Having trouble analyzing user behavior? Meet the beginner's guide to creating an effective conversion funnel

Original author: DAVID ARNOUX
  • Transfer

You have probably already read tons of articles on how to start customizing the metric on your website, measure its key performance indicators (KPIs) and make smart decisions based on the data received. And, if you already apply this knowledge in practice, starting to create your own metric and track events on the resource, you know very well how difficult it is.

We spent a lot of time in order to collect all the information together. In this article we will try to explain to you how the measurement of performance indicators should be organized and which tools are best suited for this. You should not take the information below as the final guide to action, but we hope that this review will serve as an excellent additional help to increase the conversion of your business.


The first thing to understand ...

We strictly adhere to the principle of “lean methodology”: Construction-> Measurement-> Study-> Construction-> Measurement-> Study.

Thus, the company must have a person responsible for the metric.

As practice shows, such a concept as “analysis of the behavior of visitors based on events” can provoke a beginner to study only those indicators that are in the growth stage. But the most competent strategy is the use of ALL data received from users, building a conversion funnel and changing your product based on it.

Determining which aspects of a conversion funnel is most important to a startup’s success is a major concern for every established company. The following is an example in which it is clearly visible that something in the service is not working as it should, and it is time to think about the problems, and then develop effective ways to solve them. The first thing to do before starting tracking and analytics is to determine which metric is the most important for your organization. It’s not at all as difficult as designing and assembling a rocket: everyone, with a certain zeal, can become a competent analyst. What you are dealing with are two groups of data:



  • Demographic : generalized information about users (name, geolocation, age, etc.)
  • Events : what these users do on the target site (following links, subscribing, commenting, etc.)


Simply put, you have “users” and “events”

In this article, we are primarily interested in events. They include the whole history and the entire path of the user on the site: what pages he visited, where he clicked, how many invitations he sent, whether he left comments, etc. To some extent, Google Analytics (GA) can help with this, but our view is that its capabilities are very limited. This is a great tool to understand where your visitor came from, it can be used to build the highest point of the conversion funnel. To analyze how many users visited my landing page, and then subscribed or left a post - all these, of course, are important points that directly affect the conversion. But, unfortunately, GA is not suitable for an in-depth analysis of events and actions that a user performs on your site.

Without analysis based on event tracking, uncertainty is generated that leads to guesswork. Guess what really happens inside your product. But do users use the necessary functions, do they do it right, do they return to your site, do they do it regularly, do they invite new people where they go, what should be improved on the site?

We would recommend using GA to improve the strategy of “getting” users to your site, but services such as Mixpanel.com or Carrotquest.io will be much better for tracking the “events” that these users make - they will allow you to maximize your product .


What do we need to track?

bad way

There are reasons why people bang their heads against the wall when it comes to building a conversion funnel. The main difficulty is to answer the question “What should I track?” We have prepared for you a number of simpler questions that will help you come to an understanding of what should be monitored and what to analyze:

Step 1. What task does your website perform? What is the main goal of your entire business?

Let's look at a few simple examples:
  • For an online shoe store, the main goal is directly to sell shoes;
  • For community sites like Airbnb, get as many people as possible who book your accommodation;
  • For our Carrot Quest service, the task is to encourage people to communicate with their users, analyze their behavior and improve their product;


Step 2: What needs to be done to achieve these goals? What are the main KPIs that will show that your business is well developed? You do not need to invent hundreds of such indicators: it is better to limit yourself to 1-2 characteristics. Let's try to show it on our examples:
  • For the online shoe store, the main KPI is the number of shoes sold;
  • For Airbnb, the number of seats offered for rent and the number of seats reserved by guests;
  • For Carrot Quest, this is the number of events tracked and messages sent (either manually or automatically). And to a lesser extent, this is the number of user segments created.


Step 3. What leads the user to the actions indicated in the last paragraph?

Here it’s already a little more complicated: depending on the structure of your site, leading the user from the “Came to the landing page” event to the final event (“Place an order for the goods”, “Booked the accommodation”, “Created an automatic message”), which we designated as KPI, there are many ways. The trick is to find such paths (remember that we consider events such as “clicked a button”, “visited a specific page”, “filled in a specific field”) so that ALL users can go directly from the “landing page” to key action. Sometimes this can mean the absence of elements such as a search, passage through certain pages or other paths from the funnel, as it is possible for the user to switch from Event No. 1 immediately to the Final KPI event without performing such actions. Here
  • Way 1: visit the landing page> search by city> filter the search results by price> go to the page of a specific apartment> select the date> click the "Book" button> Click the "Make payment" button
  • Way 2: visit the landing page> select the date> go to the page of a specific apartment> select the "Similar place to rent"> click the "Book" button> Press the button "Make payment"

