Marketing is how you reach people, convert them and talk to them. Customer intelligence is knowing which customers come back, why, and what to do about it next month. Cande Trucco, Growth Director at the ecommerce retention agency Flowium, put it this way on our podcast: data "tells you what happened," and customer intelligence "helps you understand what might be driving that behavior, and what you should do next."
I sat down with Cande for the first episode of a series we're calling From Marketing to Customer Intelligence. Before Flowium she was director of D2C marketing and customer support at SodaStream, a business unit of PepsiCo, so she's sat on the brand side and the agency side. These are my notes from the conversation, with her numbers and mine.
What is customer intelligence?
Most ecommerce brands already have the data. Where customers come from, what they browse, what they buy, how often they buy and which channels they engage with.
The gap is what you do with it. Cande's line: "We don't need more dashboards. We need to connect the information that you already have and turn it into decisions."
Her test is simple. A report tells you what happened last month. Intelligence should help you "decide what to do differently next month."
| Marketing | Customer intelligence | |
|---|---|---|
| The question | How do we reach and convert people? | Why do some customers come back and others don't? |
| The number you watch | Cost to acquire, first-order revenue | Repeat purchase rate by channel, by cohort, by product |
| What you get | Campaigns that reach and convert | A decision about which customers to go after next month |
The two work together. Intelligence tells marketing where to aim.
Why isn't one repeat purchase rate enough?
Say your repeat purchase rate is 25%. Good to know. It also hides almost everything.
That's Cande's example. Same store, same 25% average, two channels that behave nothing alike. Now you can ask real questions. Are we attracting the right customers? Should more of the budget go to the first channel? Is something going wrong after the first purchase?
She works in cohorts for exactly this reason. Customers who first bought in January can behave nothing like the ones from June. Organic search customers can stick far better than paid social customers. One product category can bring people back more than another.
Two checks before you move budget: compare groups over the same length of time, and look at what each group is worth after discounts, not just how often it comes back.
Why does retention start with acquisition?
A lot of brands treat retention as the thing that happens after the first order. Cande's view is that it starts earlier: "The type of customer that you acquire has a huge influence on what happens next."
I had a fresh example from one of our own clients, a web app that brings people in through Meta ads. We let Meta's algorithm chase the cheapest conversions. We got the best cost per download we'd ever seen, and first-week retention flatlined.
Our target was iPhone users, roughly 40 to 55. Meta had opened it up to people over 70 on Android, who were 70% more likely to download, never take an action and leave. Cheaper installs, worse customers. We pulled the targeting back to iPhone users only.
Not every sale is the same, and you only see it once you look at the acquisition number through a retention lens.
Can a subscription hurt retention?
Cande had her own example from an audit she ran the day before we recorded. A wellness brand with a subscription offer. Everyone assumes a subscriber is the goal.
When her team looked at how subscribers behaved, the people who subscribed on their very first order tended not to stay. They dropped off around the third shipment. Her read: maybe don't offer the subscription at the first checkout. Offer it after the second or third purchase, once the customer knows they like the product.
That only shows up when you look at the cohort. In Cande's words, these are the "tiny little things" you usually don't see in a report.
How do customers find a brand through AI?
Cande said about 40% of Flowium's leads now come directly from AI tools. A CMO opens ChatGPT or Claude, types something like "I need an agency that can handle our retention system," and Flowium comes up.
Her takeaway is that you now have a second audience: "You're not only selling to customers. You're selling to AI."
My take, and it's an opinion: it's still early. I think making your site and content easy for AI tools to read and quote is table stakes now, and I think the brands that start today have room to get ahead before everyone else catches up. It still feels like a blue ocean to me.
Where does AI still need a person?
AI makes the work faster and lets you go deeper. Cande was clear about the limit: "The senior level still needs the human brain." Feed it the wrong strategy and it can be very fast at destroying your strategy. You're just going really fast in the wrong direction.
What should an ecommerce brand do before Black Friday?
This was Cande's one action item, and it has a deadline.
- Pull everyone who bought from you for the first time last Black Friday and Cyber Monday.
- Track what they did next: average order value, whether they came back, and whether they ever came back at full price.
- Find the moment you lost them. A lot of them came for the discount.
- This year, plan something for that exact moment, so the customer you win on Black Friday is still buying from you next year.
Frequently asked questions
What is a cohort analysis?
You group customers by something they share, such as the month they first bought, the channel they came from or the product they started with, and compare how each group behaves over time.
Is one repeat purchase rate enough?
No. The spread matters more than the average. In Cande's example a store at 25% overall had one channel at 40% and another at 12%.
Do I need new tools for customer intelligence?
Usually not. Most ecommerce brands already hold the data. The work is connecting it and turning it into a decision for next month.
Cande and her team are at Flowium.
Full episode transcript
The episode, lightly cleaned of filler words. Times link to that moment in the video.
