skip to Main Content

Google Analytics Reporting

Automated cleaning and reporting of web ad and search data

Inconsistent web analytics data

A marketing team of a large corporate client had issues with tracking the performance of its web marketing activity:  it’s existing process was labour intensive, inflexible and inaccurate

CPRA was asked to automate and improve the process

It’s the data, Stupid

Again, it’s the data.  Multiple users, no formal data model and a difficult data structure.

The marketing data is in the form of an ID and a long string:

cpra_uk_financial_services_regulation_600x300_3p_wallstjournal

But this structure (or rather lack of structure) allows data variations and so inconsistencies:

calimere_point_gb_reg_fs_600x300_3p_wsj

calimere_point_gb_regulatory_fs_600x300_3p_wallstreetjournal

calimere_point_gb_regulatory_fs_600x300_3p_the_financial_times

calimere_point_it_regulatory_fs_600x300_3p_financialtimes

cp_es_financial_services_regulation_160x600_NEW

Business Rules

And this is where the problem starts: it makes consolidation of data difficult; data is inaccurate; difficult to reconcile to other data sources and hard to discern a hierarchy.

And what does “new” mean?  How new will new be six months’ time?

On top of this the client’s existing data aggregation process in excel based off downloaded reports introduced further errors.

The Solution

Three key parts:

  1. Create a process that can flexibly identify and normalise the data from long strings into a structured format
  2. Create an automated data process that feeds a dynamic dashboard
  3. Most importantly, ensure the data model and rules are kept to on an ongoing basis

Whether it’s web data, portfolio data or physical inventory, we’ve seen the same issues and the solutions are similar.

The Data Model

Our data tagging experience was invaluable.  We created a structured data model and a rule set that would take all the client’s data variants and mapped them to the new data standards.

So “es” and “sp” becomes “es” for Spain.

PBOC, PBC and The PBC becomes “the_peoples_bank_of_country”

wsj; wall st journal becomes wall_street_journal

but wsj.dnwsj.dwsjd becomes wall_street_journal_digital

and google affinity newsjunkies is nothing to do with the wall street journal

But such tagging can’t be blind, it has to take context into account.

Automated Data Process

Almost the easy bit.

API connections to google (Search Ads 360; Display & Video 360; Campaign Manager) to remove the manual downloads.

Automated run of CPRA’s data model using our modern data analytics tools.

Output to dynamic dashboard on a secure website.

And either on-premises or on CPRA’s cloud platform.

Regular Communications

Turning off the tap of bad data: regular calls are scheduled to ensure data is on track and to give notice and agree the naming of new marketing activity.

This is helped by CPRA’s automated data process that automatically extracts and highlights i) new data and ii) existing data that has changed.

Such calls take progressively less time as clients and agencies get used to the new data discipline.  Without this it’s was a waste of money for a one time clean.

Benefits

Flexible, easily accessible and accurate information on marketing activity and responses in a timely fashion.

Transparency of marketing spend and marketing effectiveness.  Provides accurate, timely information on the relative performance across regions, channels and formats.

More efficient use of spend; in one case a client’s marketing spend had to be diverted by their agency into crunching the (Inaccurate) numbers.

Back To Top