Data-driven marketers rely on facts, not assumptions. We all want to trust our analytics but what happens when your analytics configuration is broken? As head of Analytics at In Marketing We Trust, I’ve worked with large brands including eBay, Coca Cola and Expedia to name a few and other medium and smaller businesses. The sad reality is that when I start working with them, most analytics configurations are broken one way or another. In this post, I’ll show you how to conduct a Google Analytics audit in under 30 minutes.
The less the data is reliable, the harder it is to take action and to base your marketing decisions on facts.
Google Analytics Audit
Independently of the scope of work we have with our clients, we will always start by auditing and assessing their current analytics configuration. We can not take the bet of basing our future decisions on broken analytics. And I believe that nobody should take that bet.
Now, if we were to do a full analytics audit (>60 points to cover + delivery of measurement plan), this would require a budget that not every company is ready to unlock straight away.
That is why we developed different levels of a Google Analytics audit. Each type delivers a different bucket of insights but we want to provide as much value as possible. To each need, to each budget, there is an ‘audit’. Usually, one unlocks the next level if errors are discovered.
Initial analytics health check
This is what we are going to cover in this article. It usually takes less than 30 minutes. This will provide you with a good idea of your analytics state. It is not going to cover everything and you won’t have a full understanding of the whole configuration. But this will enable you to quickly detect any major issues. While you don’t want to find any major issues, unfortunately, not finding anything wrong does not necessarily mean that your analytics is set up correctly.
Data Integrity and accuracy report
This is the second level of auditing that we perform. It takes 4 to 8 hours depending on the complexity of the setup. It enables us to score the current configuration (from 1 to 100).
- Data integrity: Is the data being tracked properly?
- Data quality: Can we add granularity to improve insights?
- Data maturity: From a strategic approach, are we tracking things that are crucial to the business?
Full analytics audit
This is part of our analytics implementation framework which is composed of 4 phases:
The audit is performed during the discovery phase. This helps us identify gaps and fixes but also opportunities for improvement. Coupled with the client’s knowledge and requirements, the audit will help us to put together the best measurement plan.
Initial analytics health check: 30 Minute Google Analytics Audit
This is the process that we use internally.
Whenever we start working with a new client, someone in our team will go through the following 7 steps. This article will mainly focus on conducting a Google Analytics audit, however, the basic process also applies to other analytics tools.
- Understand the environment
- Define tag deployment method
- Validate Google Analytics general settings
- Validate the sources of data
- Review Channels and Campaign tagging
- Review existing Events report
- Review Conversions
This 30 minute Google Analytics audit does not answer and cover every question we may have but will provide you with a general understanding and overview of the existing setup.
1. Understand the environment
Understanding the environment is the first step. This is obvious but fundamental. By knowing what tools the marketing team use, we learn what is important and what matters (at that particular moment in time). What is the source of truth?
Any easy way for me to find out the tools they use (without asking), would be to use an extension such as Build With.
In the example above, in a matter of seconds, I know that:
- 2 analytics tools are used: Amplitude and Google Analytics. I will need to figure out later which tool is used as the primary data source (their source of truth).
- Marketing campaigns are potentially run on Facebook and Linkedin. This means that during the tracking assessment, not only will I verify whether GA is properly setup, but also that the right data (conversions) is being collected for these advertising tools.
- Hotjar and Inspectlet are used. Qualitative data points are being collected along with quantitative data points (heatmaps, user recording sessions, click maps). I can infer that this client is already analysing user behaviour and tries to optimise user experience using qualitative and quantitative data.
Other extensions such as Tag Assistant by Google or Facebook Pixel Helper will help me troubleshoot Google tags including Google Analytics, Google Tag Manager and Facebook tags, respectively. I will get a sense of what and how data is processed and collected.
2. Define tag deployment method
By knowing how the data is currently collected (how the tags are currently deployed), I will define our capacity of updating, changing and fixing the tracking. This is important to understand how easy or hard it is to change the current implementation. It is also the reflection of the client’s digital maturity and how the data collection is managed.
- Is the implementation of tags native (i.e. directly in the source code)?
- Is a tag management system in use?
- Do they use a plugin/integration through a third-party tool to deploy Google Analytics or other tags?
For Google Analytics, the extension Tag Assistant by Google can help. You can also go through the source code of the page by yourself and search for tags and snippets of code.
In the source code, I detected that the legacy Google Analytics tracking code is being used. The GA tag is natively deployed which means that if any change is required, we will need to contact IT which complicates things and slows down any fixes or future implementations.
3. Validate Google Analytics general settings
Most of our clients use Google Analytics as their primary analytics tool. We typically start by conducting a Google Analytics audit. The integrity and accuracy of their analytics data in Google Analytics is often a reflection of the rest of their analytics tools. If GA is not properly configured, there’s a very good chance the other tools aren’t either.
Here are the 3 main things that I will check in the administrative panel of Google Analytics:
- Review the main property architecture and views
- Check existing integrations
- Review Goals, Filters and Channel Settings for the main reporting view
Questions to ask:
- Are we following best practices?
- Do we have a clear naming convention?
- Can we easily identify the main reporting view?
- Do we have a testing and a raw view?
- Do we have reporting views dedicated to specific teams or regions?
- Can we understand it without additional explanations?
- Is there any native integration in use?
- Do we use Google Ads, Adsense or other tools?
- Was the channel grouping modified?
- Is there any active goal?
- Is every conversion set as a goal?
- Is e-commerce supported?
- Is the view using any filter?
The idea is to detect if there is an obvious mismatch between the current configuration and the needs while getting a sense of what’s important.
