Call tracking

Call tracking tasks include evaluating the effectiveness of advertising channels, monitoring the work of a contact center (analyzing sound recordings , monitoring the number of missed calls), calculating ROI , measuring the cost of attracting customers among any advertising media, and optimizing the marketing budget for more efficient media channels.

Using this technology, it is possible to record data such as the phone number from which (and to which) they called, the result of the call (received or missed), the duration of the conversation, the sound recording of the conversation, who called and on which advertising channel. Based on this data, it is possible to generate reports on the total number of calls, the number of calls for each advertising channel / monitored traffic source, as well as the number of unique and repeated calls and calls for each advertising channel every day up to an advertising element, which is assigned a replacement phone number .

In the basic version, one VOIP or DID number is enough to track calls. All calls to this number are displayed in the call log, which allows you to control the number of missed calls, their duration and other parameters. However, tracking the performance of advertising channels requires multiple numbers. The essence of the functioning of calltracking is that for each advertising channel that needs to be analyzed, or for each one-time user on the site, a telephone number is assigned, and the potential client sees only this number. When a user calls by phone, his call is forwarded to a call center. Thus, you can correlate a call to a specific number with a specific advertising source (advertising campaign) or visitor to the site. The scheme of work with calltracking can be described in three steps ^[1] :

1. You need to determine how many advertising channels or how much traffic you need to track. The number of monitored advertising channels is limited only by the number of rooms, and as a result, by their cost.
2. Each advertising channel or one-time site visitor is assigned a separate phone number. Each advertising source publishes a number dedicated to this advertisement.
3. On the site where the advertising traffic is directed, a JavaScript code is put in, which remembers each visitor and replaces the phone number. Calls to phone numbers used to track calls are forwarded to the call center, and the data of each call is recorded for further analysis.

All methods used to track calls are divided into two groups: automatic, which have gained the greatest popularity, and semi-automatic ^[1] .

The automatic types of call tracking include tracking calls by source (static method) and tracking by user session (dynamic method). Tracking calls by source is a classic method in which each phone number is assigned to a specific referral source (channel, type of traffic , company, key request or a combination of them), which makes this method expensive. As a result of this, this method is used to analyze traffic sources, but is not suitable for tracking on key requests. Accordingly, tracking by source does not provide in-depth data for analysis, most often it is used by online stores with a small advertising budget and turnover, which do not need an in-depth analysis of the received data ^[1] .

When using dynamic call tracking (per user session), the phone number is assigned to a specific user session. When tracking calls over a user's session using a relatively small number of numbers (depending on how many people visit the site at the same time), it is possible to analyze all the factors that led to the conversion to a phone call, that is, the alleged key phrases, the source of the users’ transition, which they called what they did on the site and which account was the user's visit to the site. Also, phone call data can be transferred to Google Analytics and other web analytics systems, which allows you to set goals in the system for phone calls. Thus, for analytics of advertising campaigns, it is possible to obtain an even larger volume of source data. As a result, ad campaigns can be better optimized. In addition, in this method, an additional number is allocated that is necessary when traffic increases. The essence of dynamic call-tracking is that when making a call, the program captures not only the traffic channel, but also the keywords by which the request is generated, which provides in-depth data analysis ^[1] . The disadvantage of dynamic call tracking is the decrease in the accuracy and reliability of the data received, since the pool of phone numbers is relatively small and all numbers are in continuous rotation between different advertising elements, the same number can be assigned to many sessions of different users over a very long time interval . When the system registers a call, it has to make a choice from a large number of potentially suitable candidate visitors who once saw this number from the moment the tracking was launched. The final choice is made according to probabilistic-behavioral models. According to various estimates, its average accuracy is 70–90%. The first company to offer a dynamic call tracking service in Russia is Calltouch ^[2] .

There is also a combined method, which is a combination of dynamic and classical technologies. This method allows you to use dynamic substitution of numbers on channels (which need to be analyzed in detail), and classical (where it is enough to analyze at the source level) ^[3] .

The semi-automatic method is the promotional code method, which does not involve the use of phone numbers. The essence of the method is that for each site visitor on each page of the site, a unique code is placed (as an option, a promotional code that gives a discount, etc.). The code is unique for each page viewed, that is, if the same visitor opens the same page several times - he will see different codes. During the call, the operator asks for the promotional code, and thus it becomes possible to associate the promotional code not just with the visitor's session, but with the specific location of this session. The semi-automatic method provides the most accurate data. Moreover, it is not fully automatic and requires manual work (operator's question), which significantly limits the scope of its application ^[1] .

