Evaluation of the quality of advertising traffic

At one of our client sites, we have provided the ability to track calls to the company through the feedback form and calls. We used a simple mechanism: each visitor is assigned an individual number, which is visible when entering the site. If, after visiting the site, the customer made a call to the company, the sales manager will recognize it and enter it into the database. If the visitor contacts through the feedback form, the number was recorded automatically. Using the identifier obtained in this way, you can determine where the person came from, on what request, which pages of the site you looked through, etc.

We will not go into the technical details of the implementation of the system, we will present the conclusions that were made on the basis of the collection of this information.


')
We calculated the number of calls to the company from various contextual advertising campaigns. The statistics fragment for the period from March to May looks like this (screenshot from the internal accounting program):



This is not a complete statistic, but for illustration it is quite enough.

Let's take a closer look at the statistics of requests from Yandex.Direct over the period 1.03-31.05. First, let's calculate how much it costs to turn on contextual announcements from Yandex and from the Yandex contextual network.

Baseline data (General performance of the advertising campaign for the period 1.03-31.05):

Consumption	158982.60 rub.
Transitions	4016
Cost per click	39,59 rub.
Number of accounted applications	46
Of which the search	32
Sites	14
Total hits online	1268
Number of collected and tracked numbers	442

It can be seen that the number of calls is not equal to the number of numbers collected: unfortunately, the sales department did not always specify the individual number of the caller. Perhaps in the future we will be able to collect more accurate data, at least we are trying to work with the customer sales service in this direction.

Determine what percentage of the numbers were collected:

442/1268 * 100 = 34.86%

In order to calculate the final cost of circulation, we will have to assume that the remaining 65% of requests are distributed in approximately the same way. This, of course, is not true, but, unfortunately, we do not have more accurate data. Let us estimate the true number of advertisements in Yandex.Direct:

The absolute number of recorded hits: 46 / 0.35 = 131
Of which the search: 32 / 0,35 = 91
Sites: 14 / 0.35 = 40

Further, according to statistics Yandex Direct, we look at how much money was spent on the search and on the contextual network:

Search:
Consumption: 142383.60 rubles.
Transitions: 3107
Price per click: 45.83 rubles.

Sites:
Consumption: 16599.00 rubles.
Transitions: 909
Price per click: 18.26 rubles.

Now let's calculate the final cost of contacting the company:

The cost of handling advertising: 158982.60 / 131 = 1213.61
Of these, the search: 142383.60 / 91 = 1564.65 rub.
Sites: 16599.00 / 40 = 414.98 rubles.

Conclusion

Contextual networks work, and work quite well. And in our case, the cost of circulation was two and a half times lower than with the search.

Therefore, they must be used. Moreover, recently this situation has developed with this company: in the low season, competitors with their inexhaustible budgets began to make large investments in advertising. For some requests, the rate at the entrance to the guarantee reaches 10 USD. In this case, everyone ignores the context network. We, having budget constraints, instead of fighting for high positions in search, make small bets per click and quite comfortably feel ourselves in the contextual network, receiving transitions and appeals for much less money from there.

Also, as a result of the experiment, it is clear that Runner and Advords also bring appeals to the company. At the same time, both search and partner networks work.