Humans of Martech

What's up everyone, today we have the pleasure of sitting down with Gary Amaral, marketing leader at Docket.

(00:00) - Intro

(01:15) - In This Episode

(04:35) - What Lead Scoring Was Really Hired to Do

(12:13) - Why Lead Scoring Models Turn Into Black Boxes

(16:04) - What Is an Agent Qualified Lead

(24:02) - How Agentic Qualification Goes Beyond AI Scoring

(27:48) - What AI Agents Do That Drift Style Chatbots Never Could

(34:35) - Why Buyers Tell AI Agents What They Hide From Sales Reps

(40:05) - Who Needs an AI Agent on Their Website

(42:48) - How to Keep the AQL From Becoming the Next Vanity Metric

(50:11) - How to Introduce AQLs Without Breaking Your Forecasting

(56:30) - Will AI Agents Replace Intent Data and Marketing Automation

(59:41) - Why Work Life Balance Is the Wrong Goal

Summary: Gary built lead scoring so ambitious it crashed a Marketo pod and took dozens of companies down with it, then co-founded a scoring company, sold it, and switched sides of the argument. Now he says the whole category rests on guesswork: scores infer what buyers want, while an AI agent can just ask them. He explains why every rep would trade 20 scored leads for one hand-raiser, how Docket's agent Aura lifted engagement 8X by moving into the homepage hero, and why he refuses to be measured on MQLs anywhere he works. Along the way you get a Waymo analogy, a Meat Loaf methodology, and a bet that agents will eventually replace your bloated marketing automation platform. By the end you'll know exactly how to pilot AQLs without giving your CFO a forecasting heart attack.About Gary Amaral
Gary Amaral has spent his career on the question lead scoring was invented to answer: which buyers actually deserve a sales rep's time. He chased it at BlackBerry, then at Hootsuite, where the demand center his team built included a scoring system so complex it took down an entire Marketo pod. That week of hand-routing leads became the genesis of Breadcrumbs, the lead scoring company he co-founded and later saw acquired by MadKudu.
Today Gary leads marketing at Docket, the AI agent platform behind Aura and the agent qualified lead, where he is making the case that asking buyers what they want beats inferring it from their clicks.
What Lead Scoring Was Really Hired to Do
Every marketing ops team has a scoring model somewhere in their instance that nobody fully trusts. Points for a webinar, points for a pricing page, a threshold someone picked in a meeting 3 years ago. Strip away the arithmetic and the model was only ever hired to answer one question: who should sales reach out to right now?
Gary has been chasing that question since the BlackBerry days. By the time he landed at Hootsuite, lead scoring had grown into something much bigger, a full demand center with scoring at its core. And the system his team built grew so elaborate that it broke more than his own instance.
He then spent a week manually sorting and routing leads while Marketo dug itself out. Imagine the tedium: a demand leader hand-triaging spreadsheet rows because the machine built to do it had flattened the neighborhood. That week became the genesis of Breadcrumbs, the scoring company he went on to co-found. His pitch was simple. He describes himself as "just an average guy" without a PhD in statistics, carrying years of experience feeding sales teams quality leads. Recency, frequency, fit, and intent, combined in a way that tells you where to send a lead and which seller should get it. Breadcrumbs set out to bring that sophistication to everyone else.
Then the market moved. Gary names 3 headwinds that hit the company:
AI-native competitors arrived, and if you squinted you could already see how agents would change qualification entirely, Legacy platforms finally invested, with HubSpot shipping scoring that was still not great but good enough for teams already paying for it, Consolidation swept the category as vendors hunted for efficiencies and expanded offerings
