Rather than drowning in data, marketing teams should focus on asking precise questions that drive business growth. That's the core philosophy of Keith Perhac, founder of SegMetrics, a marketing analytics platform specializing in tracking complex, multi-touch customer journeys.
Key insights from this interview:
- Most teams track too many metrics (aim for 3-5 KPIs max per department).
- Valuable insights come from specific, smart questions, not gathering more data.
- Long customer journeys require different attribution approaches than traditional analytics tools provide.
- Marketing teams should spend 80% of their analysis time on what's different instead of routine reporting.
Keith breaks down how to define and track your most valuable customers, shares a framework for effective marketing reporting, and explains why monitoring the middle of your funnel is crucial for content marketing success.
Editor's note: This interview has been lightly edited for clarity and readability.
Tim: What's the key to getting valuable insights from your data?
Keith: The crucial thing to understand is that data isn't a singular thing that shows absolute truth. It has too many facets. You have to approach it with a specific question to get a factual answer. If you don't have a question when you're going in, there's no way to get a good answer out of data.
What are the main problems people face when it comes to data analysis?
There are typically two main issues. First, many people simply aren't watching their data at all. Second, once they decide to start looking at their data, they often don't know what questions to ask.
Our favorite customers at SegMetrics are those who say, "I've never been looking at data. Show me what I need to look at." They don't have any preconceived notions. The more challenging customers are those who come in saying, "I have all this data. What should I do?"
Why do you think some people don't look at their data?
People often don't look at data when they're successful. In the early stages of a business, you can be successful without it. Usually, people start looking at data when they hit a bump in the road. But if that's when you start collecting data, you're in trouble because you need historical data to make informed decisions.
You can't just start looking at data today if sales are down and expect immediate insights. You need to start collecting data and analyze it for the next six to nine months before you can actually do anything meaningful.
How do you help clients find the right questions to ask about their data?
Many people come in with vague questions like, "Why doesn't your data just tell me what I need to do with my business?" or "What should I do to make more money?" These aren't great questions. If I knew the answer to that, I'd be a billionaire. [laughs]
The questions you need to ask are more specific, like "Where are my best customers coming from?" And even that isn't straightforward. You need to define what a "best customer" means for your business. Is it someone who converts the quickest? Someone who spends the most money? There are many factors to consider.
How do you define a "best customer" then?
It's different for each company, but at the end of the day, it usually comes down to who spends the most money with you. How that's measured varies by business model.
For e-commerce, it might mean someone who makes a large one-time purchase. For a subscription product, it's someone who stays subscribed for a long time.
You need to define what a valuable customer looks like for your business. Is it based on the length of time they stay with you? Is it a big initial purchase? If you sell a one-time product that people don't come back for, you want to maximize that first sale. But if you're running a subscription box, you might prefer a lower initial cost if you know customers are staying for 6, 9, or even 28 months.
Beyond identifying your best customers, what other key questions should businesses be asking about their data?
I'm always a big proponent of looking at outliers. When you're looking at average numbers, all the good and bad get averaged together, and you just see a middle number. This can hide important insights.
For example, let's say you have an email sequence where people from Google convert at 100% and people from Facebook convert at 0%. The average is still 50%, which looks great on paper. But you'd be missing a huge opportunity by not digging deeper into those outliers.
Whenever you have a number, you want to find something you can break it down by. This is a core principle of SegMetrics. You find a group of people and then break them down by factors like where they came from (Google, Facebook, organic, etc.) or what their industry is (copywriters, designers, etc.).
You're essentially looking for a pivot point in your business that makes people either worth a lot or less depending on what they did with that information. This allows you to identify areas for improvement and tailor your approach to different segments of your audience.
Can you give an example of how this segmentation might work in practice?
Sure. Let's say you offer a course for freelancers, and you have copywriters, designers, and programmers signing up. You might find that programmers spend the most money but convert the worst, while copywriters and designers spend less but convert very well.
