Ever stopped to think how accurate your data is? If you are making business decisions on data, it's obvious that you will want that data to be correct. With businesses losing an estimated $100 of revenue for each dirty record, data collection and accuracy becomes increasingly important to business success. It doesn’t matter if you're collecting ecommerce sales data or gathering metrics about your website performance and its visitors, it needs to be reliable so that you can make the appropriate adjustments to your strategy.
Chances are you are already using Google Analytics as one of the main sources to evaluate the performance of your marketing performance with KPIs. In an ideal world this would work flawlessly and you would never have to manipulate data because someone didn’t realise the UTM Campaign tags were case sensitive or a developer pasted the Google Analytics script tag onto the site when you already had it in Google Tag Manager.
Unfortunately small errors like these happen from time to time, but it’s when you are aware of the issue and what you do to fix it that becomes the difference. The adage of ‘garbage in, garbage out’ certainly rings true here. To ensure you get the most out of Google Analytics, you have to ensure that your data is as clean as possible.
Not all metrics in Google Analytics are equal, as some of them are more susceptible to giving misleading information from poor implementation and website structure. Knowing where these metrics go wrong is key to unlocking the full power of Google Analytics so that you can more accurately analyse your site's data.
Google Analytics uses logic to calculate certain metrics and dimensions from the raw ingested data. But this has lots of limitations if you rely on that data to make decisions and you aren’t aware of how the logic works. For example, Google Analytics classifies traffic into Channels so you can understand what is driving visitors to your website. One often misunderstood channel is Direct, which as its name suggest is traffic from users going directly to your website. However, it's important to understand that if Google can’t determine the source of the visit then it will classify the channel as Direct.
This might seem harmless at first glance, but take the scenario where you are actively doing email marketing and do not tag your links with UTM parameters, then you lose the ability to know that your email campaign is the source of website traffic and the recent increase in conversions.
Large spikes in Channel specific traffic would be a suitable candidate for monitoring anomalies. Although the Anomaly being detected won’t directly solve the issue of your data not being tagged, it can lessen the impact if caught early.
Google Analytics has a wealth of information when you log in, but the ability to log in to each client’s account frequently is a luxury not afforded to most marketers. If a change were to happen on the website from an update and as a result your pageviews collection was affecting and dropped, you would have no way of knowing unless you log in and check. As a result, you may not have enough time to catch the problem and correct it before data loss becomes a real problem.
To combat this, Google Analytics allows you to set up Custom Alerts for certain occurrences, such as a change in the number of visits/visitors, pageviews, bounce rate, average visit duration, percentage of new visits, etc. But you'll only get these alerts if these metrics change by a certain pre-set percentage. This often causes issues when your website receives a small amount of traffic during the early hours of the morning and your thresholds can be met, giving you false positives. Quite often this leads to alert fatigue and the system that tries to warn you about issues becomes ignored.
Jepto takes a different approach and applies a Machine Learning model for Anomaly detection so there is no need for static percentages or absolute values. Upon setting up a metric to be monitored for Anomalies, you will see how the algorithm has identified anomalies over the past 3 months. Giving you the ability to adjust the sensitivity up or down, which can be used to increase or decrease the amount of alerts triggered. For Agencies the sensitivity can be turned up for high profile or important clients.
There are a plethora of Google Analytics audit tools available on the market today, and they exist for a reason. Most Google Analytics accounts are setup and then rarely checked or adjusted as the business objectives shift. Simple things like setting the correct Time Zone and internal traffic IP filtering affect the data being captured and the ability for the coveted 'insights' to be derived. Take the Time Zone setting, if you have a single location business you can simply select your location so that website visits are recorded at the time local to you, not a big deal. However, consider that you have multiple locations and they operate in different time zones.
Your ability to understand traffic on an hourly basis, which is often used to help determine the optimal time for sending email marketing campaigns, across your different locations is not so straight forward. It's important to remember that view settings are non retrospective so if you do make a change it will only affect data moving forward, so setting an Annotation might be helpful to know when changes have been made.
If the Custom Alerts that you have set up in Google Analytics do get triggered you will get an email to let you know. Unfortunately the information is not very intuitive and is hard to understand if you have multiple accounts. Alerts are also user specific, so you cannot see alerts that have been setup by other users which may cause unnecessary alerts being setup when collaborating with your marketing team.
Google has improved on the Custom Alerts with a feature they call Insights, which was released in early 2018. Unfortunately it hasn't seen any noticeable updates since. Insights also utilises a Machine Learning algorithm for Anomaly Detection (Bayesian state space-time series model), however there are no email notifications when they are created. Therefore you must login to Google Analytics and then look to see if any new Insights have been created. Whilst this is a step in the right direction it falls short of being useful because of the lack of notifications and the ability to adjust the settings, making it unsuitable to rely upon for critical sites in its current form.
Modern Marketers are using a variety of collaboration tools, like slack to communicate as a team, and as such emails might not be the preferred channel for these alerts. Bringing these alerts into your real time chat allow for quicker awareness of an issue and reduce the time to be resolved. The actions you take after a data issue incident will either allow you to be proactive with your client or stakeholder and get on the front foot, or erode confidence and create doubt about your ability to be up to the task.
Jepto understands that you have your own way of working and so Alerts can be sent by a variety of channels:
Anomaly detection won't prevent you from making mistakes, but it will give you the opportunity to limit the damage done.
Google Analytics is a great tool on its own, but Jepto can make it work for you. Get started with a free 14 day Trial to see what anomaly detection of your Google Analytics data can do to improve your data accuracy.
Thanks a lot for your feedback! If you’d like a member of our support team to respond to you, please send a message here
Oops! Something went wrong while submitting the form.