0:00 40% of Flowium's leads come from AI
Cande: 40% of our leads are coming directly from AI. AI tools, people directly, CMOs going into ChatGPT or Claude: "Hey, I need help. I need an agency that can handle our retention system." And that's how they find us.
0:15 Who we are
Dave: I'm Dave. I'm the founder of BigWork AI. I build AI products and services that help businesses save time and make more money.
Cande: I am Cande. I'm the director of growth at Flowium, where we do retention strategies for ecommerce brands. A little bit of background: I was director of D2C marketing and customer support at SodaStream, a business unit of PepsiCo. So I have both ends. I come from the brand side, but also the agency, and I know what is challenging, and I understand ecommerce brands.
0:55 How Flowium went from email agency to lifecycle retention
Cande: At Flowium, we originally were an email marketing agency, but 18 months ago we started the shift to become full lifecycle retention marketing, because we wanted to become lifecycle growth partners to our clients and to the ecommerce ecosystem. It's more about understanding the entire customer journey and using the data, the strategy, technology and AI, now that AI is in the center of everything, to improve how brands acquire, engage and also retain their customers. That's what I'm particularly excited about in this shift and where we're moving. I also think the role of agencies is changing significantly. Nowadays the agencies that will be creating most of the value will be the ones that can connect marketing strategy and execution to real business outcomes.
Dave: They don't care if the numbers look good. The numbers they want to look good are money in their pocket.
2:11 Marketing vs. customer intelligence
Dave: Cande, what's the difference between marketing and customer intelligence?
Cande: For a long time, marketing has been primarily focused on how to reach customers, how to convert them and how to communicate with them. And yes, of course, that is important, but I think we're entering a stage where the real competitive advantage is going to come from how well brands actually understand their customers. Today most ecommerce brands have access to an enormous amount of data. You can know where your customers come from, what products they're browsing, what they're buying, how frequently they purchase and how they engage with different channels. But the challenge is how you manage that data. Because that data tells you what happened, but customer intelligence helps you understand what might be driving that behavior, and what you should do next.
3:16 One repeat purchase number, two very different channels
Cande: So for example, knowing that your repeat purchase rate is 25% is of course helpful. But understanding that customers acquired through one channel have a 40% repeat purchase rate, while customers acquired through another channel have a 12% repeat purchase rate, that's where things start getting more interesting. Because now you can ask better questions. Are we attracting the right customers? Are we creating the right experience? Should we be investing more in one of those acquisition channels? Is there something happening after the first purchase that's affecting retention?
3:59 What AI speeds up, and where it still needs a senior human
Cande: AI is really accelerating this. It's helping us answer those questions faster, and not only faster, but go deeper. So to me, the shift from marketing to customer intelligence is about moving from execution to continuously learning from customer behavior, and using those learnings to drive decisions. We don't need more dashboards. We need to connect the information that you already have, and turn it into decisions.
Dave: I love that.
Cande: Because we love reporting, we love to review reports on what happened last month. That intelligence should help you decide what to do differently next month. And AI, of course, is incredible, because it makes things faster and more efficient. Automating, producing content, it's way faster. But it also depends on the data you give it. If you have a good strategy, the senior level expertise, the decisions: the senior level still needs the human brain. Because if not, if you are feeding the AI with the wrong strategy, with the wrong things, it can be very fast. It can be very fast also in destroying your strategy.
Dave: You're going really fast in the wrong direction.
5:46 The customer intelligence loop
Cande: This is the customer intelligence loop that I want to go a bit deeper on.
Dave: Yeah, we have a little diagram right here. Let's pull this up. This is just visual. We've got four buckets here, and then a nice little loop. It's the same stuff, right? How do you get found? Who's converting? How are you converting somebody who's making the first purchase? How are you keeping them? And then how are you analyzing who they are, what they like, and how to give them more of what they like? It's a loop that's been going on for a long time, but there are different pieces now in the loop.
6:28 Getting found through AI search
Dave: And one of the first pieces is, do you notice too that more people, and more D2C brands, are getting found via AI? People are not Googling things.
Cande: Actually, I was going to bring that data. For us, as an agency, 40% of our leads are coming directly from AI. AI tools, directly CMOs going into ChatGPT or Claude: "Hey, I need help. I need an agency that can handle our retention system." And that's how they find us. The other interesting thing is how important it is to be selling to AI. It's challenging, because you're not only selling to customers. You're selling to AI. So we have an absolutely new customer on the radar, and we're learning a lot of different things. Now we are all technical people.
Dave: AI is its own customer, actually. A lot of the internet is getting repurposed a bit for agents to be able to search it better. So making sure your content and your website are AI optimized is absolutely table stakes right now. I think the people that do this now still have this blue ocean in front of them. Companies and brands have the ability right now to get ranked a lot faster if they're following AI search visibility best practices.
