Data Pattern Analysis: Learn from a coaching session with Flint McGlaughlin
UPDATE 8/6/2021: This video was part of the testing and development process for MECLABS’ new, free conversion rate optimization course set to be released in the Fall of 2021 – Become A Master At Creating And Optimizing High-Converting Web Pages. Here is a quick excerpt – How to Create High-Converting Landing Pages: Ask this important question.
We have kept this video live because it still has helpful content. And now, more about this video…
How would you translate your company’s data and analytics into a step-by-step plan to grow the business? Where is your company’s most significant conversion opportunity? Can you detect any intriguing patterns in your company’s metrics?
To help you organize your data, pinpoint funnel leaks and increase conversion, MarketingExperiments held a live, interactive coaching session with Flint McGlaughlin, CEO and Managing Director, MECLABS Institute.
This first-ever coaching session from MarketingExperiments focused on practical application of a methodology to help you get more value from your data. Participants from around the globe interacted with MECLABS Institute’s CEO and Managing Director Flint McGlaughlin, Director of Research Services Matthew Klein and Director of Hypotheses Development Danitza Dragovic to discuss specific data challenges, ask questions about the lessons taught in The Marketer as Philosopher Episode 2, and get direct help in using the free Data Pattern Analysis Tool to increase conversion.
See below for specific questions that were addressed, along with time stamps in the video for those questions, if you would like to jump to a specific topic that addresses your immediate needs.
- 00:20 What you will get in this series
- 00:44 How can I simplify the way I view my data?
- 3:33 Can I really use the DPA to drive major conversion increases?
- 5:25 How many numbers do I need before I can detect a pattern?
- 6:22 What period of time should I use for data analysis?
- 8:47 How can I recognize a meaningful pattern?
- 9:25 How can I get my data out of Google and into the DPA tool?
- 10:15 How do I ensure that the data is accurate?
- 13:42 Why do you insist on visualizing the conversion journey?
- 14:47 Marketing is missing its way…
Watch the edited, condensed replay of the coaching session now to simplify the way you view and use your company’s data.
If you need help filling out the Data Pattern Analysis Tool or would like to have a MECLABS Scientist certify your data, email our sciences team directly with this link: Have a MECLABS Scientist Certify My Data
This transcript provided by GoTranscript:
Flint McGlaughlin: In this series, we’re going to give you a complete set of tools that we’ve developed in our research program over the last 20 years. For those of you that are new, we conducted 20,000 experiments. We invested $130 million-plus into a research program to answer a single question. “Why do people say yes?”
Karen asked such a good question. I want to share something that I’ll be teaching in future episodes, but this is a coaching session. All of you should think about your conversion and your data by using a tool we call the discovery triad. When you as a leader or your boss comes in and ask a “how” question, for instance, “How can I get more sales? How can we increase revenue? How can we get more subscribers? How can we generate more leads?” That’s what bosses ask. They ask “how” questions and what they’re really implying is, “Go fix this.”
Never let your mind stop at “how”, but instead think of this as a triangle. At the top of the triangle is “how” and then I’m going to move over here to the other side of the triangle, and that question is, let’s say, “How can I get more leads?” Here’s the “what” question. “What does the data tell me about customer behavior?” That is the primary question that you apply to your metrics program. When you apply that to your metrics program, it’s going to put a lens on so that you can suddenly see into your data something more important which is the customer patterns.
For instance, let’s suppose you notice that people are getting on this page, but they’re not completing the form. Now, I’m oversimplifying. Let’s suppose you see, “I’ve got great clickthrough from the ads. It looks like they’re engaging with the page,” but when I get to the form, many of them start, but don’t complete it. Let’s suppose your data showed you that. You went from “how” to “what”. “What does the data tell us about behavior?” That leads you to a “why” question now, and that completes the triad. The “why” question is, “Why are people behaving that way?” That typically leads you into the psychology of the offer.
For these almost 30 years, I’ve been asking, “Why do people say yes?” To get to the answer, I had to ask a more profound question, and that is, “Why do people say no?” Because you get a lot of nos before you get a yes. If I can understand the nos, I can move people up the micro “yes-chain” to the ultimate “yes” that turns them into a customer. Always think in terms of– Right now is your website, your challenges, all your “how” questions.
