Mobile Menu


Online marketing has the edge over the traditional forms of advertising: it allows an in-depth examination of its efficiency. Many marketers still make a pretty serious mistake by treating each marketing activity as a standalone and independent part and as a consequece, separately assess efficiency of various promotional activities. This error stems from the failure of two very important issues. First of all, various marketing channels (traffic sources like SEO, AdWords, Facebook, mailing) interchange and influence with each other. Secondly, before our customers buy something they pass through several different states of mind: first, there is the need, then they get acquainted with their desired item, they look for opinion about its versions, select offers, finally they purchase it. After the purchase they share opinion about their product. Different marketing channels interact on customer's various states of mind.

Analytics demonstrates how the various marketing channels interact with each other!

There are two groups in Conversions reports being problematic for beginners. Today we will learn how to use them. There are respectively:

  1. Multi-channel funnels - you'll see the path of your customers having the first visit and the visits ended with a purchase,
  2. Conversion attribution - assign a different and better success attributing model to individual marketing channels.

Let's begin with MCF (multi-channel funnels).


A lot of invaluable answers can be found in this group e.g. how does the customer behave - a user that brings conversion, namely bought an item or achieved one of our goals - beginning from the first visit on out website. How did he initially found us? How many times and from what sources did he entered the website? Which marketing channels were involved in conversion?

Let's have a look at the Overall report:

A combined number of conversion is available: sales and goal conversion rate. Second information is something new: assisted conversion. It refers to the number of conversions credited as a part of particular marketing channel but includes only cases where this channel wasn't the last one - in other words, the channel comprising visits to purchase.

Sounds quite complicated but it may be clarified by a simple example. Assume that a customer who had bought a product had visited our website 5 times in total. For the first time it was via AdWords, second time via Facebook, third time another AdWords entry, fourth through mailing and the final fifth again via Facebook. The result is one conversion registered for the Facebook channel. AdWords and mailing channels are recorded as assisting channels. It may be assumed that if the company hadn't set their ads through other two means the purchase had had fallen through. As a result, AdWords and mailing are also counted as channels that generate conversion.

It is presented with a graphic of overlapping collections.

Another reports displays specific conversion paths: The most important conversion paths:

By this example we may conclude that the overall customer journey starts through means of free and paid search. They buy, enter the site directly. By reading the reports of the last channel success assignment it becomes obvious that the user deriving from direct traffic is pointed as the best customer. However, the above reports show something completely different. Before the customer enters the store's website directly, he becomes acquainted with it by search engine results page visit. In other words, if it wasn't for the visibility at Google search engine there wouldn't have been no conversion.

Assisted Conversions report tells us only about channels being a support for conversion. However, we may see which conversion assistants are salient.

Because of this report nothing is as simple as it used to be. It's no longer easy to eliminate marketing channels. Each channel gives a final value. Size of this value can set the basis to determine the marketing budget.

On this occasion it is crucial to mention one very important issue. If one of the channels is more supportive and generates less sales, than its content should be adapted to this role.

By this example it is noticeable that social networks play a large part in supporting the conversion but if it comes to sales, it's null. This means that Facebook ads (the shop has a Facebook profile and uses its ads) should be based upon information and solution content without any product descriptions. Issue of this kind is extremely wide and requires separate chapter to be covered.

Timelapse report shows distribution of path time duration to the conversion number and value.

Conversion Path Length is the last meaningful report to inform about how many times does the user interact with our website from the first visit until the purchase is done:

The conclusion you should get from the above example is that the store has the biggest day's earnings on people who had paid at least 4 visits. What's even more important: customer had at least 4 opportunities to find his desired deals elsewhere.


The same question may be analyzed from a completely different perspective. Instead of keeping track of individual paths we can change the credits distribution to conversion. Standard model adopted in Analytics distributes 100% of conversion credits to a channel that brought a customer directly before the purchase. As we already know, in such a fast-living world such a model does not correspond to reality at all. Although there is no perfect model but we have to choose some alternatives.

Here's a list of standard attribution models in Analytics:

  1. Last interaction - 100% of credits for the last channel even if it's direct traffic,
  2. Last Non-direct Click - a default model - all credits are being distributed among a channel other than direct traffic from which a customer had converted directly.
  3. Last AdWords Click - 100% of credits are distributed among the last click on an AdWords ad even if a customer had came from other channels - a good model to assess profitability of AdWords campaign,
  4. First Interaction - all credits distributed among the channel that brought the customer on the website,
  5. Linear - it equally distributes credits among each channel that took part in bringing the conversion about,
  6. Time Decay - assigns credits to channel being the closest to conversion,
  7. Position Based - allocates pool of credits to the first and last channels. You can set the percentage of credits. The standard is 40% for inputs deriving from initializing channels, another 40% for channels being before the conversion and 20% for median channels.

Report comparing the attribution models allows us to collate various models and to analyze marketing channels from different perspectives.

Let's compare 2 models: standard "last non-direct click" and "linear":

You can easily spot that if you flatten the credits equally to all customer visits, the essence of specific channels is considerably modified. Social Media traffic drastically decreases whereas the direct traffic reliably mounts up. Was it the case of our example? No. By analyzing the multi-channel path we've concluded that direct traffic is result of other activities performance. One should test diverse attribution models for a reliable comparison. I encourage to operate on your own statistics. Please note that our examples provide shops with relatively small conversion number. Analysis of this kind are best for large numbers.

Finally, I've left another question concerning the reports configuration. Until now, we've studied all conversions: sales and goal conversion. However, it can be changed to analyze sales or selected goals only. You can use the menu located under the title of the reports:

  1. Conversion menu - decide which conversions should be included in analysis,
  2. Type - type of channel for analysis - you may select AdWords as the only channel to be analyzed e.g. campaigns affecting conversion in various models,
  3. Validity period - how long shall a path last - 90 days is the maximum value.

Multi-channel Funnels have athe same configuration.


Mainly, before deciding on the budget strategy and the scope of work of various marketing channels. For example, before finishing your AdWords campaign try to find how has it benefit the store sale and whether it has brought any negative effects. This way you'll easily come up with a channel worth of spending more resources on. You'll also find out whether a certain campaign or a promotional activity results in conversion.


© 2016 - professional SEO for e-commerce