With the use of a call tracking tool, the ability of marketers to calculate the return on investment is growing significantly. With a click-through scheme for paying for advertising and contacting customers by phone, it is almost impossible to calculate the real cost of attracting a customer . Many advertising sites abuse this by selling clicks that are never converted to orders (clicks from botnets , popups ). By analyzing the number of calls from each advertising channel, it becomes obvious the feasibility of marketing investments for each channel, which makes it possible to effectively redistribute the budget. Also, by tracking calls, there are great opportunities for experimenting with advertising.

Despite a number of advantages, call tracking has its limitations and disadvantages. So, for example, the need to use separate phone numbers for each advertising channel leads to the “erosion” of the main phone number (large businesses, as a rule, use “beautiful” phone numbers consisting of repeating numbers). Also, if the task is to monitor the effectiveness of a large volume of traffic, the cost of dynamic call tracking can be very impressive, but if the bulk of the calls to the company are applications or purchases through the basket (not calls), the economic feasibility of using call tracking is rapidly falling. In a situation where 80% or more customer calls do not take place over the phone, but through online forms, as a rule, the capabilities of web analytics are sufficient to evaluate the effectiveness of advertising and the use of call tracking is optional.

The disadvantages include the error in the analytics of incoming calls, when users after a single session can save a replacement phone number that they saw on the site for a certain period of time, and later dial this number, which will no longer be tied to the same visit session in the system and in analytics, it’s impossible to correctly bind the user’s transition source and bind it to the order if it was completed.

There are a number of algorithms that affect the accuracy of dynamic call tracking and are today industry standards.

1. Algorithm "Woodpecker - number dialer". The essence of the algorithm is that the system regularly automatically dials phone numbers from the substitution pool. The end user will not notice them, but in case of inoperability of the phone number will receive a notification. If during this test the call does not go to the call tracking platform, this phone number is temporarily excluded from the substitution. As soon as the number becomes available, the system automatically includes it in the dynamic substitution pool.

2. "Selective call tracking" algorithm. This dynamic call tracking algorithm allows you to use a specific number pool for users from a given region or traffic source, as well as limit the substitution under these conditions.

3. Algorithm "Protection against repeated calls" The peculiarity of this algorithm is that if a call passes from a phone that has already called, then information on this transition is not displayed in the analytics. However, if the same user is on the site at the time of the call, then this transition is added to the user's history.

4. Algorithm "Windhover (" kestrel ") - extension of the session on the movement of the mouse." Thanks to this solution, the phone number is assigned to the user as much time as he actively uses the site (mouse movement is taken into account), and a fixed time is fixed for fixing the number upon completion of work. If the user is distracted and returns to the site after 25 minutes, his session has already ended, and the phone number has been freed. But as soon as the site visitor continues to continue to move the mouse on the site, the code will be updated. If the number previously assigned to it is free, it will be shown to the user, if not, a new one will be highlighted, but the main point is that in this way the session will last and the transition source will be determined correctly.

5. Session Bonding Algorithm This solution makes it possible to set a certain time in the call tracking settings during which extended sessions will be combined into one transition. This method also takes into account that if the user returns to the site from another advertising channel, but within the specified time, this transition will already be considered new.

6. Algorithm "Hybrid Sessions" This algorithm takes into account the different activities of users on the site and thus fixes the phone number at different times. Such a solution makes it possible to save a pool of telephone numbers, reach a larger number of active users, assign them numbers for a longer period of time, and thereby increase the accuracy of call tracking.

The first call tracking services began to appear in Russia and began their development in the 2010s. In 2011, the Target Call service was launched, which was created as one of the tools of the Yandex.Metrica site analytics system. In 2012, Mikhail Fedorin launched the Calltouch web analytics platform , which also includes a call tracking tool for evaluating call tracking performance. Also in 2012, Dmitry Kudinov launched the end-to-end analytics service CoMagic. In 2014, Callibri was opened by Ivan Shkir in Yekaterinburg, and her first service was call tracking. The main market players that provide dynamic telephone number swapping and accounting of call statistics of advertising channels also include Alloka, Roistat, Mango Telecom , Ringostat, PrimeGate, QUON, Zadarma .