Episode 547: Analytics API Request In Monday Morning (2)

Sunday, August 6, 2017

Ranking on popular pages

In the last article, I introduced you a way of automatic generation of "weekly reports" by automated communication with "Google Analytics Reporting API".

The point was the automated Step (Service Task Addon) which automatically retrieves data aggregated in the following rule.
  • Dimension: ga:hostname, ga:pagePath, ga:pageTitle
  • Metrics: ga:pageviews, ga:sessions

In this Step, "List of web pages that got a lot of access" is gathered into a multi-line text. If you set to "filter" something like "ga:pagePath=~/blog/" you can also automatically get the ranking of "web pages under blog folders".

Notification email that is embedded these texts would be very useful information for Marketing team.

Tens of thousands of other patterns of aggregation method

But now, I would like to aggregate data of Google Analytics in more different perspectives.

While in the previous example, ranking was retrieved with the above "three dimensions" and "two metrics", Google Analytics has about 260 dimensions and about 230 metrics. In other words, a wide variety of data aggregation is available, by changing its combination.

For example, when you aggregate not only on the aspect of site contents ("Behavior (BEHAVIOR)") but also "AUDIENCE" or "ACQUISITION", you should be able to extract information such as "What kind of people are accessing?" or "What kind of site has it been derived from?".

Reference: Dimensions & Metrics Explorer

[Website Operation Report 2]

Reading "curiosity of customers"

In the Business Process Definition introducing here, one automatic Step (Service Task Addon) is added, and the second automatic Step generates a list by "content of dimensions" that different from the first one.
  • Dimensions: ga:networkDomain, ga:pageTitle
  • Metrics: ga:pageviews, ga:sessions

I suppose people who use Analytics every day would understand, but this "ga: networkDomain, ga: pageTitle" expresses an aggregation method (perspective of aggregation) for seeking for "accesses from which network to which page".

Of course, "source information" is aggregated by Internet providers, so it may not be possible to extract significant information in the case of sites where access from ordinary households occupies a lot.

Whereas, if the site has many "access from the corporate network", you will be able to extract information such as "Which company is looking at which manual page?", or "which company is viewing which case study page?"

That will be very useful information for the CS team and Sales team.

Incidentally, in order to filter out "access via internet provider", the following filter setting is effective.
  • Eliminate "(not set)"
  • Eliminate "unknown" network
  • Eliminate networks of ".net"
  • Eliminate networks of ".ne.jp"
* ga:networkDomain!=(not set);ga:networkDomain!~unknown|.net|.ne.jp

<Email Sample>

Increasing productivity by information sharing

Regarding businesses that have the possibility of productivity improvement through "regularly sharing the latest information", I would like to positively consider about combination of "dimension" and "metrics", or "filter". Practically, you will only have to try this and that for combinations which will be likely to provide useful data, in "custom reports" etc..
  • Which company accessed with an old browser?
  • Accessed from which link in which external site?
  • To which pages accessed from Facebook and Twitter?

Once the format of the latest information to be shared is decided, then it is necessary to acquire "Service Task Addon" that matches it.

If you do not find "Service Task Addon" that matches your aggregation policy, you need to create it separately. If there is someone with JavaScript knowledge in the company, it should be made in-house (by arranging existing Addon, etc.), but even if there isn't, you can get it in about one day when you make request to crowdsourcing, etc,

Note that, the "Service Task Addon that communicates with the Google Analytics API" which introducing here has its dimensions and metrics fixed (hard-coded). In other words, all you can set up by config are as follows.
  • View ID: Access log to be aggregated
  • Number of items to display: Number of items in ranking list to display
  • Filter: Filter for ranking list display (optional)

In addition, the reason why it is not in a "general-purpose structure" is mainly due to the fact that the "display method of aggregation results" differs slightly depending on the combination of them.

for( var i = 0; i < reportObj.reports[0].data.rows.length; i++ ){ 
  rankingStr += reportObj.reports[0].data.rows[i].metrics[0].values[0] + "\t"; 
  rankingStr += reportObj.reports[0].data.rows[i].metrics[0].values[1] + "\t"; 
  rankingStr += reportObj.reports[0].data.rows[i].dimensions[0]; //★Eliminate separator for being clickable URI★ 
  rankingStr += reportObj.reports[0].data.rows[i].dimensions[1] + "\t"; 
  rankingStr += reportObj.reports[0].data.rows[i].dimensions[2]; 
  rankingStr += "\n"; 

[Website Operation Report 2:"1. Report preparation" screen]

[Data Items list]

[Free Download]
<Similar Models>
<<Related Articles>>

[Japanese Entry (ε’Œζ–‡θ¨˜δΊ‹)]