With this information, you might decide to create special content specifically for programmers to improve their conversion rate, since they're potentially your most valuable customers if you can get them to convert. Without breaking down the data, you might have missed this opportunity entirely.
How did you end up creating an analytics tool like SegMetrics?
It was actually a bit of a happy accident. We had a conversion rate optimization agency for several years, focusing on personality brands, info products, and creators. Our specialty was turning website traffic into leads through newsletters or opt-in magnets, and then nurturing those leads to convert them into customers.
At the time, there were plenty of web tracking tools like Google Analytics and KISS Metrics, but nothing for the middle of the funnel. We couldn't measure things like webinar attendance, email views, or clicks, or any of the segmentation data. We were working with various tech platforms and realized it was all just data in a database. So we thought, why not put it into our own database and analyze it instead of relying on spreadsheets?
How does SegMetrics handle long customer journeys, especially for content marketing where someone might interact with your content for months before converting?
This is where we really shine compared to other analytics tools. Many systems, like Google Analytics, can't identify individual users over long periods, while others, like Facebook, have limited attribution windows.
We track the full funnel. If someone signs up today, we can see every click they've made, even if their first interaction was months ago. We track anonymous users, and as soon as they provide an email address, we connect that information to the anonymous user. Then we tie that back to payments, credit cards, and everything else.
We're not just web-based; we connect to platforms like Keap, HubSpot, ActiveCampaign, Stripe, and others. This allows us to tie together web data with native data from these platforms, giving us a complete picture of a customer's lifetime journey.
What about privacy concerns with this level of tracking?
We're not grabbing anything that isn't already there. When tracking the web, we're using an IP address and a first-party cookie on your own site. There's no sharing with anyone else. We track anonymously until someone signs up for your mailing list, at which point they've given permission to tie that to an email address.
We're also fully compliant with privacy regulations like GDPR. When someone requests to be forgotten, we sync with your email platform to remove their identifying information. We can still see the user's journey because we need to track purchases, but all identifying information like email address and name is removed.
Are there certain industries or products where SegMetrics works particularly well or less well?
SegMetrics works well for most industries, but it's particularly valuable for businesses with longer nurture sequences. If your customer journey from initial contact to purchase takes more than a day or a week, we're really the best in class because we're one of the few tools that track that middle-of-funnel activity in detail. We excel there because that's what we did as a middle-of-funnel conversion rate optimization agency. It's been our focus for 20 years.
What common mistakes do you see people make when it comes to monitoring their data and reporting?
The biggest mistake people make is trying to measure everything. They say, "I need to look at this number and this number and this number." Suddenly you have a dashboard with 800 metrics on it, and you're never going to look at it. The perfect number is about three to five, five at the very highest. You can't do anything after that.
We recommend that in larger companies, every department or group has three to five KPIs (Key Performance Indicators) that they are watching. These should be the three to five things that are most important to their business. Then they leverage one or maybe two of those things to the next person up the tree. This way, anyone is only looking at three to five metrics at any time that they are monitoring.
How should teams structure their reporting to make it more effective?
I always see data as two parts. One is the KPIs that you are monitoring to make sure you're going in the right direction. And then you explore with breakdowns and outliers to try and find places to update. Because a KPI [isn't] going to tell you what to do. All it tells you is if you're going in the right direction.
As you're going up the organizational hierarchy, it's not always the same number being reported. Let's say you have a team in charge of ads. If ad spend is the same across the board, you don't need to tell the higher-up when you report that ad spend hasn't changed. It's not important. But if there is a change, then you leverage that up. It's about making the information actionable.
How can marketers present their recommendations more effectively?
Make sure it's simple. Make sure it's drilled down. Don't overwhelm [people] with unnecessary details. I get on calls all the time with my team and my ad team, and they'll say things like, "We increased ad spend by 5% and we had 12 clicks on this ad." I'm like, dear God, I don't care. I know I should care, but I just really don't.