In this case, only events marked in color are common to two paths. There is no need to track other events. An example may not be the most successful, but the basic principle on it is not difficult to understand. But back to our three examples and create funnels for them:
  • Conversion funnel for a shoe store: Visit the page> Go to the specific product page> Add to cart> Click "Order"> Click "Make payment";
  • Conversion funnel for Airbnb (lessor): Visit the page> Click "Add accommodation to the list"> Click "Next"> Click "Confirm";
  • Conversion funnel for Airbnb (guest): Visit the page> Go to the specific housing page> Click on "Book"> Click on "Make payment";
  • Conversion funnel for Carrot Quest No. 1: Visited the landing page> Registered> Put the code on the site> Created an event (started tracking the event);
  • Conversion funnel for Carrot Quest No. 2: Visited the landing page> Register> Put the code on the site> Created an automatic message (any of 3 types);
  • Conversion funnel for Carrot Quest No. 3: Visited the landing page> Register> Put the code on the site> Created an event (Sent a message manually - to one user or segment);
  • Conversion funnel for Carrot Quest No. 4: Visited the landing page> Register> Put the code on the site> Created a user segment;

These are just the most basic examples. There are quite a few other events that you may also have to keep track of. Now that we know what to track, we should eliminate errors in obtaining specific data.


How to read the necessary information?

Naturally, we do not want to check the same event a huge number of times. And we are not so interested in the number of events (for example, 432 messages were sent today) - it is much more necessary to analyze who initiated these events (for example, Ivan sent 8 messages today, and Olga Vladimirovna - 53). Events are step by step connected with each other by the same user ID - thanks to this, it is possible to analyze certain steps of each site visitor. That is why it is necessary to clearly distinguish between the number of users and the actions that they perform.

But do not worry! Correct demarcation does not require significant programming experience, but the person responsible for collecting information about user actions should be technically savvy (or at least he should not be afraid of encountering difficult tasks).
On how to set up events, read one of the previous articles: Creating site funnels and measuring conversion over a cup of coffee without a line of code .

In order for this data to be of some use, you should go a little deeper into the analysis. Why only 10% of visitors visit the registration page? Did the “right” people visit the homepage? Are CTAs (call-to-action, calls to action) organized attractively on the site? And finally, why only 1% of visitors make a purchase? Are we asking too much from them? Perhaps the registration form is too complicated? Start from the very beginning of the funnel and analyze the entire user’s path: from the first visit to the site to registration and purchase of your product.

Focus on the highest interrupt rates. Try to understand why users are leaving your site and find a way to keep them. For a more accurate analysis of why users leave the landing page, the Yandex Webvisor service is suitable (though they won’t show what happens after the user has registered and entered his personal account).


Great and terrible analytics tuning

Most analytics tools today allow you to get started as quickly as possible and without significant pre-configuration. The best and most advanced tools have round-the-clock technical support via email, instant messengers or telephone. Setting up the lion's share of events is no more difficult than adding a couple of lines of code to the site page. And sometimes it’s even simpler to use the visual interface.

Now you need to make sure that the most important events are correctly tracked. Back to our Carrot quest example. When you first encounter Carrot quest, you need to register and install the code on the site. This is important for the service, because without the installed code, further work is not possible.

The funnel in this case is as follows:


The example clearly shows that the main problems for users are created by Step 1 - Step 2. Thus, it is easy to identify where the zone of greatest interruption in the funnel is located, and proceed to fix this problem.


If you still think that funnels are not that important ... think again.

Even the smallest problem in your funnel can deprive you of a significant amount of income in the long run (if you achieve a positive business result at all). Take a look at the statistics presented below - average performance per 1,000 visitors.

Using such schemes will help to understand why the interruption occurs and make the cost of customer acquisition (Cost of Customers Acquisition, CAC) significantly lower.

Perhaps you are targeting the wrong people - check demographic factors. Perhaps your CTAs are unclear, incorrect or incorrectly located - you should change them or move them, conduct an AB test. It is worth remembering that analytics shows only one side of the coin. With its help, you can find out what is happening on the site. But the most important, in our opinion, factor is an understanding of why this is happening. The analyst will not give you an answer to this question. The only way to understand WHY conversion is reduced is to ask your users. But this is a topic for another article ...

Cool products.

The user analysis and communication service team is Carrot Quest.

Also popular now: