Subscribe
Share
Share
Embed
"There is a version of AI adoption where efficiency just becomes a euphemism for doing more with less: you automate 30% of somebody's work, immediately fill that capacity with more work, and just move on." — Elizabeth
Elizabeth Martin's team at Babylist cut newsletter production time from 2 hours to 15 minutes, freeing up 600 hours a year, using Claude, an Iterable MCP, and a Notion-based copy pipeline. Her AI-lifecycle-marketing system has been called one of the most advanced MCP implementations among her ESP's clients, built in three months, with every failure documented along the way.
This isn't a theoretical AI conversation. Elizabeth walks through QA failures, an API key gap caught live on air, and what "30% more output" really costs a team, inside a company expanding into new verticals while the U.S. birth rate hits a generational low.
Timestamps
03:00 — The Role-Blurring Tax: The same AI capability that grants independence also creates friction when a team is expected to be experts in everything; the cognitive and labor load quietly expands alongside the power.
07:10 — Independence as the Real AI Dividend: Beyond workflow automation, Claude gave Elizabeth's team direct access to their data layer and API triggers, blurring the line between engineering and lifecycle marketing.
12:11 — Hiring While Using AI: Babylist is growing headcount, not cutting it. AI as a tool to do more with more, not to replace workers.
16:30 — Security, HIPAA, and the Line You Don't Cross: How Babylist keeps registry data and HIPAA-protected health data cleanly separated across systems.
22:11 — What the MCP Still Can't Do: Journey tile QA, audience field definitions, and multivariate notification tests are the real gaps behind the "most advanced" headline.
23:30 — The Moat Is the People: Asked whether the competitive advantage is the tool or something else, Elizabeth's answer is unambiguous: it's the people. Technology augments the engine, but the engine is human.
25:45 — The Pod Model and the Seat at the Table: Embedding lifecycle marketers directly into product team standups changed what the function owns, driving end-to-end strategy through production rather than being pulled in at the end to send a triggered email.
30:05 — Anyone Can Create an MCP: A real governance risk hiding inside MCP democratization: the line between empowerment and security exposure is thinner than most teams realize and requires active, ongoing conversation with an AI enablement team.
33:22 — Efficiency as Euphemism Warning: Why AI efficiency can quietly become code for doing more with less and the better bet marketing leaders should make instead.
35:55 — Slow Down to Speed Up: Real AI automation requires upfront investment in documentation and process clarity. If you want to automate a year from now, start writing everything down today.
45:36 — Fastest AI Wins Are in Verification, Not Creation: Babylist cut newsletter production from 12 hours to 15 minutes mostly by automating quality assurance, reframing where teams should look first when adopting AI.
Sponsor
Brought to you by Hightouch, the leading composable CDP and decisioning platform trusted by brands like Domino's, Chime, and Aritzia. 90% of customers have a real use case live within their first week, delivering world-class personalization at scale. Learn more at hightouch.com/msom.
Connect & Subscribe
Creators and Guests
What is Making Sense of Martech?
Unfiltered takes on the biggest shifts in marketing technology. We spotlight what matters, who's leading (or lagging), and what's next. In Martech, clarity is power — and we're here to deliver it.
00;00;01;08 - 00;00;24;10
Speaker 1
Welcome to the making sense of martech for the rabbit hole goes deeper than the headlines. This is a show that puts ideas to the test, not just on display. On the hotseat today, we have the illustrious Elizabeth Martin. Her team at Babyliss recovered 600 hours a year through AI automation and called it a gift. But when your company is growing into new verticals, your market is structurally shrinking and you've got a new CMO to impress.
00;00;24;10 - 00;00;48;18
Speaker 1
More time has a way of becoming more expected. Today we're asking is this an empowerment story? Or is it the number that justifies not hiring more? A little bit about Elizabeth First. She's the director of Lifecycle marketing and CRM at Baby List, the leading registry and commerce platform for growing families, where she oversees the email and customer engagement infrastructure that touches 9 million parents.
00;00;48;21 - 00;01;16;16
Speaker 1
One of the most high stakes moments of their lives. In roughly three months, her team built what one ESB called the most advanced MCP implementation among its clients cloud powered, end to end email production system that cut build time from 2 hours to 15 minutes, handing back those precious 600 hours a year. This isn't her first rodeo. She's led lifecycle and CRM teams at some of the biggest names in e-commerce, and you'll hear about them in a few moments.
00;01;16;19 - 00;01;39;21
Speaker 1
Welcome, Elizabeth. Thank you for being here. I'm so excited. Thank you for having me. I it's like a long time coming since when we finally met to I know. All right. So to dive in, let's start off with some rapid fire. What was your first martech tool? Salesforce. Back in the day. Ooh. Marketing Cloud or Pardot.
00;01;39;24 - 00;02;02;20
Speaker 1
I think part I actually I'm the black sheep. I love her. How is your token budget these days. One of my team members who's my heaviest user and where she's on track to spend about $1,200 this month. Okay. That's it's pretty reasonable all things considered. Yeah. There's to be curious to follow up in a couple of months. I'll let you know.