I just really want to know what changed, what should I be doing, and what did you find when you were looking at the data that didn't make any sense or that we need to change or stop. Did ad spend go through the roof? Did click rate drop? How many people [did] we [get]? If it's the same as last week, I don't need to know about it. I want to know what's different and what we're doing to improve.
How should marketers approach data analysis and reporting to make it more valuable?
People often spend a lot of their time just collecting everything that's the same every week or every month as they're preparing reporting or recommendations. They should get that over with as soon as they can, just make sure it's the same as last week, and then move on.
The real value comes from seeing what's different. You should spend like 80% of your time figuring out that thing that's different so you can have a good recommendation or story around that.
That's why I always say five KPIs, because you're not going to spend a bunch of time on that. And then where you should be spending your time is seeing those breakdowns and looking for those outliers. It's like, okay, here [are] the leads that we got this week. Where are they all from? And how did that break down compared to last week? Look for differences and places we can improve.
How can marketers ensure they're asking the right questions when interpreting data?
We do this internally whenever we're optimizing things — we're always asking, what is the question we're asking and how did we get the data to make sure that the way we got the data matches with the question that we're asking.
For example, if someone came to me and said, "Our best customers are all e-comm customers," [I'd ask], how do you know that? How did you find out they were e-comm customers? What's your definition of best? What are the filters or data points that you used to prove that? It becomes an obnoxious question, but if you don't do that, then the politics of that really start to [clash].
Can you share an example of how data interpretation can lead to conflicts within an organization?
Back when we were doing the CRO agency, [we knew] there's a very specific number that every field you add to a form decreases the chance of someone filling it out by 5%. We were working with the sales team to build their intake forms. The sales team, of course, wants 800 different fields because they need all those answers [before] they get on the call. The more answers they have, the more effective their calls are. But [there's a] lower chance people are going to fill them in.
We had this really good form that was really short and they were like, "No, we can't do that." I'm like, we have like an 80% submission rate. And they're like, "We don't care. We need that information." So we added them all in and submission rate went down to 5%. But they got all the data they wanted. That was part of the politics.
What worked as the next step was them and us working together [to figure out] how [to] trick the user to start filling it out and then give them more information afterwards. So they think that it's a short form and they're already invested in filling out the form. So they'll continue to fill it out the rest of the way. [That way] you get all the information you need. [We asked], what are other things we can do to get what we both want? A lot of places aren't willing to do that. They're just like, "No, I want all 800 fields."
What factors influence the sophistication of attribution models in organizations?
In some large organizations, leadership might not care about metrics or understand them, which limits the need for sophisticated attribution.
There's also the issue of perceived complexity. Sometimes companies think their cases are too unique to measure, when in reality, if they got data around it, they would see that it's not all unique. They just don't know how to measure it yet.
In the end, it comes down to people who care about data and those who don't.
—
Keith Perhac is the founder of SegMetrics, a marketing analytics platform that specializes in tracking complex customer journeys and middle-of-funnel metrics. Before founding SegMetrics, Keith ran a conversion rate optimization agency focused on helping creators and information product businesses improve their conversion rates through data-driven optimization. You can follow Keith on X and LinkedIn.
More Animalz interviews
Stop Chasing Perfect Attribution: Deborah Carver on Making Analytics Work for Content
From Zero to 40k: How Sara Stella Lattanzio Built Her LinkedIn Empire
Founder-Led Content: How Christophe Pasquier Built Slite’s Brand Through Consistent Storytelling
Founder-Led Content: Blitz-Scaling Social with doola’s Arjun Mahadevan
Founder-Led Content: How Skiplevel’s Irene Yu Turns Personal Experience Into a Growth Engine
Founder-Led Content: Lessons From Reclaim’s Henry Shapiro on Authentic Engagement
How Jess Cook Turns Content and PR into a Power Couple at Island
Why Megan Morreale Thinks Content Marketing As We Know It Is Dying (And What’s Next)