Listen, as Peter Drucker, my favorite business philosopher said, “The purpose of a business is to create a customer, and it’s the job of every single person in the company and it’s the number one job of the CEO.” The number one way you accomplish that is by crafting a strong value proposition. The only way you can craft a strong value proposition is to take the “how”. “How can I create a customer?” And complement it with a “what” question and then with the “why” question.
This is a big company. This is Aetna’s HealthSpire. If you look on your left, something happened that changed the control’s performance. Do you see the treatment? It represents something we call a signal set. You would call it a web page, but we view it as a signal set. It’s not a page, it’s not a web, it’s zeros and ones turning on twos in the mind. Look at the results, 638%.
Here’s another. We’re going to move on to another, 166%. This is another organization. That’s Toll Brothers, huge nationwide builder. That’s 166% conversion rate and that results in 167% increase in leads. Here’s another, 96%. That’s Fluke by the way. Defense contractor, big organization. This is another. 1600% increase. Oh, by the way, that’s The New York Times. These are all part of the research program at MECLABS. PR Newswire, 202%. Here’s another piece, CBS Sports, 44.5%.
Whether your business is big or small, it is possible to see exponential impacts if we can understand customer behavior. We can understand customer behavior if we can get into their behavioral traces. I like to think of them as brain tracts. Beneath those results, was a deep-dive into the data to understand how people were thinking and thus how they were behaving. With that in mind, I’m going to talk with you about a tool that we applied in those situations to get an increase. That tool was the subject of Episode 2 in our new series, The Marketer as Philosopher.
In the spreadsheet tool that we downloaded, you’ll see that we’ve found in one line, a break in patterns. All you need to establish a pattern is a majority. When I say majority, I mean critical mass occurs in the pattern. This is pure philosophy. When you’ve reached enough nodes that you can see anything weighted one direction or another. It’s hard with two, it’s easy with three. Sometimes three numbers can give me a pattern, but the more numbers, the more reliable our understanding of that pattern can be.
Look for the broadest threshold you can bring into your pattern recognition piece. Look over a year instead of over a week. A month is better than a week. This accounts for seasonality also. That’s a technical answer to a technical question, but I’m going to keep trying.
Matt, just go into your experience and tell us what you think about that.
Matthew Klein: We’ve done data analysis as short as just a few weeks even. We’ve done data analysis as much as three years. It’s somewhat contingent on the organization. Generally speaking, if you do experience any type of monthly seasonality, it would be helpful to have more than one year’s worth of data. 12 months of data is a really good starting place. You also have to account for– This is a really important consideration for any kind of timeline amount of data. You have to consider for any significant changes that may have happened either in your organization, with your ideal prospect, or within your website.
Let’s just say I take 12 months of data, but six months ago, I redesigned my website. I’m only going to have really six months’ worth of relevant data to inform how customers are behaving or interacting with my current website. There’s no golden rule in terms of the amount of data. It’s really contingent on how the uniqueness of your business as well as the things that you may have done, which may have caused changes in your data.
What you’re really trying to capture in your data analysis is a representative sample of customer behavior over time so that you can start to predict if based on what I’m seeing in the data, this is how my customer is behaving with my website right now. You want to be able to analyze that over, let’s just say periods of seasonality, whether that be weekly seasonality, monthly seasonality, et cetera. Does that make sense?
Flint: I think it’s a great answer. Philosophically, the foundation of data analysis comes from one sort of thinking skill, that is comparison. If you’re comparing future performance with previous performance, any of those different items you’re comparing these elements, they determine really how long that data needs to be. That’s the philosophical answer. Matt gave you the very practical answer, and I hope that helps you with that question.
We’re going to charge double, and double times nothing is nothing Michael, but still, [chuckles] that’s probably what the advice is worth.
Michael: Really, you clarified the fact because I had seen some other people had asked questions about if you have a minimal amount of traffic, what constitutes a pattern as opposed to just a new one?
Flint: Good. It’s a great question, Michael. If you see Episode 2, you’ll actually see us. We will highlight the numbers and we’ll show you the pattern and then we show you where we found the missing money. Hopefully, that will help.
