About the company. Samolet Plus is a service that creates convenient digital products and multifunctional offices within walking distance to cover all user needs: from choosing an apartment to living comfortably in it. Igor Morozov, Head of Digital at Samolet Plus, talks about the details of the case.
How it was: downtime due to standard callback
Our client ordered a call back on the website. Then the standard mechanics of the Calltouch call back worked : the system called the call center, when the operator answered, it called the subscriber. The subscriber picked up the phone, and the two calls were connected. However albania phone number resource the CC employee often had to wait a very long time for an answer. The waiting time reached 2 minutes, and the client might not pick up the phone at all.
What we did: implemented reverse, removed voice notifications
We have implemented a reverse callback Calltouch. How it works: first, the system calls the subscriber, and when the subscriber picks up the phone, the call goes to the call center.
We have also sped up the process of connecting the subscriber with the operator. According to the standard scheme, when the operator picks up the phone, a short message is broadcast, for example: "A callback was ordered from the site samoletplus.ru, connection with the subscriber has been started." Or, if the connection is set up by pressing a key: "Press "0" to connect with the subscriber." We asked to remove these notifications. As soon as a connection is made with the subscriber who ordered the callback, we immediately automatically connect the call center operator. The waiting time is 2-3 seconds.
The broadcast of messages for operators has also been cancelled. Previously, the employee listened to introductory information about what kind of call it was - direct or return, from which widget it was coming. For example: "Call from a client, from a mortgage application form, to accept the call, press "0".
This did not affect the call recognition in any way, because our contact center is designed to receive calls only from Samolet Plus clients. The operator does not care whether he is receiving a return or direct call. In addition, as practice has shown, no one listened to the message to the end. The operator pressed "0" as soon as the connection opportunity arose, because it is important for him to process the call faster.
This does not mean that operators do not receive any information about the client now. The data does not disappear anywhere, it comes to the CRM through the integration of Calltouch and the site. And at the moment of connection, a card automatically pops up in the program, which indicates from which callback widget the call is coming. At the moment, the mechanics have been implemented so that the card shows what services the client was interested in - rent, mortgage, repair or something else.
Non-stop communication
We listened to calls before the changes were implemented, and we listen to them now. And we see that we have managed to optimize and improve the process. Firstly, the employee does not need to press the "0" or "1" button to connect with the subscriber. Secondly, the number of calls processed per hour has increased.
When calls come one after another non-stop, the operator enters a good working pace, and productivity increases. And how it was before: the phone rang, the employee picked up the receiver, listened to the information message, pressed the connection button, listened to the beeps... That is: either you communicate with clients without pauses at a good pace, or you listen to IVR or beeps for 2 minutes, which slows down the pace.
Callback
Increase the number of requests from the site
New clients get 50 minutes free
Read more
Evaluation Method: Traffic and Criteria
We evaluated all callbacks that came to Calltouch. But it turned out that we only implemented a reverse callback for calls from contextual advertising and the site, but did not switch calls from VK lead forms. These calls are easily tracked and visible in analytics. So, we evaluated all callbacks for two periods:
January`23 — February`23. First a call to the Call Center, then to the subscriber — the reverse mechanics have not yet been introduced. In February, the numbers are not quite "clean", since this month the reverse was launched in test mode
March'23 — June'23. First a call to the subscriber, then to the call center — a reverse callback is launched for calls from the site and context. There are distortions here too, since VK advertising is not covered.
The results were considered according to two criteria:
successful call,
connection wait less than 3 sec. and more than 3 sec.
A successful call in our terminology is a call lasting more than 2 min. 15 sec. Why is that? The call can last up to 2 min., then Calltouch disconnects. And we add another 15 sec. to eliminate the possibility of error. If the call lasts more than 2 min. 15 sec., then there was definitely a dialogue with the client.
It is clear that among calls lasting over 2 minutes and less than 2 min. 15 sec. there may be short conversations with clients. For example, a person picked up the phone at the last moment, answered "not interested", and hung up. But, firstly, the share of such calls is small. Secondly, such calls can also be considered unsuccessful, because there was no normal conversation with the client.
Analytics of successful calls
Waiting for the connection between the client and the operator for less than 3 seconds and waiting for more than 3 seconds is the second criterion of interest. It helps to check how quickly the connection occurred, how long the operator had to wait until the client answered.
We looked at the statistics on successful calls. In January-February, the CC employees spent 178 minutes (~3 hours) waiting to be connected to the client. I would like to emphasize that these are statistics only on successful calls. In less than 4 months of using the "reverse" mechanics, the wait for successful calls was only 71 minutes in total. Moreover, this time is for forms that have not been reconfigured.
Statistics improved already in February, because this month we launched reverse in test mode (in February there were 49 calls with a wait of less than 3 seconds). And since March, a clear result is visible: calls with a wait of more than 3 seconds have decreased to a minimum, and the share of calls with a wait of less than 3 seconds prevails. The share of calls in which the connection time between the CC and the client is less than 3 seconds has grown to 86%. At the same time, the subscriber also does not wait long for the connection. This is the first proof that the reverse call technology works successfully.
End-to-end analytics
Evaluate the effectiveness of all advertising campaigns in one window from click to ROI
Invest in advertising that brings in customers
Read more
Failed Call Analytics
Failed calls were analyzed in two sections:
1. Operators' time saved. This is the time that the robot spends dialing. Previously, this work was done by call center employees (the same "parasitic load"), so this indicator is important to us.
Note. The line "Waiting for more than 3 seconds" since March (and partially, since February) are calls from VK advertising. Let me remind you that we only connected the reverse mechanics to this channel in July. If we had done this immediately, then, most likely, since March these values would have tended to zero or to ones, like the top line. That is, the wait would have lasted a total of 0.5 - 1 minute per month.
Waiting time for failed and erroneous calls is the time that operators spent calling customers.
In January-February, operators spent 1025 minutes waiting, or more than 17 hours. This is an average of 8.5 hours per month. And for 4 months of using the "reverse" mechanics - 325 minutes, or less than 5.5 hours. This is an average of slightly more than one hour per month.
Purity of the experiment: confirmed
The results seemed impressive to us: changing one mechanic - and such a simplification of the process and reduction of labor costs. To confirm the purity of the experiment, we assessed the overall consistency of the call traffic.
Let's say that in January-February we would have launched network campaigns for 100 million rubles, and would have received low-quality traffic or some programmers. This could explain the high percentage of refusals and waits in the first period. Then, if in March we would have turned off all networks and driven up high-conversion hot traffic, this could have justified the decrease in the number of long waits. However, we did not change the traffic, which is confirmed by the figures:
The total number of callbacks has a positive MoM trend.
The proportion of successful and unsuccessful calls remained virtually unchanged.
The traffic consistency is the same, so the experiment is clean. External factors could have influenced the result, but not significantly, since this is not visible in the figures. The figures show that it was after the introduction of the reverse that the call processing process was simplified and labor costs decreased.
