Who is guilty? Measure last click and first click conversions
Companies that work with online advertising, more and more attention is paid to each step of the user, because such information is not only interesting from the point of view of research, but also important for evaluating the advertising channel and reallocating costs to achieve a greater number of conversions.
Remember how the move in billiards happens: the cue hits the ball, the ball hits another ball and eventually one or more balls hit the pocket. The result is important - to count the ball. What turned out to be more important for victory: hitting a cue, a cue ball, or a random ball that pushed the winning ball into the pocket? Hardly lovers think about it.
Attribution is the rule for distributing the value of a conversion to individual points of interaction in the process of achieving this conversion. Basically, it comes to the rules for assigning individual channels to achieve a certain goal. In electronic commerce to purchase or registration, as a rule, there is a chain of steps and actions performed on the site. These are the so-called microconversions, each of which has its own weight in achieving the goal. With the increase in the number of different channels and devices, there is an increasing need to track the entire chain of interactions on the way to conversion, as well as to use different ways of value attribution. Google Analytics has long introduced various attribution models into its interface.
Each user on a commercial site must achieve a goal and usually this goal is set up and counted in Google Analytics as a conversion. Conversion is usually achieved through multiple interactions between the user and the advertiser. For example, when choosing a car, the buyer can go to the site with an advertisement, then go to it from the bookmarks and eventually go to the remarketing banner and sign up for a test drive. To analyze such chains, Google offers to set up various attribution models in the Analytics tool.
')
Last interaction - in this model, 100% of the value of the conversion is assigned to the last channel in the chain of interactions. This model is good for those where the purchase decision (other action) is made immediately.
The last indirect click is a type similar to the previous one and used by Google Analytics by default for all reports. When it is used, direct visits are ignored and 100% of the value of the conversion is assigned to the last channel in the chain of interactions. This method is quite simple and does not take into account a certain part of visits to the site. It is best used in cases where the user is as close as possible to the final transaction (a reminder about the renewal of a subscription to the media, about forgotten goods in a basket or unpaid purchase for technical reasons).
Last Click in AdWords - 100% of the conversion value is assigned to the last click on an AdWords ad in the interaction chain. This model is used if you have an AdWords advertising campaign and users from your ads come to the site to complete transactions.
The first interaction - 100% of the conversion value is assigned to the first channel in the chain of interactions. This attribution model is less commercially applicable than all others. It is better to use it during the initial entry to the market to track the dynamics of interest in a new brand or a new company.
In the linear model, all channels in the conversion sequence are assigned the same value. It is usually used in cases where the user is exposed to advertising throughout the entire purchase cycle (another transaction) and all points of interaction with a potential client are equally important. Obviously, this situation is rare in commerce, so this model can be used, for example, to analyze advertising on a site or publish it on a blog.
Taking into account the duration of interaction - a model that works when a client makes a purchase decision in a short time. Google Analytics in the statement states that the model is based on exponential decay. This complex term came to Google Analytics from nuclear physics and describes the essence of the attribution model as precisely as possible: the closer to the conversion is the interaction point, the more valuable it is. The remaining points lose value with an increase in the time interval. The model is applicable, for example, to analyze purchases resulting from promotions.
Positional attribution is a hybrid of the First Interaction and Last Interaction models. Instead of assigning all the value to the first or last channel, you can divide it between them. Usually it is distributed as follows: 40% for the first and 40% for the last channel and 20% for the rest. Such a model is usually used if it is interesting to track all points of interaction: from the first interest to the last action that led to the conversion. Perhaps this is the closest to real life model - it is applicable in almost all areas of business.
The listed models are standard solutions offered by Google Analytics. However, users can create their own attribution models. And for this they need to own a number of terms related to attribution models.
Multichannel sequences are sets of user paths before conversion. For example, you advertised in AdWords, the buyer saw it, went to the site and left it. Then, remembering the name of the site, he hammered it into the address bar of the browser, but made a mistake when typing. The search engine guessed what he needed and issued the site first in search results. The user went and bought. The sequence was formed: advertising in AdWords → site → search → site. But this does not mean that the search leads to purchases, the most likely trigger was the ad. Based on such a sequence, it is worthwhile to build an attribution model. Multichannel sequences can be represented graphically in the Google Analytics interface (conversion paths):
In the attribution model comparison tool, you can compare various types, including custom ones. This will make it possible to determine the most optimal sequence and find out which channel is more efficient and requires additional investments for the growth of commercial indicators.
As the experience of RealWeb shows , many marketers stop at one of two options: either last click or first click. And each of them is right.
Previously, when measuring the user's path was a difficult process, the last click was always used (last click) and the marketing channel that brought the click was considered the most effective. If we neglect a number of factors, such a model can be represented if a letter with a discount coupon was sent out and a purchase of an item already known to the user was made directly from the letter. Closer to reality - the activation of a coupon in the event that this is a macro goal. That is, there are situations when it is justified. However, the last click does not reflect the user's entire path.