That last headwind doubled as an exit, and MadKudu acquired Breadcrumbs. The pattern says something uncomfortable about martech point solutions: when your product's output is a number that feeds another team's queue, you are one platform release away from becoming a feature.
Key takeaway: Audit your scoring model against the job it was hired to do. List the last 20 leads it routed to sales and ask a rep which ones deserved the outreach. If the rep's answer and the model's answer disagree on more than a handful, the model is scoring activity, and you should rebuild the weights around closed outcomes instead of page values someone negotiated in a meeting.
What Made Advanced Lead Scoring Different From Point Systems
The classic model runs on decree. Somebody decides the pricing page is worth 20 points and a blog post is worth 10, usually during a horse-trading session where sales insists that anyone who looks a certain way and clicks a certain thing must be hot. Marketing goes along with it because keeping the peace is cheaper than arguing. The weights encode office politics, and everyone quietly knows it.
Advanced scoring, the kind Breadcrumbs built, flipped the direction of authority. Instead of a human declaring what matters, the model watched signals over time, tied them to actual outcomes, and surfaced the combinations of behavior that genuinely preceded a purchase. More data sources feeding the model, and math deciding the weights. Harder to explain, Gary admits, but honest in a way the point system never was.
That last part deserves more attention than it got at the time. Because the tool learned from outcomes, you could point it at any outcome you cared about. A model for webinar attendance, another for churn risk, a third for upsell readiness. The same machinery, aimed wherever the business needed a prediction, while most teams were still arguing about whether an ebook download deserved 5 points or 15.
Key takeaway: Pick one outcome beyond lead routing and build a scored view of it this quarter. Churn and upsell are the natural candidates because the outcome data already sits in your CRM. Pull 12 months of closed outcomes, find the 3 behaviors that most often preceded them, and hand that shortlist to the team that owns the motion.
Why Lead Scoring Models Turn Into Black Boxes
If you have ever stared at the stars and flames in Marketo Sales Insight, or an intent dashboard announcing that an account is "surging," and wondered what any of it actually means, you have met the transparency problem. Darrell raised it directly on the show: as a user, he often had no idea what was going on inside these tools and no way to change it. Gary's answer starts with an admission. Breadcrumbs never fully solved this either.
His best explanation for why the totems exist comes from riding Waymos in Austin. The first driverless ride freaks you out. By the third, you barely notice, and Gary credits the giant screen in the front of the Jag that renders the cars and pedestrians the vehicle sees.
The flames work the same way. They translate a complex system into something a human can glance at and trust. The trouble is that really smart marketers refuse to stop at trust. They want to know what drives the representation, and then they want to override it. So every scoring vendor ends up refereeing between what the algorithm proved from historical data and what the marketer feels should count, and Gary says there is some validity on both sides of that fight.
The model only knows where the business has been. A 10-year-old company that trains on all 10 years of data is muddying the waters, because who you sold to in year one barely resembles who you sell to now. And the moment you make a big deci...

