Imagine you are in a meeting with a new client. Everything is going well, almost too well. Your suggestions are well received. Your geeky jokes get the laughs you anticipated. You are on top of the world.
Then it happens. Just as you think you will be able to escape the meeting unscathed, a suggestion comes through. One that you are not sure is useful, advisable or even technically possible.
The client, nonchalantly asks a question that will send your head spinning for the next several days.
“I would like to know the path that every visitor takes to purchase our products”
A simple request with all kinds of baggage attached to it.
As a seasoned analyst, you have found that visitor pathway analysis is mostly useless. But how do you tell that to the client in a constructive way?
You have your work cut out for you.Understanding the appeal of visitor pathway analysis
When I first started trying to measure websites, the reports looked like this:
With Web 2.0, tools started to offer better visualization of these reports. The data tables were pretty much the same (and they still are), but additional bells and whistles were added to our reports. This included pretty pictures like visualizations on where someone clicked on-page and how they traversed the website.
Can you guess what digital marketers showed clients when trying to secure budgets for investing in online marketing programs? The pretty pictures. This was our shot at bringing web analytics into the mainstream – showcasing visuals that everyone can understand. These pictures were so cool at the time that we had to share!
Are clients on to something?
I think about Web Analytics every day. I read about Web Analytics every day. I write about Web Analytics once a week. You could say that this topic is important to me.
Many clients do not think about Web Analytics every day. Years may pass before an executive sees a web analytics report. Their only memory may be of the “aha!” moment they had when viewing the visitor flow report 10 years ago. Or the first time they saw a shopping cart funnel visualization.
It is easy to dismiss their knowledge as being outdated or inadequate. But maybe they are on to something?
Making the case for pathway analysis
Clients are smart in so many ways that it’s exhausting to list them all. They need to be masters of budgets, strategic planning, vendor relationships, technology selection and revenue generation.
There is a reason why they are asking you to help analyze the path visitors take to purchase. That reason should be obvious: they view it as a path to growing revenue.
Strategy involves starting with your desired outcome and working backwards. By retracing the steps that successful customers took to buy their products, you can grow revenues by finding more customers with similar backgrounds.
The appeal of visitor pathway analysis is that it provides a blueprint for how an ideal customer goes through your website and purchases.
Only it doesn’t really work that way.
The web is many things, but linear is not one of them
Clients have the best of intentions when asking for pathway analysis of their visitors. They are taking a strategic approach to solving their problems (finding new sources of revenue). But there is one major miscalculation here: the web is not linear.
Visitors come to our site through search. Other times they type in the URL directly.
Visitors are completely familiar with our brand. Others are seeing us for the first time.
Visitors can find their answer by clicking our navigation links. Others go straight to the search box.
For a website with thousands of pages, we quickly start to see that there is no linear path to success. Our reports quickly relegate any possible insights into the “>100 more pages” category. I have struggled to find a way to make these reports useful for the past 10 years.
Taking this into perspective, I believe that the client is asking for the right thing for their business. Our tools are still not providing the appropriate level of insight.
This has been a problem for a long time
For as long as these visitor path reports have been available, their utility has been debated. Avinash wrote a post on this way back in 2006 called Path Analysis: A Good Use of Time? Spoiler alert: it’s not. There is a good debate in the comments section of the site saying that these reports may be useful with segmentation applied, which is true.
Nearly 10 years later, and the same argument can be made. The report looks nice, but once you get beyond a linear set of 5 paths, it becomes impossible to manage.
We can get around the paths problem by applying content groupings and custom dimensions, but there is still a long way to go for success. Here is a sample behavior flow report based on year published. To learn how to do this, check out my recent Moz post on the topic.
This is a step in the right direction, but it takes too much configuration to be mainstream. We either need the tool to bake this type of analysis into the core product, or we will need a custom solution.
If a client is looking to use web analytics to find their ideal customer, I’m afraid that our web analytics tools still have a long way to go.
The sanity check
Google has made significant improvements to their visitor flow reports over the years. They have become more aesthetically pleasing and also have more segmentation functionality. Yet I still have not found them to be particularly insightful for analysis.
