Case author: Elena Berezovskaya, head of the agency’s web analytics department.
Client. Telecommunications (young internet provider).
Time of cooperation. Five months.
Have you ever had a situation when you have a big difference in the number of requests from the Internet in the reports of a marketer and a sales manager? Probably, yes.
This problem can have many reasons, starting from those associated with different methods of calculation, ending with trivial errors. But whatever the reason for the discrepancy in the data, the main thing is that it does not lead to the loss of physical leads.
In this case study, you will learn how we prevented a loss of 20% of leads by simply answering the question: Why does the data from different reports differ?
After replacing the order form, the discrepancy between the number of Google Analytics goals achieved and the actual leads in the client’s CRM has sharply increased. It was necessary to determine the cause of such a problem.
- The reason for the loss of leads was found and eliminated.
- The number of sales and satisfied customers increased by 20%.
Initially, we added a multi-step form for ordering service on the client’s website. Since the number of incoming calls significantly outnumbered the number of completed applications from the site, we suspected that users did not have the patience to fill in so much information.
We hypothesized that if we shortened the current form, the number of submitted applications would increase by 40%, while the number of calls would drop by 10%.
Then we decided to test how this would work in practice.
To do this, we set up tracking of the user’s transition from one step to another when placing an application. As a result, it turned out that a large percentage of users abandon the registration of an application at the second of four stages:
Graph of the results of the analysis of the effectiveness of the service order form
Based on this, we decided to shorten the form. The new form contained only four fields – a phone number according to a given mask, full name, address, and comment.
As a result, after this update, the number of applications on the site increased by 45%.
Everything was fine until, when comparing the data on the number of requests from CRM and Google Analytics, we saw a discrepancy of more than 20% (with an acceptable disparity of up to 5%).
The number of leads sent from the site to Google Analytics and CRM
First of all, we checked the correctness of the event settings and the Google Analytics goal, which are responsible for submitting the form. It worked correctly.
Then we decided to check the sending of data. That is, to answer the question – which applications do not get into CRM.
As a result, we got the following datasets:
- anonymized data in Google Analytics;
- data in a temporary table;
- data in CRM.
Form Submission Data Scheme
We compared data from different sources and saw that there are more of them in our “parallel” table.
Comparing the data in the table with the data from CRM, we got a list of contacts that are not in CRM.
After analyzing those contacts that were not sent, we found out that data is not sent to CRM in the following cases, if:
- the user started to enter the number with +380;
- the user has specific codes of telecom operators, for example, numbers starting with 067;
- It also turned out that the main loss occurs from devices with a screen resolution of less than 360 × 640, which is about 45% of all users. As it turned out, a notification about the incorrect filling of a field was displayed above the form after clicking on the “Confirm Submission” button, and users with low screen resolutions did not notice this message.
- Improve the verification of the correctness of entering the number;
- Move the comment about incorrect input to the corresponding field:
- Make the button for submitting data inactive until the data is entered correctly.
This update allowed us to prevent the loss of 20% of applications from users.
Identifying and solving the problem showed that:
- data collection logic should always be transparent;
- it is necessary to prescribe data quality requirements;
- a process is needed that allows you to find data loss (for example, comparing the amount of data at different steps);
- if the data collection logic or any inconsistencies cause you questions, do not stop until you find the reason for this.
Remember, it is crucial not only to guess the reason for the discrepancy in the data but also to confirm its existence. A hypothesis without confirmation or refutation is dangerous because it is misleading and can lead to financial losses.