What's up everyone, today we have the pleasure of sitting down with Christina Garnett, Fractional Chief Customer and Communications Officer at Neuemotion.

(00:00) - audio-Christina

(01:19) - In This Episode

(05:37) - Why Your Tech Stack Cannot Fix Siloed Customer Teams

(11:10) - Why Getting the Name Right No Longer Counts as Personalization

(18:08) - Why A/B Testing Programs Ignore Qualitative Customer Data

(24:14) - How Brands Build Core Memories Customers Cannot Buy

(29:55) - How B2B Brands Build Core Memories Without a Budget

(35:46) - What to Do When Transactional, Behavioral and Feedback Data Disagree

(42:12) - Why Spotify Wrapped Works and What Gen AI Took Away

(49:17) - Why Personalization Feels Predatory Even When You Have Consent

(53:58) - Where Influence Ends and Manipulation Starts in Marketing

(58:57) - Why Audiences Are Going Back to the Humanities in the AI Era

Summary: Christina spent 5 years teaching math before she ever wrote a subject line, and it shows in how ruthlessly she takes apart the personalization playbook. Getting someone's first name right used to be impressive and now it just proves you didn't get fired, which leaves marketers chasing hooks and tricks while customers quietly learn to read the pattern. She makes the case for core memories a brand cannot buy, walks through what actually happened when Spotify Wrapped got the AI treatment, and introduces earned context, the idea that having someone's data and having the right to use it are 2 completely different things. Along the way she calls CX and brand the same job, refuses to bet her life on NPS, and explains why the B2B move nobody is making costs almost nothing. If you have ever shipped a perfectly segmented campaign that landed with a thud, this one will tell you why.About Christina Garnett
Christina Garnett is a fractional Chief Customer and Communications Officer at Neuemotion and the founder and principal at Pocket CCO, where she builds customer trust systems for agencies and SaaS companies. She's the author of Transforming Customer-Brand Relationships, which won the Independent Press Award and the International Impact Book Award and was a Foreword INDIES finalist, and she's currently writing a second book built on the Customer Trust Equation she created.
Before going independent she built customer advocacy at HubSpot, growing the HubFans program from 0 to 37,000 members and more than 4,000 active advocates, and she led social and customer strategy at global scale through ICUC. She's a contributing writer at Campaign US, was previously a byline at Adweek, and she started her career as a teacher.
Why Your Tech Stack Cannot Fix Siloed Customer Teams
Every few years a software category shows up promising to end the silo problem. The CRM was going to do it. Then the customer data platform. Then the customer 360, the composable stack, and whatever the warehouse-native vendors are calling it this quarter. Marketing, support, social, and CX all get wired into the same tables, and everyone assumes shared data will produce shared behavior. Buy the integration, get the alignment.
Christina has been on enough of those teams to know how that ends.
She came into marketing sideways, an English major who taught math for 5 years, including a stretch at a school built for kids with learning differences where she trained in the Orton-Gillingham method. That method exists to move information from short-term into long-term memory. Marketers have a word for the same thing and call it brand recall. Her husband told her early on that she should go into advertising because she couldn't watch an ad without rebuilding it in her head, tweaking the copy, cutting the music, fixing whatever was broken. She went and took every course, read every book, and kept the teacher's instinct the whole way through: make the hard thing make sense, find the right words for the room you're in.
Then she sat at basically every table that claims part of the customer, working as a social listening strategist, a social media creator and a social media strategist, and putting time into support, CX, community and advocacy. What she kept finding was the same organizational failure repeating under different logos, and no amount of shared tooling moved it.
Here's the argument that runs under her whole book. Nobody is solving customer experience alone, because nobody is the only person reaching the customer. Support talks to them, so support gets blamed for the relationship. But your UX team is shaping what the customer can and can't do, whether they think of that as CX work or not. Your legal team is writing the fine print the customer reads at the worst possible moment. Finance sets the refund window. All of it builds the relationship, and most of it never appears on anyone's CX dashboard.
The result is a company that's genuinely excellent in one place and quietly hostile in another, and it shows.
That's why the material became a book rather than a framework or a course. A framework asks you to already agree with its premise before you'll use it. Christina wanted the thing she'd been hunting for on shelves and never found, written for the people at all of those different tables at once. Her words on it: "I think we all wind up writing the books that we wish already existed." The structure follows the problem in 3 moves:
Every moving piece that actually builds a brand relationship, most of which sits outside the CX team., Community as the silo killer, the thing that finally forces those teams into the same room., What happens when something lands that you had no way of seeing. A brand crisis, an acquisition by a company your customers hate, or AI arriving and throwing the whole operating model in the air.
That third one is where most companies are living right now, and it's the least rehearsed. The martech buying cycle keeps funding integration projects because integration is something you can purchase, and cross-functional accountability is something you have to negotiate. Until one person owns the customer relationship end to end on paper, a new platform just gives 4 teams a shared place to file work they still refuse to coordinate.
Key takeaway: List every team that shapes your customer's experience, including the ones that never speak to a customer. Legal writes the cancellation language, UX decides how buried the downgrade button is, finance sets the refund window, and none of them get counted. Put all of them on one page and mark who currently owns the customer's experience of each moment. The blanks are your real CX roadmap, and no platform purchase will fill them in for you.
Why Getting the Name Right No Longer Counts as Personalization
Accuracy in a lifecycle program has never been cheaper. You can segment by behavior, by lifecycle stage, by predicted churn, by what someone hovered over last Tuesday, and the tooling will do all of it on a mid-tier plan. Ask a marketer whether their email is personalized and they'll show you the merge fields and the branching logic and say yes.
Ask the person receiving it and you get a different answer. Christina's position is that martech made accuracy so cheap that creativity started to look like a nice-to-have, and customers noticed long before marketers did.
Her frame for the whole industry is the Jeff Goldblum line from Jurassic Park, the one about scientists so preoccupied with whether they could that they never stopped to ask whether they should. We're living inside that meme, she says, running on a lot of we could so we can. And the cost shows up somewhere marketers rarely look, which is in what we've trained people to expect.
Conditioning is real, and it doesn't require a marketing degree. Somebody w...

