smell of our own farts.Tagsanalytics · data analytics · artificial intelligence · marketing analytics · creative testing · advertising · media planning · YouTube · CRO · data analyst · analytics podcast · AI This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.analyticsunscripted.com
• 
  Fish where the fish are. 15.07.2026 35m

  OverviewWe're back! After a year away from the mic, Bav and I finally hit record on episode one of Analytics Unscripted - the podcast where we don't prep, don't script, and don't have an editor to save us. No agenda, just two data people falling straight into the deep end of a conversation about AI, the future of the analyst, and whether the humble dashboard is quietly dying.We get into whether AI is really coming for our jobs, why "self-serve analytics" was the great white whale of Bav's career, the beautiful chaos of "okay thanks" questions, one-dimensional vs. multi-dimensional analysis, and my little narration → commentary → insight framework that I forced everyone to listen to. We also somehow talk about vanity metrics and why we're both a bit needy. No conclusions were reached. That was the point.LinksForm: Confess, rant, or simply grass someone up (anonymously ofc) using this form!LinkedIn: Daniel Perry-Reed @ Kickflip AnalyticsLinkedIn: Bhavik Patel @ HuelPodcast: Comment, subscribe and listen to other episodes over on analyticsuncscipted.comChapters00:00 — Welcome to the inaugural chaos. What "unscripted" actually means, and a promise that we won't solve anything.01:47 — A whole year since The Measure Pod. Where the time went, why there's no editor, and a moment of silence for Will.05:19 — Plugs, updates & career whiplash. I ramble about Kickflip Analytics (and forget to name it), Bav goes in-house.08:48 — Is AI coming for the analyst? Straight in with the heavy question, no messing around.10:23 — Self-serve analytics, reframed. Bav's "biggest failure" wasn't a failure — the ceiling was just really low. Plus: fish where the fish are.13:36 — FONCU, the "non-technical" myth & AI use cases. Fear of Not Keeping Up beats FOMO, and why automating the boring admin is the real win.17:02 — "Okay thanks" questions vs real analysis. One-dimensional vs multi-dimensional, and me poking Bav about it.22:04 — Are dashboards dead yet? Where dashboards actually belong now that we've got LLMs, MCPs and Slack pinging us.26:05 — Narration, commentary, insight. My three-tier framework, analytics debt, and why I reject text boxes in dashboards.31:05 — Wrapping up: vanity metrics & the secret confession form. Likes vs a nice LinkedIn message, and why we're both needy.Tagsanalytics · data analytics · artificial intelligence · data teams · self-serve analytics · dashboards · business intelligence · data analyst · analytics podcast · AI · analysis This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.analyticsunscripted.com