00;02;02;21 - 00;02;21;11
Speaker 1
I'll let you know if Claude disappeared tomorrow, and you had to go back to the old workflow for one one. What's the first thing your team would actually miss? And what, honestly, might they not? I think the first thing we would miss is our independence. Claude has actually, outside of just, automating our workflows has actually made us a lot more independent.
00;02;21;14 - 00;02;41;26
Speaker 1
We have access to cloud code. We have access to our data layer, snowflake. So we're able to really sort of blur the lines between engineering product and lifecycle on CRM in a really fun way for the team. And so I think we'd really miss that independence, being able to create API triggers and really think about our workflows differently.
00;02;42;03 - 00;03;05;12
Speaker 1
I would miss what I know about our program, or being able to dig deeper into what to discover about Life cycle zero the last. What we wouldn't miss is sort of the role blurring that comes along with that. We're expected to be experts in everything we do, and that can cause, a lot of extra work, honestly. And so I think that that causes a little bit of friction, for sure.
00;03;05;15 - 00;03;30;20
Speaker 1
And speaking of friction, which might be one of the words I would use for some of these, but you have worked at Amazon, Zulily, Real Self and all recipes. What is one word for your experience at each of these companies? I would say for Amazon, surprising Zulily. This is tough, but I would say bumpy. Since I went down with the ship, their real staff was Thrashy and all recipes was just fun.
00;03;30;21 - 00;03;52;25
Speaker 1
It was one of those jobs that really just so fun. I cosign each one of those without even knowing the context. Thank you. Okay. So now Baby List has a shop, a health vertical, a fulfillment center, a financial education hub, and a podcast. As a life cycle marketer, which one of these expansions is the hardest to write a welcome email for?
00;03;53;02 - 00;04;12;16
Speaker 1
We just launched what we call Early Investor, which is a really exciting product. It is a gifting platform for college funds, and it's really sort of the first of its kind. One of the main benefits, baby list in general and this product is it really takes the awkwardness out of our parents of asking what they need. And it's a new product.
00;04;12;16 - 00;04;36;14
Speaker 1
It's really industry leading. And because it's just so fresh in the industry, we're iterating and learning. Now that we brought it to market. That's wonderful. And last but not least, what's your hottest take? My hottest take today. Is that anything we built ourselves, my team personally for the workflows that we manage, is better than any into all AI workflow that we've come across.
00;04;36;19 - 00;05;01;08
Speaker 1
When you get to control all of the guardrails and variables, that makes total sense. But there's a steep learning curve I would imagine there. Yes, yes, that's true, that's true. Okay, I want to dive in. So setting the stage your team's framing of this AI is explicitly optimistic. So recovered time to better strategy, deeper personalization, smarter experimentation you name it.
00;05;01;10 - 00;05;32;21
Speaker 1
But you're doing this inside a company that's aggressively expanding into these new verticals and all these different changes. And to add to this, you're selling into a market where the U.S. birthrate just hit a generational low. That context matters for, what, 30% more output in the same team actually means. So in 2025, a Pew survey found that 52% of U.S. workers are worried about how I will be used in their workplaces going forward, and only 6% think it will create more opportunities for them.
00;05;32;23 - 00;05;55;18
Speaker 1
A third think it'll mean fewer. So have you actually protected that reclaim time from your team, or has it already been quietly folded into the roadmap? I think about it a little bit differently than that. So with the things that are really automatable or easily automatable with AI are clearly documented things that are like really rote process these things that you do over and over again.
00;05;55;18 - 00;06;33;28
Speaker 1
And so those are the things that are so easy to automate and really like kind of take off your plate to enable you to do higher level thinking. And when we talk about sort of 30% more capacity or be able to do 30% more, that reclaim time is being spent on sort of that higher leverage work. So I think I should if I back up a little bit, what we're doing now is use leveraging the AI tooling to do those rote processes so that the team members can be more deeply embedded with our partners and understand their processes and help make better decisions about the lifecycle implementations and really end to end own strategy through
00;06;33;28 - 00;06;56;09
Speaker 1
production, rather than just owning that technical, that tactical execution of the production they're doing the more strategic layer of work, which enables us to make better decisions for our users. So the AI piece is sure that the work is enabling us. The AI is enabling us to do a lot more, but it's enabling us to do like, better work.
00;06;56;11 - 00;07;18;10
Speaker 1
I love that. So building on that, does that mean your roles and job descriptions have changed as a result? Yeah, a little bit. When we think of sort of the evolution of sort of the AI workflows that maybe last, we really started a little bit earlier than most. And I think I know we'll get into this a little bit more later, but we started about a year ago and then really accelerated this year.
00;07;18;17 - 00;07;40;27
Speaker 1
And the tech team really led the charge. Obviously, like, you know, product and tech and engineering, they taught that those workflows are changing rapidly and that industry is changing so rapidly, and we partner so closely with them in terms of execution. And there are stakeholders and where they're stakeholders. And so we really kind of recreated our roles on our team.
00;07;40;29 - 00;08;09;01
Speaker 1
And to be more end to end. So now we have sort of what we call an embedded pod model where, you know, I'll have there's the celebrations and events, team AI baby list. And so we have somebody on my team who goes and works with that team and attends their stand ups. And so rather than them pulling us, then at the end when they need us to send a trigger to email my team members, then, you know, owning end to end, how are we going to bring this to life from a transactional perspective?
00;08;09;01 - 00;08;28;28
Speaker 1
But then how also are we going to bring it to to life from a product marketing perspective? So it's really enabled us to, in a really fun way, participate more in the product development cycle, but also make really good decisions for our baby less users, which is always the goal. And we have a seat at the table more than ever before.
00;08;29;02 - 00;08;48;18
Speaker 1
We have permission to be builders, we have permission to participate, and we have permission to use cloud code and try things out. And they've told us we can't hurt anything, which is comforting to hear when you're kind of down. Yeah, when you're doing something you've never done before and submitting peers and all of those things. But yeah, it's been transformative.
00;08;48;20 - 00;09;12;26
Speaker 1
Not without its challenges. You know, I as you mentioned, people do have some uncertainty. And, you know, when you're going through rapid periods of change that can be uncomfortable. And we've had our own moments of that, of course, as well, for sure. And I guess the question I have is with the same amount of headcount, does that mean it's a case for not increasing headcount or just mean we're optimizing our current team?
00;09;12;28 - 00;09;30;03
Speaker 1
We're actually hiring. Our baby was quite a bit. Our CEO likes to say we're not going to do more with less. We're going to do more with more, which I think is a great distinction. I actually just hired on the team somebody new who's going to be our marketing technologist and help us continue to uplevel our workflows.
00;09;30;03 - 00;09;59;25
Speaker 1
And, you know, we're expanding into doing more integrated marketing, which is exciting. Our engineering team is growing, rapidly. And and I think as we evolve our workflows, we're constantly evaluating what the needs of the team are. And as the product team doubles in size, like I'm evaluating what my team needs to, we we check in regularly to make sure everybody's bandwidth is that they're not working outside of work hours or yeah, that they're not too overwhelmed.
00;09;59;28 - 00;10;22;08
Speaker 1
And that's fabulous. So it actually is the accelerant we all want and help as opposed to something else. Ideally. Do you have to slow down to speed up in the beginning. Right. It does take a little while to give for us. That's clogged everywhere or uses different tools, but it takes a while. So you could plot all the contexts that she needs in order to execute, the campaigns.
00;10;22;08 - 00;10;41;18
Speaker 1
And so that can take some time. So at this point, we all know as it relates to prompting garbage in, garbage out, it's it's own craft and requires real human investment. Where does that labor live on your team? Is it distributed fairly or has it quietly landed on certain experts, certain heavy users? We have a lot of skills we leverage.
00;10;41;24 - 00;11;09;24
Speaker 1
So across the team. So that sort of try and minimize or sort of flatten out that work so that we are requiring new prompts for, leveraged all the time. Those skills are transferable. And now you can sort of house them centrally and they update all the time. We also are really fortunate APB list because we believe so deeply in the, future of AI and enablement that we have an AI enablement team.
00;11;09;26 - 00;11;31;06
Speaker 1
And so they've helped us with other skills, which are things like a prompt improver, which help me translate this document into a doc for senior leadership. And then I'll say, now run the prompt improver, and it will be like, I'm a senior strategist on Elizabeth's team, creating a deck for Xyzzy and really sort of helps improve. And then, yeah, we go from there.
00;11;31;06 - 00;11;53;20
Speaker 1
So we're doing our best to evenly distribute it really by sharing information and sharing skills and keeping those up to date. That's really lovely and thoughtful and I appreciate that. There's an AI enablement team. Well, because you could have one employee, you know, become an expert, another fall behind because they didn't understand anything. They didn't have the same resources and access.
00;11;53;20 - 00;12;14;08
Speaker 1
Yeah, absolutely. That's a nice democratization of how the team will work together and be able to troubleshoot and navigate problems and challenges along the way. Well, it's such a new way of working. We're all learning this together. I think. You know, I ask somebody who reports to one of my managers the other day, it's like, can you teach me how to do this cool thing you just did?
00;12;14;11 - 00;12;46;03
Speaker 1
I mean, we're all learning together. It's very fun. Every day there's something new and no one knows what's going to change in two weeks. That is true. And also troubling. So speaking of the tech, I want to get into the actual architecture. And because the details are where the real trade offs sometimes hide. So you have described the system as an elegant notion form Claude and Iterables MCP existing templates and a two hour build cut to 15 minutes.
00;12;46;08 - 00;13;10;19
Speaker 1
But template based speed is only as good as the templates. So what's the failure mode? Is there a source code to QA against? Yeah. Good question. So for us what the way it works is that the pass through. So we're taking notion and we are really using a notion page as a copy doc. And then using the iterable MCP to really publish to the template.
00;13;10;19 - 00;13;37;08
Speaker 1
And so we have a variety of these templates that live in notion. And we just use the copy doc to create the notion page that then passes through. Where this becomes challenging is if the template fails QA because it is code, right. We're not using the drag and drop editor, or we're not building it in our old way as a building where we would just go back into the template, make a small change or tweak, then are just going back out and or rebuilding the whole thing from scratch.