What's up everyone, today we have the pleasure of sitting down with Aditya Vempaty, VP of Marketing at Coram AI.

(00:00) - Intro

(01:07) - In This Episode

(05:05) - Why Expansion Revenue Beats New Logos When Channels Are Saturated

(06:07) - Why Customer Empathy Means Knowing What Gets Your Buyer Promoted

(12:38) - How to Tell If Your Case Study Makes the Customer the Hero

(18:16) - Why Single Channel Attribution Hides the Journey That Closed the Deal

(22:00) - How to Find Real Customer Problems by Paying for 30 Minute Interviews

(30:29) - The 3 Levels That Connect Marketing Strategy to Tactical Execution

(36:32) - Why Implementation Speed Is the Real Answer to a Migration Objection

(42:12) - The ABM Play You Can Build From Your Closed Won List Today

(48:46) - How Marketing Leaders Get a Team to Actually Adopt an AI Tool

(53:26) - How Marketing Leaders Decide What Deserves Their Energy

Summary: Aditya has built marketing functions and named categories from nothing at Synthego, Amplitude and Nutanix, and he shows up here to argue that the loud top of funnel work is now the cheaper half of the job. He takes apart the make your customer the hero cliche with a test you can run on any case study in about 4 seconds, then explains why he pays customers $100 for 30 minutes before he will even look at the product he was hired to market. There's a story about harvesting a competitor's free tier badge off customer websites, a telco onboarded in 6 weeks, and a VP of marketing spending 15 minutes making a poster so his team would actually use AI. He also has a very short answer for what your ChatGPT persona document is worth. Stick around for the part where 95% product parity turns into the only argument that reliably wins a migration deal.About Aditya Vempaty
Aditya Vempaty is the VP of Marketing at Coram AI, an AI physical security platform running across more than 1,500 locations. Before that he spent 3 years as VP of Marketing at MoEngage, where the company grew from $58M to $100M in ARR. Earlier in his career he built the marketing function and named the category at Synthego, Amplitude and Nutanix, work that spans 2 IPOs and several unicorns.
He also invests as an angel, mentors at First Round Capital, and is known for 2 rules he repeats often: market the problem first, and distribution is the strategy.
Why Expansion Revenue Beats New Logos When Channels Are Saturated
Every marketing team has a number on the board for new logos. Far fewer have one for what happens to a customer 6 months after the contract gets signed. That made sense when acquisition was cheap and a paid dollar reliably bought you a lead. Channels are crowded now, budgets are flat, and the math has moved.
Aditya has spent his career at the loud end of marketing. He helped name the product analytics category at Amplitude. He helped define the agentic customer engagement platform at MoEngage. Those are top-of-funnel plays, the kind where you plant a flag and get the industry to repeat your words back to you. So it's a fair question why the category creation guy now spends his time talking about heroes, lifecycle and retention.
His answer is arithmetic.
For a marketer defending a budget, the second half of that sentence matters more than the first. Cheaper is the obvious part. Easier to show impact is the part that should change how you plan. Every acquisition attribution fight you've ever lost, where the CFO wants to know which of your 6 channels made the lead buy, goes away when the account is already in the building and you can see exactly what they did and didn't adopt.
The shape of the products matters too. Amplitude and MoEngage are both multi-product suites. Customers land on one thing, get value, and then expand into the rest. So the second, third and fourth sale are already sitting inside your customer base, and nobody has to be convinced you exist. Aditya points to HubSpot's version of this, where the funnel is a loop rather than a line, and treats the loop as the actual revenue machine instead of a diagram in a deck.
Here is the uncomfortable version. Most B2B marketing orgs are staffed and measured around the most expensive, least provable part of the job. If expansion is where the margin actually is, the demand gen team should carry a retention or adoption number, and most demand gen teams would fight you hard on that.