8:04 Retention starts with acquisition
Cande: And talking about this loop, and making the distinction between acquisition and retention: a lot of brands still think about retention as something different that happens after a customer makes their first purchase. But retention starts with acquisition. That's where the loop goes, because the type of customer that you acquire has a huge influence on what happens next. Not every customer comes to your brand with the same motivation, the same expectations or the same potential lifetime value. That's why acquisition and retention should be much more connected from the start than they traditionally are. To give an example, let's say your ecommerce brand is investing in Meta, in Google, influencer marketing and organic content, and you might be measuring each channel on how much the customer acquisition cost was, how many customers you acquired and the revenue of that first purchase. But what happens if the customers coming from one channel are significantly more likely to purchase again? Suddenly your acquisition performance looks very different when you evaluate it through a retention lens. That's why they should be analyzed as a connected system. Understand where customers come from, what influenced their first purchase, what they bought, how they behaved afterward and what eventually drove them to purchase again. And of course you want to attract more customers like that. So you start: okay, this group, let's do an analysis of this group so we can get more of these kinds of people. That's the intelligence loop, and the more data you have, the deeper you can go.
Dave: That makes your entire budget and money go further, because you're tapping into the most valuable cohort. You want to find your top cohort and spend all your money there. The point is that not every sale is the same, and you only see it once you actually dig into the data.
10:39 Cohorts: what a 25% repeat rate hides
Cande: That's why I am a fan of working with cohorts, cohort analysis. Imagine you have an ecommerce business and you see that the repeat purchase rate is 25%. That's a really good starting point. People ask, "Do you know your repeat purchase rate?" "Yes, I know my repeat purchase rate is 25%," and they think that's it. They have the number. So what does that 25% actually tell you? Maybe your customers who purchased in January are behaving completely differently from those who purchased in June. Or maybe customers acquired through organic search have a much stronger retention rate than the customers you acquired through paid social. Or the customers that bought a specific product category are much more likely to come back than customers who purchased something else. Then you have these insights, and you build your strategy around them.
11:53 Cheaper installs, worse retention (my Meta ads story)
Dave: Speaking to that exact point, I just noticed this with one of our clients. It's a web app. We bring people in through Meta ads. We opened up Meta's Andromeda algorithm and just let Meta start feeding us the cheapest conversions. Well, the onboarding and the first week of retention flatlined. We were getting just as many people, even more people, and we were like, what's going on? One of the things we dug into: we had initially identified iPhone users as our target market, people between about 40 and 55. Meta is always searching to give us the best result at the cheapest price, so it opened everything up and started pushing it to people who were over 70 and had an Android. They were 70% more likely to just download it, never take an action and then leave. So even though we were getting cheaper installs, we were getting less retention. What we're doing this week is pulling all that out. We're only allowing iPhone users now. We're not letting the algorithm do what it wants to do. That's the kind of insight you're talking about. We had a bunch of people downloading for the best cost we ever got, but the retention tanked.
13:28 Why subscribing on the first order can backfire
Cande: I have this fresh in my mind, because yesterday I did an audit for a lead. It's a wellness brand, and they have a subscription. So, okay, let's see how these subscription people behave. Sometimes you think, oh, they subscribed, that's the goal, I reached the ultimate goal. But we went deeper and analyzed how these subscription customers were behaving. We noticed that people who subscribed on their first purchase, and a lot of people were doing that, tended not to stay for long. On the third shipment, they went down. So the insight is: maybe we should not be offering a subscription on the first order, at checkout. Maybe it's after the second purchase, once they really like the product, or even the third. With cohort analysis and going deeper, you find those tiny little things that you usually don't see in a report.
Dave: Yeah.
Cande: So that's the interesting part.
14:57 Do this before Black Friday
Dave: Let's give people some action items, high level. What should people be doing when they're analyzing their loop, and how should they be looking at these things?
Cande: My first, especially in this season of the year: I would do a whole cohort analysis on BFCM. Take BFCM of last year, the people who bought for the first time, the ones you acquired on Black Friday and Cyber Monday, and track how they behaved. What was their average order value? Maybe they had a bigger average order value, but then they come back less, or they don't come back at full price the way someone who bought at full price does. When did you lose that BFCM customer who came mainly because of your discount? Where do you lose them? So this year you can target that moment, and go with a strategy to get them back and keep them in your lifecycle.
16:17 Our favorite AI tools
Dave: So really quick, wrapping the episode up, rapid fire. Give me three of your favorite AI tools.
Cande: Well, I'm a fan of Claude.
Dave: All right. Love it. Me too.
Cande: I love how it connects to everything, when each platform has its AI connector and the AI systems talk to each other. That's what I like most. What about you?
Dave: One that I've been using for a while and love is HyperFrames, made by HeyGen. It lets you do animations and different video cuts. One video probably saves you three hours of work. It's been awesome having you on. Thanks, Cande.
Cande: Super happy to be here.
Want to see what your own numbers are hiding?
We build the AI systems that turn the data you already have into decisions. It starts with a 30-minute fit call.
Book a 30-minute fit call