Why you can’t trust your website analytics: Part Two

Written by Edward Kay on December 18, 2018

Part one of this report revealed how tools such as Google Analytics are not accurate. Technology such as ad blockers means up to 78% of your website visits are not being recorded.

In this second and final part we look at the implications of these findings and the information associations can still trust when making data-driven decisions.

Affected tools and data

It is not just Google Analytics data that are being under reported. Other third-party services that rely on the user’s browser to send them data can be affected by ad blockers. These include:

  • Remarketing code – used to target ads to people who have already visited your site. e.g. Google Ads remarketing tags and Facebook Pixel.
  • Marketing automation tools – used to manage marketing activity based on website interactions e.g. Drip, HubSpot, Pardot, Eloqua, ConvertKit, Infusionsoft
  • Content personalisation and split-testing tools – used to adjust website content based on user data and/or to test content variations e.g. OptinMonster, RightMessage, Optimizely
  • Other web analytics tools – e.g. Mixpanel, Piwik, Segment, New Relic, CrazyEgg, Hotjar
  • All code managed through Google Tag Manager

None of these tools should ever be considered completely accurate. There are many reasons why they don’t work all the time: e.g. dropped connections; users having Javascript disabled; someone using multiple browsers/devices; navigating to a new page before the code has executed. This has always been the case.

The problem now, based on these data, is that the proportion of our visitors for whom these tools actually work can be tiny. Less than 22% in the case of the PSA.

I now believe trend-based tools, including analytics, are now largely meaningless.

Tools that work on single sample points, e.g. heatmaps and visitor recordings, are less affected. We just have fewer data points to use.

The value in web analytics comes from using relative changes (i.e. trends) in the data to drive further action.

This works fine if we are confident the tools are capturing data from a large and consistent proportion of our audience. We have now shown this is not the case.

But there is more to it than this.

The proportion of visitors being tracked is not consistent. All of the datasets evaluated here show strong declines in the percentage of visitors being captured by Google Analytics – especially Scottish Association of Landlords and Year Out Group.

For Year Out Group in particular, the rate of decline in tracking exceeds the visitor growth rate. This means Google Analytics is showing a downward trend in visitors when visitor numbers in the logs are growing.

What to do about ad blockers and analytics.

It can be tempting to follow the path of various media outlets and put up a fight against ad blockers. But I believe that’s a losing battle.

I love my ad blocker. I’m not going to turn it off just so the sites I visit get better stats.

My motivations for using an ad blocker are better security, removing interruptions and improving speed. These mirror the findings from PageFair’s 2017 report into ad blocker usage. Only 6% cited privacy as a reason for blocking ads.

Threats from malware are increasing so the ad blocker is here to stay.

I recommend two routes of action:

1. Keep the analytics, but understand the limitations

Web analytics still have a place, despite these serious shortcomings. They still provide some useful data on the visits that are tracked, such as measuring the relative performance of different content, e.g. Google Ads. (If someone is clicking on your Google Ad, they probably aren’t using an ad blocker).

We just need to remember that we’re only seeing a relatively small portion of the overall traffic.

The proportion of traffic not captured by your analytics is unique to your site. Running a log file comparison like this will provide an indication of the accuracy of your data.

2. Focus on the metrics that actually matter

Web analytics has long been a black hole. It is full of fascinating but often useless stats. It takes sustained discipline and effort to use the information effectively.

A far more effective strategy is improving the metrics that actually matter. For the membership sector these include:

  • Membership numbers
  • Membership renewal rate (or membership churn)
  • New member renewal rate – i.e. the number of memberships that continue beyond their first period
  • Event registrations
  • Event attendance
  • Enquiries received
  • Website logins
  • Use of self-service functions
  • Email opens and clicks
  • Member benefits used, such as:
  • Member-only resources accessed
  • Purchases made with member discounts

The good news is none of these metrics (with the possible exception of email opens) are impacted by the use of ad blockers.


This is a long article with a lot of information. To sum up the key points:

  • Standard tools used to measure website usage are missing lots of data due to the use of ad blockers.
  • Blocked tools include Google Analytics (which also reports conversions back to Google Ads), Google Tag Manager, Facebook Pixel and Google Ads remarketing code.
  • The proportion of visitors not being measured can be massive – over 78% in the case of the Professional Speaking Association for example.
  • There is a strong downward trend in the proportion of visitors that can be measured.

This presents challenges for all organizations. I recommend:

  • Understanding and monitoring the level of misreporting of your users. This will inform the level of trust you can have in any tools affected.
  • Focusing on the most important metrics. e.g. email clicks, membership numbers and event ticket sales. These are not affected by ad blockers.
  • Consider using alternative technology. e.g. website log file analysis to understand website use. But only if these data provide insights that drive useful actions.

This article was first published by Tall Projects. Please see the supporting notes for details of the methodology used and other important information.