Key takeaway: Pull the last 4 quarters of revenue and split it into new logo, expansion and renewal. If expansion is the bigger line and nobody in marketing carries a number against it, claim it before the next planning cycle starts. Pick the one product line your existing customers adopt second and build the campaign for that.
Why Customer Empathy Means Knowing What Gets Your Buyer Promoted
Customer empathy is the most repeated phrase in B2B marketing and the least examined one. It shows up in every kickoff deck, every messaging workshop, every values page. Ask what it produced and you usually get a persona document nobody has opened since it was made.
Aditya spent years on the vendor side of exactly this, at 2 companies that sit in the middle of the funnel where marketing ops and lifecycle people live. He saw how the sausage gets made, and he came out of it with a low opinion of the word.
What replaced it is much more specific and much more useful. Sitting inside those companies taught him that the thing worth understanding about a buyer is their job justification. How does this person prove they were worth their salary this year? How do they get promoted? Your product either helps them say that sentence out loud or it does not.
At Amplitude the answer was retention. Product managers in B2C couldn't track what they were doing and couldn't justify their own existence and investment. The industry was obsessed with mobile acquisition at the time, so the question that actually kept a PM up at night was how to keep the users they had already paid for and how to show the product was giving those users value. Amplitude built the whole marketing motion around that metric because that metric was the PM's career.
At MoEngage the pain had moved. Retention was table stakes by then and everyone was already on mobile and web. The problem had become execution across channels, and the version Aditya describes is painfully familiar to anyone who has run campaigns at a big company. You have every channel available to you and no ability to say what you actually did. He describes marketers who legitimately can't come back and report which campaign hit, where it hit, and which users it reached across the full cycle. Teams operated as channel owners rather than as one coordinated effort.
Then he gives the number that makes it real. One customer took 6 weeks to launch a campaign, because getting the segments required 20 people on a SQL query team pulling across 5 databases. 6 weeks. For a campaign. That's the daily texture of the job for the person you're selling to, and no amount of empathy language on a website touches it.
Here is the payoff. When your product lets someone explain their value in a simple way that other people understand, they stop being a customer and start being an advocate. As Aditya says, they will rave about you because you made their life easier and their job explainable.
Most vendor positioning is aimed one level too high. It addresses the company's pain, churn and efficiency and cost, while the person signing the order form has a smaller and more urgent problem, which is that they can't prove what they did last quarter. That mismatch is why so much martech messaging reads as true and lands as nothing.
Key takeaway: Write down the exact sentence your champion would say in ...

What's up everyone, today we have the pleasure of sitting down with Mary Keough, Fractional Demand Gen for B2B at DemandLoops.

(00:00) - Intro

(01:12) - In This Episode

(04:30) - How to Write LinkedIn Posts That Hit a Nerve Without AI

(06:39) - How to Run an Honest Goal Setting Conversation With Your Founder

(14:46) - Why Marketing Always Catches the Scope Creep at Startups

(18:39) - How New Marketing VPs Establish Credibility in Their First Month

(28:30) - How Women in Marketing Leadership Get Out of the Marketing Girl Box

(33:49) - How to Tell if a Founder Who Doesn't Get Marketing Will Trust You

(36:54) - What a VP of Marketing Job Description Should Actually Say

(39:12) - When to Fight for a Broken Marketing Role and When to Leave

(44:43) - Why Fractional Marketing Leadership Works as a Stopgap and Fails Long Term