This view does not use any filter. There is an important risk of corruption of data: internal traffic, UAT or dev sites data, bot traffic, fragmented data …
This view has 13 active goals. All have been active in the last 7 days and it seems to match with what I know from their site. There is no red flag at this stage.
4. Validate the sources of data
This fourth step enables us to validate the data we see in Google Analytics. We have to make sure that the data shown in the Google Analytics reports is only coming from legitimate websites and from all the sites we want to report on.
- In GA, access Audience > Technology > Network
- Change Primary Dimension to Hostname
- List the different Hostnames being tracked on the view
Questions to ask:
- Are all hostnames legitimate?
- Is data being collected from unexpected domains or subdomains?
- Are all sites covered?
- Is any site missing?
This is a recurring problem. More than people think. We regularly observe developer and testing environments polluting the data. Sometimes there are missing sites, e.g. in the case of a subdomain dedicated to the deployment of campaign landing pages.
Location (the original location of the traffic) also has to be validated. Conduct a rapid check to ensure that we are not facing any obvious bot traffic coming from very specific locations.
- in Google Analytics access Audience > Geo > Location
- Review the incoming traffic by location
- Review generic engagement metrics such as bounce rate, conversion rate, pages/sessions and conversions
More than 45% traffic on this Australian company site is coming from the United States. The average bounce rate for the US is 98.92%. The engagement seems abnormally low compared to other countries. Is this traffic legitimate? If not, this is biasing the overall performance of the site and might lead to making a marketing decision based on misleading data.
5. Review Channels and Campaign Tagging
It is not possible to report on performance without proper tagging. This is so important yet this is often missed!
By reviewing the channel report, you get a sense of the marketing activity and can detect errors in tagging that could lead to misleading decisions.
- Has the default channel grouping been modified?
- Is (Other) channel receiving a lot of traffic (>10%)?
- Is there any channel driving an unexpected amount of traffic, conversion or showing particularly low engagement?
- Has Direct Traffic fluctuated outside the normal range?
Direct Traffic was showing important spikes of traffic. By breaking down the performance by landing pages, we could identify that more than 15% of the Direct traffic was in reality, traffic coming from Newsletter campaigns.
The (Other) channel received 24,611 sessions (7.25% of the traffic). Every session that is not respecting the defined rules for GA default channel grouping ended up in the (Other) Channel.
By digging into the (Other) channel, we observe that the majority is coming from email campaigns. If we want to properly report on email campaigns, we will have to use Source / Medium and Campaign dimensions rather than rely on the existing channel grouping.
6. Review existing Events report
The Events report will give us a good idea of what matters the most. If the team took the time to think about data collection, define business objectives and work on the measurement plan, the most interesting and important interactions should be tracked. In other words, you should see these interactions in the Event report.
- Access Behaviour > Events > Top Events
- Review the existing event category, actions and labels
- Review the data collection over the last few months to see any drastic change
- Map event with goals when applicable
Questions to ask:
- Do any of these events trigger a goal?
- Are there any unfamiliar trends?
- Are all my events collected through Tag Manager?
- Are there other sources? Booking engine, CRM, CMS (through plugin)…
On the homepage for this client, a new plugin is displaying social media image tiles. This plugin proposes an integration with Google Analytics which enables it to collect information about what tiles were seen by users. Instead of implementing a tag through GTM, the integration was used. 26.79% of the total amount of events tracked in GA were sent by this plugin. Not only did this event impact the bounce rate for the homepage (down to 0), but it also inflated the number of users in Australia by 15% as every other country’s site was using the same plugin, i.e. using the same integration and sending the data to a unique GA property (the Australian GA property).
7. Review Conversions & Ecommerce in GA
Defining goals is THE fundamental component of your measurement plan. Conversions (including Ecommerce) are the translation of the business objectives into measurements. It enables us to measure and report on the most important interactions happening on your site.
You not only need to verify that the necessary goals exist, but also that they are set up properly.
- Access the Admin panel for the main view
- Access the Goals configuration panel
- Review every active goal: type, triggering conditions
- Access the ‘Ecommerce settings’
Questions to ask:
- Are there any goals?
- Does the list of active goals seem complete?
- Are we covering the totality of what we should measure?
- Are we missing obvious goals?
- Are the goals properly triggered?
For an e-commerce site:
- Are the GA e-commerce reports being used and configured?
- Do we use enhanced e-commerce?
- How much can we rely on the dollar value?
- Can we compare the reported revenue with real data?
- Are the data layers providing the expected information?
Even though users can purchase online, the client’s analytics view had no goal and e-commerce was not active. There is a clear gap in measurements and this will prevent us from reporting on performance. We will need to spend some time defining a complete measurement plan and make sure that we can report on the main conversions (in alignment with their business objective).
In this client’s analytics view, enhanced e-commerce was active. Just by comparing the 2 metrics: revenue (sum of the revenue of every transaction) and product revenue (sum of revenue of every product sold), we observed an important difference depending on the metrics used.
The metric Product Revenue reports more than 2 times the actual revenue. If we were to report on product performance and how much revenue each product is driving, we will have to take into consideration that this particular metric is biased and not reliable.
30 Minute Google Analytics Audit
The large majority of Google Analytics configurations are broken in one way or another. Running checks, conducting a Google Analytics audit and assessing your state of analytics is a mandatory and recurring task that every marketer should do. We need reliable data to be able to make the right decisions for our marketing.
By running these 7 checks, you should be able to get a quick sense of your data integrity and accuracy. Reviewing and questioning every configuration and data point you have will help gauge where you stand. This initial 30 minute Google Analytics audit won’t cover everything, but it will help highlight potential issues, misconfigurations or gaps in your measurements.