00;13;37;08 - 00;14;02;06
Speaker 1
Again, if the template fails QA, I think that's kind of where we see failures. The other limitation that we have is we haven't quite figured out modularization yet, and some of our templates are very modular. And so we need to like there are some templates that we haven't quite figured out how to really automate. And so that's something that we're really excited to tackle or the end of this year.
00;14;02;12 - 00;14;24;09
Speaker 1
Maze, where has the system produced something that got close to going out the door but shouldn't have? If we're doing a lot with cloud or writing copy, we're creating templates. We're doing analysis where we have sigma, which is where our data is plugged in. We're pulling data from iterable. We've had some files. Clod. You know, when it writes copy for us we have to review that.
00;14;24;10 - 00;14;55;17
Speaker 1
So even if it's on brand we were laughing about this example the other day where cloud spit out something that was like air filters are on sale just in time for fire season. It's like no cloud. You have to, you know, that's not going to work. Well, that is not, except yeah, or email templates will fail QA because by the nature of our business, we really, believe that you need to have an unsubscribe link at the very top of the template just in case something happens during your pregnancy and you really need to opt out.
00;14;55;19 - 00;15;18;10
Speaker 1
That's why. Is that not a standard right thing. Yeah. Yes. And so like we have to iterate on sort of our QA skills all the time. And sometimes we had it recently we were working on an analysis of our audience health and transactional versus marketing emails, and it gave us some wrong outputs that if we hadn't verified, would have given us the wrong direction.
00;15;18;10 - 00;15;49;00
Speaker 1
So we're getting there. It's iterative, for sure, and it's nice to hear, even with such an advance set up, that truly very few teams not only are equipped, but have live and implemented that yes, we're aware, still makes mistakes. And guess what? The humans are so required in order for this to go right? As mentioned, your MCP implementation of an AI lifecycle production model has been lauded as one of the most advanced, and I agree, what would you push back on that characterization?
00;15;49;00 - 00;16;01;16
Speaker 1
Like? Where does it still feel like y'all are in V1 beta? There's a lot we still like.
00;16;01;18 - 00;16;21;26
Speaker 1
Shout out to adorable. Please make it better. There are things we still can't do like journey QA, can't read journey tiles do that API limitations I have a whole document that I run every two weeks to make sure we're continuing to uplevel our program. There are still like, I would love to be able to set up like eight way multivariate push notification tasks.
00;16;21;28 - 00;16;46;13
Speaker 1
That's something I could do with the cloud browser plug in with admirable, but should I? Probably not. I should probably use an MXGp to do that. And so there's still some areas of, gaps and opportunities. I think there's audience building. We do it in the platform because the audience, the ways that we define our audience and the data fields are still not, not quite there that make sense.
00;16;46;13 - 00;17;13;25
Speaker 1
And I love that you mentioned needing more features within an MCP. A couple weeks back, I had a conversation with Ezra about governance and security and safety. And I'm curious what your team is doing since you're, you know, you handle HIPAA. You guys have a health business. Like, for me, those are immediate red flags of, oh, no, how do we set up this MSK so secure, safe and abide by all laws.
00;17;13;25 - 00;17;34;00
Speaker 1
What is your team doing to ensure that for your customers? Yeah, our customers security is paramount. Everybody lists. Everything we do is with our users best interests in mind. We love our users is one of our core values and you wouldn't believe how much we love it. So when you set up the MCP, are you using an API key which has your specific permissions?
00;17;34;02 - 00;18;02;23
Speaker 1
And those API keys are. I'm not to get too specific, but those are specific to whatever platform instance you're in. And so for us with iterable, we have a separate instance which keeps all of our health data separate. And so that's really keeping all of our user health information separate from our registry data. And so all of that HIPAA protected data is not something that we're using with our iterable MCP, where we're using the Oracle MCP for our registry projects.
00;18;02;26 - 00;18;26;02
Speaker 1
Iterable is HIPAA compliant. And I'm sure there is a way we could use the MCP. And I have a compliant way, but we aren't there yet. And we would definitely consult our health business attorney before we did anything that would potentially touch that data. We love some compliance and actually enforcing it. Yeah for sure. And I got to ask, are you guys rotating as API keys every so often?
00;18;26;02 - 00;18;52;02
Speaker 1
Okay, Jacqueline, I listened to your podcast and I talked to my AI enablement team. And we are going to make sure we are taking all of the security precautions. Fabulous. Well, I'm glad our conversation to have a semblance of, you know, the spark of inspiration for sure, because, yeah, Ezra and I've been talking about it for a long time, and it's mind boggling how little protections are built in.
00;18;52;05 - 00;19;26;04
Speaker 1
So you really have to be at the forefront of thinking about every angle. What is a feature that you guys have not built? I think the dynamic templates would really be a big unlock for us. I think journey building automation, we built a QA tool for that that sort of sits outside of our standard QA tool. I think merging those together is something that we need to put on our list, because I think that that would really be a big unlock for us as we're building more complex journeys, especially as we're adding things we need to tell our, our users about and sort of at the beginning of their journey with us.