The last-click model allows you to identify the sources that directly failed (prompted) the client to perform the conversion. But if the company is faced with the task of attracting new users and raising brand awareness and interest in products, then it will be useful for it to use models that distribute greater value to the sources that stood at the beginning of the site’s interaction chain (first click) .
Attribution assessment is incredibly helpful. Use the method of the first and last click, assign weight to intermediate steps - and you will see exactly how your users get to the conversion. If you once bought something online, it’s easy to imagine your customer’s behavior on the Internet. For example, he chooses a mobile phone. Upon request, he gets on your site, looks at prices, leaves, reads reviews on the Market, returns, leaves, reads reviews on Mobile Review, goes to Yandex.Pod service and there he sees your store ad in Yandex.Direct, but with the model of interest, clicks on it, goes to your website and places an order. What exactly led the client in this case? We can say that the last ad served as a trigger, but it is possible that his choice started from your site, because you have excellent SEO.
In light of the multi-channel nature of the user's path to conversion and the variability of ways to achieve it, certain rules should be followed when working with attribution models.
Conversions in web analytics are divided into two types:
Analysis of micro-conversions allows you to evaluate marketing efforts, the effectiveness and characteristics of channels, sources of transitions. Each step towards conversion, that is, each microconversion, may or may not have value.
After working with attribution models, the next step in determining the effectiveness of a source may be a joint account of the achievement of macro- and micro-conversions with its participation. For this, a final indicator is formed, which can also be used to optimize the operation of the source. In this case, macro and microconversions are assigned different weights. You can also add an account of different attribution models.
That is, a formula is formed like:
1 * purchase (last-non-direct) + 0.5 * subscription mailing (last-non-direct) + 0.3 * 1 * purchase (first click) + 0.3 * 0.5 * subscription newsletter (first click).
We encourage our clients and our interested readers to build on the length of the decision-making cycle of their clients. The longer this process, the more interactions it involves, the more attention needs to be paid to the initial and intermediate stages. This approach will allow more efficient use of the advertising budget, not relying on just one online advertising and not spending every penny on SEO. Only a comprehensive promotion can give a qualitative result.
Remember how the move in billiards happens: the cue hits the ball, the ball hits another ball and eventually one or more balls hit the pocket. The result is important - to count the ball. What turned out to be more important for victory: hitting a cue, a cue ball, or a random ball that pushed the winning ball into the pocket? Hardly lovers think about it.
Attribution is the rule for distributing the value of a conversion to individual points of interaction in the process of achieving this conversion. Basically, it comes to the rules for assigning individual channels to achieve a certain goal. In electronic commerce to purchase or registration, as a rule, there is a chain of steps and actions performed on the site. These are the so-called microconversions, each of which has its own weight in achieving the goal. With the increase in the number of different channels and devices, there is an increasing need to track the entire chain of interactions on the way to conversion, as well as to use different ways of value attribution. Google Analytics has long introduced various attribution models into its interface.
Attribution models Google Analytics
Each user on a commercial site must achieve a goal and usually this goal is set up and counted in Google Analytics as a conversion. Conversion is usually achieved through multiple interactions between the user and the advertiser. For example, when choosing a car, the buyer can go to the site with an advertisement, then go to it from the bookmarks and eventually go to the remarketing banner and sign up for a test drive. To analyze such chains, Google offers to set up various attribution models in the Analytics tool.
')
Last interaction - in this model, 100% of the value of the conversion is assigned to the last channel in the chain of interactions. This model is good for those where the purchase decision (other action) is made immediately.
The last indirect click is a type similar to the previous one and used by Google Analytics by default for all reports. When it is used, direct visits are ignored and 100% of the value of the conversion is assigned to the last channel in the chain of interactions. This method is quite simple and does not take into account a certain part of visits to the site. It is best used in cases where the user is as close as possible to the final transaction (a reminder about the renewal of a subscription to the media, about forgotten goods in a basket or unpaid purchase for technical reasons).
Last Click in AdWords - 100% of the conversion value is assigned to the last click on an AdWords ad in the interaction chain. This model is used if you have an AdWords advertising campaign and users from your ads come to the site to complete transactions.
The first interaction - 100% of the conversion value is assigned to the first channel in the chain of interactions. This attribution model is less commercially applicable than all others. It is better to use it during the initial entry to the market to track the dynamics of interest in a new brand or a new company.
In the linear model, all channels in the conversion sequence are assigned the same value. It is usually used in cases where the user is exposed to advertising throughout the entire purchase cycle (another transaction) and all points of interaction with a potential client are equally important. Obviously, this situation is rare in commerce, so this model can be used, for example, to analyze advertising on a site or publish it on a blog.