(49:53) - How to Vet a Marketing Role by Talking to People Who Left

(53:17) - How to Use a Family Value System to Decide What Deserves Your Energy

Summary: Mary wrote a LinkedIn post listing every fear a senior marketer has about taking a VP of Marketing job, and it hit a nerve loud enough to become this entire episode. We take the fears apart one at a time: the goalpost that moves the week after your team clears it, the executive dinner that quietly becomes 6 more events, the launch strategy your boss never opens. Then we build the other side of it, including the 3 sentence job description she'd sign tomorrow and the written SLA that stops sales from eating 4 days of your designer's week. She also explains why the fractional model everyone is rushing toward is a stopgap, and why her own contracts are designed to end. Stick around for the performance review where nothing went wrong and it still nearly took her out.About Mary Keough
Mary Keough is a fractional demand gen marketer at DemandLoops, where she builds B2B demand programs that are designed from day one to be handed back to a permanent hire. She has spent 15 years across B2B SaaS and industrial manufacturing marketing, most recently as Director of Demand Generation at CoLab Software, and before that as Head of Marketing at Map My Customers and Senior Marketing Strategist at Gorilla 76.
Her roots are in industrial and manufacturing demand gen, a niche she cut her teeth on at Spraying Systems Co. and still speaks about at events like the Industrial Marketing Summit. She's built an audience of roughly 30,000 on LinkedIn writing candid, fundamentals-first posts, and she co-hosts the Purposeful Marketing Podcast.
How to Write LinkedIn Posts That Hit a Nerve Without AI
You can spot them before you finish the first line. Same hook shape, same clipped one-line paragraphs stacked into a wall, same closing question nobody answers. B2B LinkedIn has gotten very good at producing posts that read like posts, and the feed is worse for it.
Then something like Mary's shows up. Her post about VP of Marketing roles was a stacked list of fears, one after another, written against the line founders keep repeating about not being able to find good candidates. It went around because a lot of senior marketers had been carrying those fears for years without the words for them. That feeling of reading your own thought in someone else's sentence is rare, and it's almost never something a hook formula produces.
Originality is the entire filter for her, and applying it kills most content calendars on contact. Mary doesn't pre-plan posts. An idea sits with her for a while, usually something that's been bothering her that she can't quite name yet. Then a conversation does it: a marketer venting, a thread she can't scroll past, a call with a potential client. Something tips her over the edge and the post writes itself.
Not a single one of her posts has been written with AI. She says it flatly, and you can tell from the drafts: the scar tissue is right there on the page, and the sentences go where an editor would have smoothed them. The cost of that is inconsistency. She posts when she has something, and only then. The return is that when she does post, it carries the one thing a language model cannot manufacture, which is the specific irritation of a person who has actually lived the problem.
The volume play is finished. When any competitor can generate a competent, on-brand, keyword-aligned post in 30 seconds, competence stops being a differentiator and originality becomes the only moat left. Most B2B content operations are structurally incapable of producing it, because they're built to hit a cadence rather than to publish the thing only their company could say.
Key takeaway: Run every idea on your content calendar through Mary's filter before it gets assigned: if 10 other companies in your category could publish the same piece, kill it. Replace the slots you clear with a standing prompt to your team, asking what has annoyed them at work this month that they can't fully explain yet. Those are the posts that travel.
How to Run an Honest Goal Setting Conversation With Your Founder
Marketing targets in most startups get set the way weather forecasts get set in a bad novel, which is to say somebody decides what would sound good and then everyone agrees to act surprised when it doesn't happen. The CMO needs a board-friendly story. The investor update needs an AI angle. Almost none of it starts from historical numbers, and the marketing leader gets handed the result and asked to commit.
Mary's diagnosis starts a layer above the number. Companies keep saying they need a great VP of Marketing, an awesome VP of Marketing, an amazing director of demand gen. Very few of them are willing to do the work that makes a great one want the job. And the work starts with a company vision that has actual business goals underneath it.
The version she wants runs down a ladder. There's the grand vision, the thing that got the company funded in the first place. Under it sits a 1-year and a 3-year picture of what changes in the market if the company wins. Under that sit the business goals that get you there, segmented by department, so product, marketing, and sales each know what they own.
She walks through it with an attribution example, mostly because it's easy and partly because she can't resist. Say the company vision is to get founders, CEOs, and marketers to start caring about attribution. Fine. Now what does product build in the next 12 months to make that true? What does marketing publish? What does sales say on calls? Every function can answer that question, and the answers can be checked against each other.
That ladder is almost completely gone from the startups she sees. What's left is the revenue multiple, which is a wish, and a set of quarterly targets that get reverse-engineered from it. Marketing is usually the first function asked to sign for the gap.
The line that stuck with me is her framing of the whole discipline: everything about marketing is simple and fundamentals driven, and simple has nothing to do with easy. The hard part isn't figuring out what to do. The hard part is that you're doing it inside politically driven motivations, shifting strategies, and goalposts that move whenever the board asks a new question.
How to Push Back When Product Wants to Change the ICP
Anyone who has worked at a startup knows the move. You spend a year building a content engine around marketers, the quarter goes well, product decides the real opportunity is salespeople, and suddenly your entire library is pointed at the wrong buyer. Product changes a few things in the roadmap. Marketing starts over, and the results of that restart won't show up for 12 months.
Mary's honest about the limits here. Preventing the shift is difficult, so you mitigate it...