00;19;26;06 - 00;19;50;18
Speaker 1
I think those are pieces that we we probably need to continue to uplevel and push on. I think, too, there's this piece where we probably should go back to the beginning and see if there's like the tools change so fast, and three months ago is very different than today. And so we should probably sort of are we plan to reevaluate sort of our, the way we set everything up and is there a better, faster way?
00;19;50;18 - 00;20;11;12
Speaker 1
Or, you know, maybe we could solve this QA problem where we have to go out and rebuild all over again? There's we could do. Yeah, yeah. Within your own work and re optimizing and exactly like that's awesome. Brought to you by our sponsors. If there's one thing that's followed me at every stop in my martech career, it's trying to get good data into the hands of marketers.
00;20;11;17 - 00;20;33;02
Speaker 1
That's why I'm so excited to tell you about our sponsor, High Touch, a leading composable CTP, an AI decisioning platform. Companies like Domino's, chime, Aritzia, and PetSmart trust high touch to power their data. And here's the kicker 90% of customers have a real use case live in production within their first week. That means you can implement a world class CDP in months rather than the usual years long headache.
00;20;33;05 - 00;20;52;08
Speaker 1
That's why top brands choose high touch to personalize every customer interaction at scale. See what high touch can do for you at high touch.com/msm. And now back to the hotseat. Since you have publicly shared how you're setting everything up, are you concerned ever that a competitor would maybe take this idea and run with it? Or is that actually an encouraging thing?
00;20;52;14 - 00;21;15;15
Speaker 1
So encouraging. We would love people to do that. As I mentioned before, our goal at maybe less like everything we're doing with AI is to make things better for our families that we serve. And we are so excited when people reach out. You know, Natasha on my team presented at Adorable Activate, and she has created such a network that she is meeting with now to talk about how we've set this up, how we're doing it.
00;21;15;16 - 00;21;36;01
Speaker 1
We want to help, other businesses and other folks do the same thing because, you know, unlocking your capacity to do that higher level thinking is going to make things better for our entire industry. If people are doing that or sending better emails or push notifications, then all of our collective unsubscribe rates are going to go down right.
00;21;36;03 - 00;22;04;16
Speaker 1
The inboxes will be less crowded. So I think it's the great thing to be able to share. And we're excited. Rising tide lifts all boats and everyone thought that way. Exactly. So what you and your team have built, do you consider the tool a moat or the people, or is it something else? I think it's the people. I think people forget how important people are sometimes.
00;22;04;18 - 00;22;30;09
Speaker 1
I agree they are the engine. Even if you have technology improving upon that engine, it's an augmentation. So zooming out to the industry at large and from maybe less for the moment, because they're not the only one running this type of experiment. They just might be one of the more advanced versions. Do you think what you've built, maybe less, is actually rare, or is nearly every CRM are doing this?
00;22;30;09 - 00;22;51;22
Speaker 1
How many other teams do you know of who are thinking like this? I don't think it's rare. I think we maybe started sooner and we're a little farther down the path our AI journey at. Maybe less started in 2025. I think I mentioned that earlier on. We we had what we called project AI, where every team sort of started to think about what the future of AI could look like for your function.
00;22;51;24 - 00;23;27;22
Speaker 1
And we really had been encouraged to think about AI in the context of improving our workflows for our team, and then shift our focus to on our side, really improving our life for our users. We had help. They formed our AI enablement team, who were really critical in helping us get set up with the horrible mix. And we had really strong goals and documentation, which really enabled us to kind of hit the ground running, and we gave ourselves like a firm deadline of like we wanted to have an email sent through the MCP by Presidents Day that was like the sales and deals email that we wanted to get sent out.
00;23;27;22 - 00;23;55;25
Speaker 1
And so we were sort of ready to go. And iterable on their MCP, where it was like we we had a strong AI roadmap, and what we were sort of operating against was let's just retry all the things on this roadmap continuously because things started to change really rapidly starting in Q4 of 2025. So if so many other companies are maybe just on the differential, jittery of the same end goal, how are any of these companies actually protecting that?
00;23;55;25 - 00;24;21;04
Speaker 1
Reclaim time for strategy. Do you have specific things a baby like you're doing for your team with regard to that? Do you know other teams that have strategies or frameworks? You know, I don't know. I think for our team, it's really we what we've seen is that people are really enjoying the time to dig into the questions they have, you know, being able to say, like, how does this data trigger this email?
00;24;21;06 - 00;24;45;06
Speaker 1
How does this tie together with this? What would a good strategy be for for this? Or how can I end to and accomplish this task and so that's kind of how I've seen it come to life on my team. And then sort of with our pod model owning end to end strategy through production, it's reclaimed time upfront to do more of that, like thinking work, which I think people tend to, well, enjoy more.
00;24;45;06 - 00;25;06;17
Speaker 1
What I've seen be a little bit more unique about our workflows. That baby list is that we have a little bit more access to the tools and a little bit more freedom to experiment. You know, I recently did several rounds of interviews for a role in my team. And and what I've seen, that's a lot of people are really staying within the tools don't have permission.