Dani, go ahead. Take over.
Danitza Dragovic: We actually do have a version of the tool that’s a bit more simplified that you can download from Google Analytics solution gallery. It’s under marketing experiments. For some of the more simpler data polls, you can just download that tool. You don’t have to actually plugin manually. There’s some instances where there’s some more difficulty if you’re making custom funnels and things like that where you would actually want to go into GA and make custom segments for those. For the funnel itself, that won’t be automatic but most of the pages can be.
Flint: Dani thus is telling you, she’s created a template sort of tool inside of GA that you can use to pull your data automatically.
“Do I have a concern about the quality of data?” I run two metrics programs. Somebody asked that question. Matt, what do you think people should do to make certain the quality of their data is right? We can certify it for you if you need extra help and you want our scientist to– If you want to use the tool and say, “Did I get it right?” you can talk to us about that, but other than that, I’m going to ask Dani or Matt to answer that. Which one wants to go first to answer that question?
Matthew: Sure, I can answer it.
Flint: Okay, Matt. Go ahead.
Matthew: The first thing that I would do if I was looking at a data platform for quality is, this is incredibly basic, but does it pass the sniff test? Is there something in there that is just either completely implausible or highly unlikely? That would hint to a potential quality issue. I do believe that, as Flint mentioned, having a second tool is often very helpful to do so. You can also potentially– If you do have a testing tool, you can also run what we call dual control tests that may be able to validate or invalidate that your metrics are tracking correctly.
Flint: Matt, can I jump in there?
Flint: I want to point out what Matt has said is very important. Everybody should do this. If you have a testing tool, run a “A versus B” test but keep the pages the same. Watch and see if you see a big disparity in conversion. When he says a dual control, just to be clear to everyone here, that means testing the same page against itself. You have an A and you have a B and at the end of the test, you’re looking without saying, “Well, that number should be very close or there’s something wrong with your metric set up.” Matt, do you want to add to that anything or Dani? Add to that anything? Nope?
Danitza: I completely agree. Always having a second platform is enormously helpful just so you can see it. Usually, there is going to be a little bit of discrepancy, but if there’s a gap, you know there’s something wrong right away.
Flint: Dani, what is the average discrepancy you see? When you say a second platform, maybe you have an expensive– Maybe you’re using one of the Adobe. It doesn’t hurt to have Google running on the side especially an expensive version where you can compare page metrics and key metrics and see how close are they. What do we say on average is a disparity when we run two tools? Dani, I have a number in my head, but I want to hear what you say.
Danitza: If there going to be a big disparity and then you know there’s something wrong, but a small disparity would probably be– A regular one would probably be about 5%.
Flint: Yes, about 5%. The same number I have. All right. If you’re in 5%, you’re probably in the right zone. Now, I’m out of time. Matt, you want to say something?
Matthew: Yes. Sorry, I was just going to– One final thing related to quality of data. You can also audit your digital data with comparable metrics offline as well. If you’re driving let’s just say leads and you’re capturing leads in a CRM, audit what you’re seeing in your CRM with what you’re getting in your analytics platform to make sure that your conversion metric is accurate. Phone calls or any number of data as well, orders even, if you’re an e-commerce store.
Flint: I’m going to show you what you should be doing on your wall. Everybody, learn from this. See the funnel? It’s not just numbers. It’s written here. See it’s in the spreadsheet. You want to take those screenshots and put them in there. Now, if I were running your business or you said, “Can you help us with our organization?” I would put poster-size versions of that on my wall and I would write the conversion rate on there every day or every week depending on how much you see change in fluctuation and on each step through the funnel and then final conversion. When it change, forget all this fancy technology.
Go up there with your pen, mark through the number, write the change on there, and note the difference. You should be monitoring that number always. By monitoring that number, you’re able to adapt and adjust to changes. Candidly, a marketing department should be living around that number because that’s going to drive results for them. Make your walls talk. Don’t underestimate the power of big visuals.
Marketing is missing its way. Right now, understanding how to get the data right, and then understanding how to get the message right, and then understanding the value proposition, that’s all coming up in this series. It’s an opportunity for us not just to learn but we can come alongside of Teresa and rescue 3,500 small businesses.