Think About an Audit for Analytics

I’m impressed with the ability of web analytics tools to spot trends and identify success/failure modes. Really, I am. And if you are reading this, perhaps so are you.

But what I am totally not impressed with is the amount of inaccuracy and uncertainty many enterprises accept as “normal” even as they rely more and more on digital properties for marketing purposes.

The causes for uncertainty fall roughly into two main categories.

One: smaller companies lack the internal resources to test and verify their own analytics infrastructure. Lack of expertise and lack of exposure to best practices can hobble any chance at really understanding true traffic patterns. Often, marketers at smaller outfits simply “don’t know what they don’t know” and rely on generic implementations to present accurate data. Almost as often, that doesn’t really work.

Two: large enterprises suffer from fragmentation of effort. While expertise may be resident or available for purchase, there is rarely an analytics “ombudsman” who is charged with ensuring accuracy. Instead, accuracy is kicked back and forth between (usually) marketing, IT and in-house/third-party developers, with no group  in possession of both the raw technical skillset and the objectivity to make definitive statements about accuracy; nor the mandate to make recommendations towards better accuracy.

The result is that uncertainty and  inaccuracy afflict nearly every segment of the market.

Why this is bad:

-when you think your numbers are accurate but they are not, then you are making decisions based on false information.

-when you are pretty sure your numbers are inaccurate, you refrain from making decisions based on the numbers: thereby wasting the entire effort.

The cure really is rather simple in theory.

It’s called an “audit“.

An audit entails the following:

-a review of the desired reporting outputs

-a review of the current reporting outputs

-a thorough inspection of all tags, custom code, calls made to the analysis engine, action scripts, log files, url parameters and so on

-by an expert (!)

-a report with a spreadsheet showing what the page-level data capture schema should look like for every page and every dimension

-a detailed map of how this can be achieved (e.g. instructions for developers)

-extensive QA of the tags and parameters once the audit-recommendations have been implemented

-a final report to verify accuracy of reporting

Commonplace Problem

It gives me the shakes to think about how inaccurate data is so awfully common in the analytics world. It makes me slightly impatient (it’s a flaw of mine) to understand how simple it really is to perform an audit, and to compare that against how infrequently audits are typically performed.

My advice to all digital marketers is: if you have not done an analytics audit recently, get one done ASAP. It will help you sleep like a baby.

 

Leave a Reply

Leave a Reply

Your email address will not be published. Required fields are marked *