00;25;06;17 - 00;25;24;18
Speaker 1
Like we need to get somebody involved to send an automated slack message or need to get somebody involved to sort of either set up an MCP or right to publish to notion or whatever, and we can do all of that stuff, which is an incredible gift for myself and the team, because it just enables us to really push ourselves as far as we can.
00;25;24;23 - 00;25;45;20
Speaker 1
So I think what is particularly interesting to me is similar to you. I'm a doer. I don't like to be restricted in it. It kills me when I work with certain enterprises, and it takes two weeks to even get a field created in Salesforce, and that's a quick turnaround. What makes the culture enable this type of work? Yes, it's definitely cultural.
00;25;45;23 - 00;26;04;23
Speaker 1
Baby List was founded by an engineer, Natalie Gordon, who founded the company to suit her own needs. And really, we have a builder culture, and one of our core values is progress, not perfection. So, you know, you might want that Salesforce field to be perfect, and I might be able to give it to you in an imperfect way.
00;26;04;23 - 00;26;28;05
Speaker 1
Faster. But the faster way will probably suit your needs for what you need today. So we sort of live our values and how we build and how we execute every day. And that makes it exciting, for sure. And I guess the next immediate question I have is governance. Like who is overseeing, who gets access to what? Because it sounds like you have pretty much full access to everything.
00;26;28;05 - 00;26;54;15
Speaker 1
You need to not break anything, but to inform and have all the context. Yeah, the context is all there, which is at this point incredible. And it's all the context in a snowflake in a data warehouse. Yes. I still have stuff like basic access, but I can plug snowflake in to Claude and I can talk to snowflake about what I need to know about emails and workflows and what's happening with this user segment.
00;26;54;15 - 00;27;18;23
Speaker 1
I don't have all of the access. Right. That's still protected. That still belongs to the data team. My team has built our access with cloud code. We can't like publish PR, we can submit PR or so reviewed by engineering. So we're being and we still do have there still is a line that we can't cross. And we do talk to the AI enablement team about where if that line, how can we move that?
00;27;18;26 - 00;27;40;12
Speaker 1
We talked to the other day about I didn't know anyone can create an MCP, but anyone can. And I know your thoughts on MSPs. They're security and governance. I'm excited. I'm intrigued but worried. Yeah. Yeah, exactly. We are protecting the business in a smart way. We have enough smart people thinking about that that we feel like we're operating safely.
00;27;40;15 - 00;28;08;08
Speaker 1
I love that, okay. Let's say that this approach style to production becomes a standard practice in the next two or so years, because I think the market usually takes any room 2 to 5 years to kind of follow the real trailblazers. And I'm curious, how does this impact the life cycle, the CRM org chart, and does that look different depending on the size of the company and you name it?
00;28;08;08 - 00;28;27;22
Speaker 1
If there's 600 people at Baby Lesson, where do you foresee this? Yeah, I think we started with what we're calling our pod model, where team members are embedded with product and really end to end owning sort of all parts of the process. I wish more people would admit that there is no answer yet. We don't know until we know because I agree.
00;28;27;25 - 00;28;48;27
Speaker 1
I think that the nice thing about working in this space and having worked on it for so long, is that it's constantly changing. Channels change, platforms change, capabilities change. And so the people who know this space know that change is inevitable. And so our roles are going to change either way. This is just accelerating it. Yeah. And the only constant is change.
00;28;48;29 - 00;29;10;26
Speaker 1
So we can't truly predict what the org chart is going to bring. That said, fewer people may be doing the same amount of work or last, we don't know. Do you think there's a role that we have not thought of that will exist and come out of this new standard? I'm sure I'm sure there's a role we haven't thought of that will come out of it.
00;29;10;29 - 00;29;36;11
Speaker 1
The buzz right now is the marketing engineer or the chief marketer that reports to the CMO or the CEO who orchestrates all of the the marketing work or marketing tech enablement. So I think there's roles that are going to evolve, and then there's roles that are going to be invented. And I think that the people who adopt and use the tools are setting themselves up to be ready for whatever's next.
00;29;36;14 - 00;29;58;07
Speaker 1
Yeah, yeah. So many thoughts on the marketing engineer. It sounds like a marketing, UX or martech person this whole time, just with a fancier, elevation of title to make some people feel better. I've been in B2C marketing my entire life, so I've never held a marketing ops title or worked super closely with marketing ops, so I don't know.
00;29;58;11 - 00;30;20;28
Speaker 1
You are very marketing ops focused though, because for those who don't know, Elizabeth and I met a couple 3 or 4 years ago when you were evaluating different ISPs, and we talked to all the different challenges and, you know, potential angles to think through of what the right fit would be. And you were more informed than anyone on that call.
00;30;21;01 - 00;30;45;23
Speaker 1
Yeah. So while you maybe haven't had the title, doesn't mean you haven't had the experience. Thank you very much. It's true. So with all of the differing opinions on what the future of AI is bringing, what is your perspective on AI? Efficiency is mostly a either reduction or just a general headcount justification, just up as an innovative narrative.