What's up everyone, today we have the pleasure of sitting down with Julie Beynon, Head of Analytics at Augment Code.

(00:00) - Intro

(01:07) - In This Episode

(04:13) - Breaking Into Data Without A Computer Science Background

(06:05) - Why AI Made Building Cheap But Maintenance Expensive

(09:25) - How To Push Back On AI Projects Without Killing Momentum

(12:28) - Building AI Agents That Multiply A Two Person Data Team

(16:29) - When Not To Reach For AI At Work

(22:23) - Running A Lean Data Team As A Force Multiplier

(27:28) - The Spreadsheet Test For Keeping Your Data Stack Simple

(31:49) - How To Prove The Value Of Invisible Data Work

(39:14) - Why Data Analytics Is Becoming GTM Operations

(41:47) - What Part Of The Data Analyst Job AI Is Replacing

(47:15) - The Data Skills AI Can't Replace

(52:07) - How To Decide What Deserves Your Energy

Summary: Julie Beynon came into data sideways from content marketing, got cut down to 2 words by a CEO, and turned that into a 15-year career running lean data teams. In this episode she breaks down how she cloned her best analyst into an AI agent named JimBot, why she pushes back on AI projects without ever saying no, and how she gets companies to fund the invisible foundation work nobody claps for. She makes the case that AI knocked down the SQL wall, so the real bottleneck moved to the data model underneath, and that simple is now the hardest skill in the room. There's a spreadsheet test, a team of 2 that runs like 10, and a surprisingly hopeful take on AI making us more human. Stick around for the part where she explains why the boring option takes the most nerve.About Julie Beynon
Julie Beynon is the Head of Analytics at Augment Code, an AI coding assistant company, where she runs a lean data team and builds AI agents that let the whole company self-serve trusted answers. She's spent roughly 15 years in data and go-to-market analytics, previously leading data at Census and analytics at Clearbit, with earlier stops at Customer.io and other startups.
She's self-taught and came into data sideways from content marketing, with no computer science background. Alongside the day job she writes a series of short essays on data, AI, and the discipline of knowing when not to reach for a tool.
Breaking Into Data Without A Computer Science Background
Most people assume the path into data analytics runs through a computer science degree or at least a few years buried in SQL. Julie's route looked nothing like that. She started as a content marketer, and by her own account she was bad at it. Her CEO once edited a piece of her writing down to 2 words.
That kind of feedback breaks a lot of people. It did the opposite for her. Getting cut down that hard built a habit of raising her hand for whatever nobody else wanted to do, and the thing nobody wanted to touch was data. The real turn came at Customer.io, where a new tool called dbt showed up and someone offered to train her on it. She figured saying yes would be silly to pass up, and within a few sessions the whole job changed shape.
The moment she describes is the one every self-taught analyst recognizes. You take a pile of raw rows nobody trusts, run it through a model, and suddenly you're holding something a whole company can use. She calls it realizing what power she had just yielded. The intuition that makes her good now got built out of the failures that came first, then an open door she was willing to walk through. Coming at data from content rather than engineering is why she reads the business side so well, because she learned the questions before she learned the syntax.
Key takeaway: Volunteer for the messy, unowned work on your team, especially the data nobody wants to clean. Pair it with one modern tool you can learn hands-on, like dbt, and get a colleague to walk you through a real project rather than a tutorial. The reps you build on work nobody else will do become the skill set that's hardest to replace later.
