The typical customer funnel is easy to understand and is useful for creating a strategy for your entire Sales, Marketing, and Customer Success teams. However, you likely have many different applications serving different parts of that funnel which can make it harder to understand your entire customer lifecycle. For example, you likely have a CRM where you can track key points in your customer lifecycle but you probably have other applications like Salesforce, Drift, Intercom, or Zendesk where you are also managing customer activity.
Here, we will discuss four steps to advanced customer lifecycle analytics, using all the data points in your customer lifecycle to expose opportunities to get into a deeper analysis. This allows you to really optimize for customer experience. It ensures that you are able to gain and create trust with your customers, which is directly tied to their willingness to stay engaged with you over time.
1. Create a model for your data
Create a simple, generalized model of what your customer journey looks like. For example, in Marketing that may be simply getting leads whether that’s through contacts downloading whitepapers, from events, or list uploads. The next step is nurturing those leads into a trial so that they can understand what value your product brings to them. The third step is a really important part of the customer lifecycle - when they find success in your product. The penultimate step is getting a meeting with sales so that they can figure out the right package for them. And then, finally, becoming a customer.
In order to get a model for your business, you'd start by identifying the customer lifecycle goal that you have for them, in this case, as a customer, and then work back through the stages that are really critical stages in reaching that goal. But, most importantly, keep it simple.
We are going to use a funnel as our model. Why do we want to use the funnel? First and foremost, it’s simple and useful when you have multiple, different touchpoints that leads have to pass through and different people in your organization that manage these different touchpoints. A funnel makes it easy to understand and communicate with different people on different teams, and makes it clear who owns which part of the customer journey as they move through the funnel. Finally, funnels are easy to prioritize and enable you to make apples to apples comparisons from one funnel to the next.
Funnels allow you to compare how valuable getting somebody through the funnel is, in contrast to the amount of work that it would take to reduce the friction in any one of those phases of the funnel. If it was easier to say, “Get more people into demo calls,” than it is to have them have a successful trial, and that would be of more value, you would say, “Let’s allocate all our resources towards getting more demo calls.”
2. Gather your data
It can be challenging having your data across multiple systems. Most people know the feeling of having some sense of the way the world looks through Google Analytics, and another sense of the way the world looks in HubSpot, for example. And those two things don't always reconcile, and especially after GDPR, I think everybody's world was kind of thrown upside down in having this clear picture of unidentified users and identified users. So that's one thing that Panoply can help with.
How do you get all of the data around your customer lifecycle in one, single place? Use a data warehouse or data management system, like Panoply. You can ingest all your customer lifecycle data from your CRM to Salesforce, from Drift to Google Analytics, into one place without using any spreadsheets. Point and click to those data integrations that you want to bring the data into your data warehouse, and you’ll have all of your data that exists within those systems into data tables that looks basically like spreadsheets with a bunch of rows and columns.
3. Build your dataset for analysis
In Panoply, we bring in three data tables or three data sources. For this example, we are going to use HubSpot data, MongoDB data, and Salesforce data, and despite the fact that they're all stored in very different ways on those different systems, we've put them in tables that all resemble each other very neatly using Panoply. So once we have these three tables, we're going to have to combine them together.
In SQL, the idea of combining different tables of data is called a "JOIN," and when you join data together, you find a common key in each of the tables, and use that key to combine all the data that corresponds to that key across multiple tables.
The HubSpot, MongoDB, and Salesforce data all have an email column, and we’ve joined them all together to have the data that corresponds with those three things that we are specifically concerned with, customers: when we acquire the contact date, the date the trial started, and the date that the contact became a customer.
Now I have everything from when they became a contact, when they started a trial, when the first time they talked to us via Drift, and when their trial ends. So we always have a very up-to-date picture of all of our customers, wherever they are, throughout the entire funnel, and that allows us to do deeper analytics when we want to, or we can just put those in dashboards so that we always have that visibility at hand.
4. Analyze and Optimize
Finally, we get to do the interesting part: the actual analytics. The three types of analysis that we're going to talk about are Funnel Analysis, so answering the question: how many people start and complete all the steps of a funnel? The second one is Measuring Time Lag, so how many people get from Step A in the funnel to Step B, or any other two steps in the funnel, and how long does that take? And the third one is a little bit more interesting, I think, which is a Cohort Analysis, and then it's asking how do different cohorts, or groups of people, behave over time?
Let's start in on the first one: Funnel Analysis. We're just putting numbers to the general model of the funnel that we had before, and we do that by counting the number of people who have completed any one of those funnel events in each column, and counting them up, and grouping it by these different segments. You will identify opportunities where there are dropoffs, especially if there’s dropoffs between systems. In that case, you have a huge opportunity to help two teams or two different people in an organization work better together, and find a better opportunity to improve their customer experience and improve your overall results.
But that's just answering the "what," so what if you want to answer the questions about the “why” people are getting through, or not getting through, your funnel?
One way to do this is to segment your funnel and look at how those segments perform differently. This is an example of how two funnels look from users who came in through paid search versus organic search.
If you're familiar with digital marketing, you know that the context changes pretty significantly between these two because with organic, often there's a lot more branded search, a lot more context, people have a lot more context about your business when they're finding you through organic search, and they're engaging with trials, or your software, or your product, or whatever it is that you use HubSpot, or any other CRM, for.
Measuring Lag Time
The next way to answer the “why” is to ask yourself: what does the lag look like between any two steps in the funnel?
Here we have the number of weeks it takes for any given amount of people to get from a trial to becoming a customer. In this example, we see that there is about a 3-week lag and then it kind of tails off from there. In this case, our free trial is 3-weeks long so what this says to us is that it’s really our job to create value and instill trust early on in that engagement and then let sales do their thing when people have found the value in their trial.
If you’re looking at lag time from one point in the funnel to another you can fine tune your email nurture, and get the right messaging at the right time, as this goes back to the idea of solving the right problem is more important than solving the wrong problem the right way. If you send the right email at the right time it is more important than sending the right messaging at the wrong time.
When we talk about behavior over time, that’s where cohort analysis comes in. On the top chart here we see that the exact same time lag distribution of how many weeks it takes for people to get from Step A to Step B. And below that is the exact same distribution, so where you see those darker versus lighter, the darker is where there are a lot more people, and where it's lighter, there are a lot fewer people.
I know cohorts can be a little bit abstract, so let's think about cohorts in the wild, and the way we can actually recognize them. Generations are a really good way to think about cohorts. My parents are baby boomers, I'm technically a millennial. I am very much a millennial, as much as you want to make fun of me, that's fine. Then there are the folks who basically have their whole life documented on Facebook and Instagram, the Gen Z folks who are just being born today.
So taking us back to the idea of moving from trial to customer, what we want to do is see how the effect of our marketing activity improves or shortens the lag time from trial to customer. We also want to see if it’s had an effect on the longer tail of customers who have started a trial, say, maybe months ago, but took them a long time to engage with sales, to then become a customer.
Those are the four basic steps to customer lifecycle analytics. If you’re a Marketer and you want to find immediate value in some type of analytics, simply model your funnel; it’ll help you and everybody in different teams think about how your marketing activities work together. This is especially so across teams and systems.
The second part of that is integrating and gathering all your data in one place. You can do this in a CSV, but it's going to drive you nuts; I recommend using Panoply to do this. And then, finally, clean up your dataset, do a little bit of data modeling, finesse the data so that it's nice and clean so that it's ready for analysis.