00;30;45;24 - 00;31;11;04
Speaker 1
There is a version of AI adoption where efficiency just becomes a euphemism for doing more with less, where you automate 30% of somebody's work immediately, full capacity with more work, and you kind of like, move on, just like automate, automate, automate. And that we've seen that tension across other industry where AI hasn't broadly translated into massive headcount reductions yet, but it is raising expectations around speed, AI.
00;31;11;06 - 00;31;37;18
Speaker 1
And I think in some cases, like what's been really exciting for us is as we embed with our partners, we're able to work faster. I think what we haven't sacrificed is our commitment to like a good output and a good work product. And so for us, we've been able to spend time on the front end of what the what the end, what the output is, instead of sort of the technical execution aspect on my team.
00;31;37;18 - 00;32;02;01
Speaker 1
I don't want to say we spend 30% more campaigns because we can. I want us to be spending that time being smarter and making sure that, like, if our program can be 30% more efficient and drive the business and our communications can be 30% better for our users, that would be a better output than us just doing 30% of our work, if that makes sense.
00;32;02;08 - 00;32;28;02
Speaker 1
I am a huge proponent of send out when it's deploy something when it's actually worthy. Don't just deploy and execute something just because can not only is a not a thoughtful way, but particularly strategic. You've got very specific segments and subsegments. Use that wisely. Otherwise you're just kind of throwing gasoline on a fire and just hoping it does something.
00;32;28;02 - 00;32;53;15
Speaker 1
Yeah. And it's not great. Yeah. And I think we we really it's for us it's like baby let's really believes in AI. As for us it's really about making parents lives easier. And and everything we do is how can we make parents lives easier and serve our families, our growing families. And yes, I'm so curious, but I want to zoom back in from the greater industry and onto you.
00;32;53;18 - 00;33;19;20
Speaker 1
So a year from now, if this AI workflow is the standard across different industries and everyone's using some version of it, what's the thing about how you led this rollout that you want people to remember and or maybe some advice? People really need the space to experiment and try things. The where this has been the most successful is when you have real workflow problems, like real things that you're trying to solve a problem.
00;33;19;20 - 00;33;53;28
Speaker 1
If you're not trying to solve a problem or or automate something that is slowing you down day to day, you're not going to see the opportunity. And I think that's been really eye opening is everybody has a different moment with workflows for this specific workflow. I think I can be uncomfortable and really recognizing that, you know, changing your work really rapidly can be uncomfortable and really making space for that, but also like making space for that, that experimentation and you do have to slow down to speed up.
00;33;53;29 - 00;34;13;01
Speaker 1
And then I would also say start documenting now if you want to do this a year from now, I'll make sure all of your processes are documented, because I won't work if you don't write it down. Exactly. I mean, I've always been a huge documentation nerd. Now I feel so validated and vindicated this whole time I I'm so happy for you.
00;34;13;03 - 00;34;34;04
Speaker 1
I know I'm the rare, rare one that enjoys that kind of thing. And I think to add on to your statement about giving people the space to try and not be overwhelmed, I think there's also easy things to implement, like cook AI channel where it's like, hey, I built this cool thing. And so that it's more of a learning experience for everyone.
00;34;34;04 - 00;34;56;05
Speaker 1
And to your point, you had a quick scenario where someone presented something and you're like, wait, can you show me exactly how you did it? And really fostering these connections of guess what? Anytime someone says they're an expert in AI right now is a full because no one's an expert baby list we had every other week we have AI.
00;34;56;05 - 00;35;21;13
Speaker 1
So until we have an AI channel where people are sharing their ideas and we have on the marketing side, we have our own AI channel because the other area channel can be very engineering focused and we're marketers. But what do you mean you're also marketing engineers? We're at home. Thank you, thank you. You know, somebody might come to the show and tell and be like, oh, this is silly, but I figured out how to get to Inbox Zero using the Cloud Gemini plugin.
00;35;21;15 - 00;35;40;26
Speaker 1
And somebody else, that's their moment. And somebody else might come and say, I figured out how to create a calendar for all my kids events. And so people are using it in all kinds of different ways. Not, you know, we're using it to automate different things, but every way that people are having these larger moments, it's really fun to hear about.
00;35;40;26 - 00;36;07;04
Speaker 1
And and those should be validated too. I love that, and I guess to make it really digestible. What are some quick examples of those rote processes that and automations you guys were working on that you then automated that would translate for others? Yeah. When coworker George, that was just such a huge unlock for us. I was working on a with one of my team members.
00;36;07;07 - 00;36;34;10
Speaker 1
We were working on trying to automate QA for email and we could not get it. We could not crack it. When Co-Work logs, we figured out how to create a QA skill that could check all the links, could compare it with the copied out, could compare it with all with our UTM doc, with all of these things that really unlocked about eight hours a week for the team, and then MCP and notion and iterable that really automating our newsletter and sponsored email production.
