New buyers to a website, for example, an online store, can come from anywhere: from social networks, email newsletters, instant messengers, advertising campaigns, by clicking on a banner, etc. To correctly distribute the advertising budget between these acquisition channels, you need to see the statistics their effectiveness.
For such purposes, an analytics script is embedded in the site template. But it doesn’t always work correctly. And sometimes it “flies”. And also the setting of the advertising campaign can affect the objectivity of the data. How do I get the correct cut across all sources used?
First, go beyond automatic data collection. Yes, search engines have a good analytics system, but it is impossible to configure it so as not to miss anything. You need to additionally analyze and check everything manually.
Secondly, take into account the human factor. For example, not the most honest work of competitors, which can influence the effectiveness of advertising campaigns.
Thirdly, periodically check whether everything is configured, spelled out and installed correctly. Especially after technical changes – switching to new protocols, design edits, scaling the site structure, etc.
The key to effective automated analytics is UTM markup
If there is no UTM markup at all, then the built-in analytics system simply will not see the difference between referral sources. She will not have a list of these sources either. Therefore, there is no need to talk about any objectivity of the results of automatic analysis.
However, if the markup is done, this does not mean that all transitions will be credited to the very channel through which they were carried out. Here the point is already in the correctness of the choice of the parameters of this markup (dynamic and static). Ideally, it should be dealt with by a specialist who knows how to fill in labels without errors. Among the latter:
- autogeneration UTM without verification;
- violation of the order of labels (hierarchy);
- typos in parameter names.
System for evaluating visitor actions
Several terms need to be introduced here:
- attribution (in the context of web analytics) – assigning primary and secondary value to a visitor’s actions;
- attribution model – a system of rules by which this value is assigned;
- type of attribution (or attribution) – the choice of a set of rules that the automatic analytics system will rely on when distributing data across sources.
Google Analytics and Yandex.Direct use not only different types of attribution. Technically (logically), they have differently implemented the process of attribution. To this it remains to add that the site owner is also not free from the risk of making a mistake. For a specific case, he can simply choose the wrong model for distributing visitors by source.
- Yandex attributes leads by referrals:
- the last one;
- the last one from Yandex.Direct;
- the last significant one;
- Google attributes on a number of metrics:
- last click;
- first click;
- taking into account the prescription of the transition;
- taking into account the position;
- in linear model mode.
Not just different behavioral metrics are used, but also different logic of their accounting. That is why you should not mix the analysis of traffic sources on Yandex and Google. It should always be done separately. Perhaps as an experimental change in attribution models to obtain an objective cut.
For clarity: bounce rate in Yandex and Google
A striking example of the difference in approaches to assessing visitor actions is the bounce rate. Both services consider it as the ratio of the number of bounces to the number of all visits. At the same time, Google Analytics understands a refusal as a transition, viewing 1 page and triggering 1 event. And Yandex – the transition and presence on the page for less than 15 seconds.
Accordingly, where Yandex Direct shows 5% of bounce rates, Google Analytics will give all 25%. A beginner who does not understand the logic of data accounting might get confused. It is much easier to consider these indicators separately, given the specific actions to which they indicate.
The script, thanks to which the program – the site analytics system – gains access to the statistics of its visits, can simply “fly off”. Most often this happens when the design of the resource changes, it is transferred to another management system (CMS) or a more advanced protocol. Simply put, any technical adjustments – from CRM integration to changing the design theme or installing new plugins – can lead to data loss.
If, in analytics, the transitions on a number of previously effective channels for attracting traffic have sharply “sagged”, it is this script (analytics code) that should be checked first. Another sure sign that something is wrong with him is a sharp skew of traffic in favor of one specified source.
It is clear that if the script is not added to the template at all, then there is no point in talking about the objectivity of the data collected about its visit. In this case, all that remains is to ask each customer where he came to the store from in order to estimate which channels to invest more funds in.
“Intrigues” of ill-wishers
The sharper the competition, the more “gray” or “black” methods competitors will use. On the one hand, experience has repeatedly shown that clicks on ad clicks from competitors de facto does not increase traffic to your own website. On the other hand, there are always personal motives and a thirst for experimentation.
However, clicks are far from the only method of unfair competition that an advertising campaign may face. Is there some more:
- promotion of other metrics, except for impressions (refusals, for example, when a competitor clicks on an advertisement himself, and after spending 5 seconds on the site, he leaves);
- outright fraud (creation of sites with a very similar name);
- calls from spoofed numbers, etc.
Defending against the dark side of competition is not as easy as we would like it to be. We need special software (if clicks are suspected, although it remains a rarity today), monitoring the behavior and statistics of a competitor, dialogue with customers, checking for the presence of “imitators” in the network – sites and companies with the same name as the advertiser or signs).