Why AI Made Building Cheap But Maintenance Expensive
AI has made it trivial to spin something up. A dashboard, a workflow, a scrappy internal tool, all of it now takes an afternoon and a few prompts. The problem is that the fun part and the expensive part were never the same part. Julie's been making this argument since before AI showed up, in a build versus buy post she wrote 5 years ago, and the tooling change only sharpened it.
Here's the trap she watches teams fall into. Someone builds a slick dashboard because they can, everyone's impressed, and then 6 months later sales walks up asking why a number moved. There's data drift, nobody remembers how the thing was wired, and the person who built it isn't the one who has to answer for it. She's the one who answers for it. So before anything gets built on her watch, she runs it through a short set of questions:
Is this mission-critical, or just fun to make?, Who owns it and who maintains it when it breaks?, Is there an off-the-shelf tool that costs less per month than the tokens it takes to build and run this?
That last one lands hard. You can burn $1,000 building something that replaces a tool costing $5 a month, and feel productive the whole time. The cheap build is a real cost hiding as a win. Maintenance never got cheaper, which means the discipline that matters now is refusing to build the thing you're perfectly capable of building.
Key takeaway: Before you build anything with AI, price the maintenance, not the build. Ask who owns it in 6 months, whether it's mission-critical, and whether a paid tool would cost less than the tokens you'll spend building and running your own version. If the honest answer is that you're building it because it's fun, buy the tool instead.
How To Push Back On AI Projects Without Killing Momentum
Ops and data people carry a reputation as the team that says no. Every shiny new tool, every "we should build this," runs into someone whose job is to point out that you already own 2 things that do the same job. That instinct matters more than ever now that anyone can build anything. It also curdles fast into being the person nobody wants to bring ideas to.
Julie's answer is to almost never say no at the door. There's a real constraint underneath her patience, since on the data side you have an actual duty to protect the data and the security around it, so some requests genuinely can't fly. Outside of that, she lets people run.
That quote comes from a specific win. One of her engineers built a proof-of-concept dashboard for a new product. It worked. Instead of telling him to stop, she treated it as a starting pad and let him take the first pass as far as he wanted. AI helped him productize and visualize what he was thinking. Then, when the question became how to make it sustainable, she stepped in and rebuilt it as a stable dashboard in their BI tool with observability and role-based access control baked in. The engineer knew every data point and how it was tracked, so she pulled the context straight out of his code instead of extracting it from him in a meeting. She calls it one of the most productive data cycles she's had. Letting someone go a few steps before you take the wheel turns pushback into a handoff, and the person feels seen instead of shut down.
Key takeaway: When someone brings you an AI-built prototype, resist the reflex to kill it. Let them take the first pass all the way, then offer to productize it yourself, pulling the logic and context out of their code rather than dragging it out of a meeting. Reserve your hard no for genuine security and data-protection lines, and treat everything else as a draft you can finish.
Building AI Agents That Multiply A Two Pers...