I though maybe I was missing something, so I enlisted the opinion of some of my analytics peers to understand how they used these reports.
The question was: Visitor Path analysis: overrated, underrated or properly rated? I asked them to give 1-2 sentences, but as you can see we got some passionate responses that needed elaboration.
Generally speaking, I think that there are far too many paths that a user can take on their way to conversion to get much meaningful (and actionable) information from path analysis. Understanding exit pages and problems that users may experience at key touchpoints is critical, but traditional path analysis is not a good way to do so.
Clients do ask for visitor path analysis but in a misguided fashion leading to confusion and often disappointment. These results frequently lead to analysts shying away from visitor path analysis to avoid uncomfortable engagements. This is a missed opportunity.
Clients may phrase the brief as ‘I need to know which pages convert best’. First off, pages don’t convert – visitors do. We have a responsibility to educate clients and analysts that visitor path analysis needs to start at the outcome and work backwards. In a recent blog post by Tom Davenport on ‘light quants‘ he describes a key skill set as being able to tell a story using data. The job of an ‘analytical translator’ or ‘story teller’ “is to start at the end – begin with the impact, and then very selectively reveal how the result was achieved.
This means examining all behavioural signals – not just pageviews – that lead to the outcome and understanding how these signals contribute to success. Starting at the homepage in a page flow report and fumbling through ever increasingly spider web-like paths trying to find value or any clue at all is a proven technique leading to frustration and failure.
No, the key here is to start at the point of success and work backwards, teasing apart the useful signal from the misleading noise.
The result of visitor path analysis needs to be a solid hypothesis that drives optimisation. It’s not an exercise in pride based on hubris – the brief needs to be framed into a tangible business question.
We can think there is potential value in changing X/doing Y. Is there any evidence in user behaviour that shows Z is stopping this? Is there any evidence that suggests X or Y add value currently?
Such a question may use visitor path analysis as one technique in answering the question. It is one of the tools in our analytical armoury and is best used in combination with hard quant skills, qualitative data from users and heuristic analysis from stakeholders.
Don’t miss out on the full picture, use visitor path analysis along with other techniques and use it in the right way to get the answers that drive action and optimisation.
The whole path concept is a difficult nut to crack, both technically and analysis-wise. For example:
- What happens if the user leaves the path and then immediately returns to it? Is this a broken path or a continuous one?
- What happens if a page is reloaded? Does this break the path or should it be ignored?
- When does a path begin? From the landing page? From a conversion? When the user exits some other path?
These questions help break down the logic of path analysis – it’s really difficult to do consistently. This is why Behavior / Visitor Flow are some of the most difficult reports to interpret / understand how they’re built.
My suggestion is for the user to think of funnels and not paths. For example, in a web store there’s a clear funnel from product exposure to purchase. That can be built with Enhanced Ecommerce, some other funnel-builder, or even just with Events that you then group into a horizontal funnel in a custom report, for example.
Using funnels lets you test paths that you know should be converting, instead of waiting for some serendipitous insight from a mesh of zig-zagging paths.
I think the whole problem with path analysis is that, like you said, the expected result is insight into what might have been missed. BUT THAT’S NOT HOW YOU DO ANALYSIS! You have a hypothesis FIRST, then you test against it, and any insight is derived from the results of this test, and it’s then fed into the next test. This is way more effective and targeted than trying to go John Nash on some weird patterns, and hoping they jump out and tell you how to build your business.
Visitor Pathway Analysis in Summary
Everyone has the best of intentions when it comes to visitor pathway analysis, but that doesn’t make it a useful report in its current form.
Our clients are looking for insight into how they can grow their business, and analyzing visitor paths is a conceptually sound way to do this. Only the technology doesn’t really support that objective.
As analysts, our mission should be on helping solve the root problem – finding new customers. We have many free tools at our disposal, and countless data points to observe.
By translating what is desired by the client into the language of our tools, we can help achieve the overall objective: to find more customers.