Taking into account the duration of interaction - a model that works when a client makes a purchase decision in a short time. Google Analytics in the statement states that the model is based on exponential decay. This complex term came to Google Analytics from nuclear physics and describes the essence of the attribution model as precisely as possible: the closer to the conversion is the interaction point, the more valuable it is. The remaining points lose value with an increase in the time interval. The model is applicable, for example, to analyze purchases resulting from promotions.
Positional attribution is a hybrid of the First Interaction and Last Interaction models. Instead of assigning all the value to the first or last channel, you can divide it between them. Usually it is distributed as follows: 40% for the first and 40% for the last channel and 20% for the rest. Such a model is usually used if it is interesting to track all points of interaction: from the first interest to the last action that led to the conversion. Perhaps this is the closest to real life model - it is applicable in almost all areas of business.
The listed models are standard solutions offered by Google Analytics. However, users can create their own attribution models. And for this they need to own a number of terms related to attribution models.
Multichannel sequences are sets of user paths before conversion. For example, you advertised in AdWords, the buyer saw it, went to the site and left it. Then, remembering the name of the site, he hammered it into the address bar of the browser, but made a mistake when typing. The search engine guessed what he needed and issued the site first in search results. The user went and bought. The sequence was formed: advertising in AdWords → site → search → site. But this does not mean that the search leads to purchases, the most likely trigger was the ad. Based on such a sequence, it is worthwhile to build an attribution model. Multichannel sequences can be represented graphically in the Google Analytics interface (conversion paths):
In the attribution model comparison tool, you can compare various types, including custom ones. This will make it possible to determine the most optimal sequence and find out which channel is more efficient and requires additional investments for the growth of commercial indicators.
So last click or first click?
As the experience of RealWeb shows , many marketers stop at one of two options: either last click or first click. And each of them is right.
Previously, when measuring the user's path was a difficult process, the last click was always used (last click) and the marketing channel that brought the click was considered the most effective. If we neglect a number of factors, such a model can be represented if a letter with a discount coupon was sent out and a purchase of an item already known to the user was made directly from the letter. Closer to reality - the activation of a coupon in the event that this is a macro goal. That is, there are situations when it is justified. However, the last click does not reflect the user's entire path.
The last-click model allows you to identify the sources that directly failed (prompted) the client to perform the conversion. But if the company is faced with the task of attracting new users and raising brand awareness and interest in products, then it will be useful for it to use models that distribute greater value to the sources that stood at the beginning of the site’s interaction chain (first click) .
Attribution assessment is incredibly helpful. Use the method of the first and last click, assign weight to intermediate steps - and you will see exactly how your users get to the conversion. If you once bought something online, it’s easy to imagine your customer’s behavior on the Internet. For example, he chooses a mobile phone. Upon request, he gets on your site, looks at prices, leaves, reads reviews on the Market, returns, leaves, reads reviews on Mobile Review, goes to Yandex.Pod service and there he sees your store ad in Yandex.Direct, but with the model of interest, clicks on it, goes to your website and places an order. What exactly led the client in this case? We can say that the last ad served as a trigger, but it is possible that his choice started from your site, because you have excellent SEO.
In light of the multi-channel nature of the user's path to conversion and the variability of ways to achieve it, certain rules should be followed when working with attribution models.
- Test various attribution models.
- Change models in case of changes in the sales scheme.
- Use different attribution models for different channels.
- Compare attribution models and choose the best ones.
- Distribute marketing costs based on the weight of each microconversion.
- Experiment with your own attribution models.
Another way to determine the effectiveness of the source
Conversions in web analytics are divided into two types:
- macroconversion - the ultimate goal that must be achieved by users (purchase, ordering, creating lead)
- microconversion - steps to the goal that the user accomplishes (for example, registering on the site before placing the order or adding goods to the basket).
Analysis of micro-conversions allows you to evaluate marketing efforts, the effectiveness and characteristics of channels, sources of transitions. Each step towards conversion, that is, each microconversion, may or may not have value.
After working with attribution models, the next step in determining the effectiveness of a source may be a joint account of the achievement of macro- and micro-conversions with its participation. For this, a final indicator is formed, which can also be used to optimize the operation of the source. In this case, macro and microconversions are assigned different weights. You can also add an account of different attribution models.
That is, a formula is formed like:
1 * purchase (last-non-direct) + 0.5 * subscription mailing (last-non-direct) + 0.3 * 1 * purchase (first click) + 0.3 * 0.5 * subscription newsletter (first click).
We encourage our clients and our interested readers to build on the length of the decision-making cycle of their clients. The longer this process, the more interactions it involves, the more attention needs to be paid to the initial and intermediate stages. This approach will allow more efficient use of the advertising budget, not relying on just one online advertising and not spending every penny on SEO. Only a comprehensive promotion can give a qualitative result.
Source: https://habr.com/ru/post/293610/
All Articles