00;36;34;12 - 00;36;55;20
Speaker 1
That has been a big unlock for us, unlocking like 2 to 4 hours to just like 15 or 20 minutes. So we've really been able to to cut down time with some of those processes. And then as we continue to push on different things, like for me, I have a strategy skill I built to evaluate all of the things we want to do against our life cycle roadmap or our life cycle.
00;36;55;20 - 00;37;19;09
Speaker 1
Figma, which has all of our, all of our comms and our all of our UX research, like, is this a good thing to do? Is this something we should or so there's just so many interesting ways to move forward for sure, and build your own skill that is based on your company's current objectives, OKRs, goals, and ensuring that is considered as part of every part of your process.
00;37;19;09 - 00;37;42;28
Speaker 1
How does this ladder up? Absolutely, absolutely. And our AI enablement team, actually, one of the first things they started with was like create your boss a skill that's like automate your boss. So like all of the feedback you've ever gotten, like put it in and then start like leveraging like, yes, business goals, but also, you know, compare all the documents you have to write with all the feedback you've ever gotten, even just iterating on that.
00;37;43;00 - 00;38;20;16
Speaker 1
Take your most recent annual review and say you're wanting a promotion, and these are the steps outlined and you are required to make happen. Make that part of it where it's like, hey, where are we at with this? Did I do things unintentionally that I'm not recognizing later into those objectives? Yes. Get that promotion. Absolutely. And so in hindsight, having built this entire program from the ground up, what is one thing about running this fast that you would have done differently if you had to start all this over again tomorrow, maybe not done anything different?
00;38;20;16 - 00;38;44;14
Speaker 1
Well, I think I might have maybe cleared a little bit more capacity to slow down, to speed up. So actually like, made a little bit more space because you think about automation and you don't think about the time it takes to invest in automating processes, especially when you think about AI and you think about how, you know, everybody thinks you just automated process, but you don't.
00;38;44;20 - 00;39;04;11
Speaker 1
You have to give cloud the context. You have to give the tools all of the context and then context, and then you have to iterate and you have to work through it, and then you have to understand how this works in your current workflows and how those have to change in order for you to adapt. And that's hard when you have a lot of other work to do, and you do need to slow down.
00;39;04;14 - 00;39;29;12
Speaker 1
I think of it a lot like the discovery process, whether you're in a sales cycle or if doing an audit of your own system and programs. It's like the hard work is actually just the research, the investment in getting it right from the start, because you're going to pay for those consequences later, whether it's be a tech debt or just improper thought processes and ways in which the solution changes that.
00;39;29;13 - 00;39;58;12
Speaker 1
We've said all of this and you've got this unique AI enablement team, what is something that they are and passionately requesting? And the team has yet to maybe satisfy? That's a good one. They are working really hard with us on making sure that we have our cloud set up with all of the right context, so a task versus the skill projects are so powerful for sort of long running things that you're doing.
00;39;58;12 - 00;40;23;19
Speaker 1
But I think there's there's just more learning we need to do about when's the right time to use each one. Yeah. It's like the next level of taxonomy and organization of documentation and user, a different approach, but still the same core concept. That's the Lily. Well, I was about to thank you so much for this conversation. You are doing something most marketing leaders are still theorizing about.
00;40;23;22 - 00;40;43;08
Speaker 1
You're actually running it, measuring it, and being honest about where, you know, things aren't fully ready yet. And this is a conversation I think everyone should listen to. And everyone should be having internally as well. So thank you for bringing the real version of this to the world. And before I let you go, who is someone we should have on the podcast?
00;40;43;15 - 00;41;03;27
Speaker 1
Yes, I recently got a chance to talk to is Megan Teep from Rover. She is their chief growth officer. Really amazing and growing company. They're an 18 countries co headquartered in Seattle, which is where I live in Spain. And she is just super impressive. And I think she would be fascinated to talk to you. Amazing.
00;41;03;27 - 00;41;37;29
Speaker 1
I'll take you up on that introduction. Well, thank you so much for coming on. This has been an illuminating conversation. Where can folks find you? I'm on LinkedIn at Elizabeth Martin. Before we wrap, here's what I want people to actually walk away with. Not the headline, but the underneath at all fabulous AI system cut newsletter production from 12 hours a week to 15 minutes, most of that from automating QA rather than copywriting, which means the fastest AI wins are often in verification, not creation.
00;41;38;02 - 00;42;01;08
Speaker 1
If you're deciding where to start, maybe the CEO frames this as not more with less, but more with more. And the company is hiring even as output multiplies. Which means the efficiency story here is a reinvestment, but not a headcount justification. And it's worth knowing which bet your own leadership team is actually making. And also, the team didn't eliminate manual work.
00;42;01;08 - 00;42;29;18
Speaker 1
They redistributed it a centralized skill set, prompting labor doesn't concentrate on 1 or 2 people. Well known gaps like Dani could still sit outside of the main system. This means real eye roll out look like an ongoing rebuild and not a finished state. Thanks for tuning in to making sense of more tech. See you next time! A special thank you to Christine Ritter, who edited this episode, and an extra special thank you to Jenna Carter for believing in this passion project means business.
00;42;29;20 - 00;42;30;09
Speaker 1
Stay curious.