7 Mistakes to Avoid with Google Analytics

As a digital marketer, one of the first steps in taking on a new client is to do an audit of their programs and their analytics to see how they’re performing and understand how they’re tracking and measuring. One of the things that we commonly encounter in these audits is that companies are not making good use of Google Analytics (GA).

No matter what you have in your martech stack, GA is still very important. Other platforms give you a lot of insights into what happens once someone becomes a lead or a customer, but GA is going to give you the most pre-conversion data. That’s critical knowledge for helping fill the top of the funnel. And even though it’s free, the stuff that you can do with GA is every bit as sophisticated as any platform on the market. Here are seven common mistakes most companies need to fix to improve not only their use of Google Analytics, but their entire analytics program:

Mistake #1: Looking at Analytics in Silos

The Marketo person looks at the data in Marketo while the Google Ads person looks at Google Ads data and they each make decisions based on what they see, but neither has a complete picture. If you want to make the best decisions, it’s really important to have a data analyst who understands analytics and measurement across all systems in a business. Ideally this person would be part of the marketing team, but what seems to happen is that there’s one analyst serving many groups and marketing is low on the list.

I read recently about a company whose data science team was building a model to look at incrementality across all of their marketing efforts. Now that’s a dream team! But for a lot of marketing organizations, just having a dedicated analyst to get all your systems set up and connected and configured right is a big step forward. Then you can start to move towards doing some more advanced analysis.

Mistake #2: Not Speaking to Business Objectives

Your KPIs should be tightly tied to things people care about, and you need to speak their language, not analytics speak in order to get support and resources for data initiatives.

For example, let’s say you’re doing B2B lead gen. In Q1, you need X number of customers acquired at Y acquisition cost. That would be the primary KPI; working backwards you’d have secondary KPIs around MQLs and SQLs.

You can configure Google Analytics to record “events” that would indicate whether or not you’re on the path to meet your goals, such as form fill or form abandonment, and also things such as scroll depth. Then, you can drill in and try to figure out why.

Here’s an example: Leads from all channels were as expected except from Facebook, which were unusually low. Drilling into all of the pre-conversion data for that channel, we found we drove 100 people to a landing page, and only two of them actually made it to the area on their screen where they could actually convert. Obviously, we needed to fix the layout of the page.

If I just said, “Hey, we need to set up a scroll depth event in Google Analytics,” that sounds like geek speak. But if you lead with the objective and explain how the scroll depth event is going to help you predict if you are going to hit your goal and help you diagnose your problem if you don’t, then you’ll have no problem convincing people you need the resources to do this.

Mistake #3: Not Using a Tag Management System

There are many tag managers on the market, but Google Tag Manager is the most commonly used. It’s also free.

A tag management system (TMS) is a tool that helps you manage all of your different tracking code snippets in one user interface, and it helps you install tracking code much more efficiently. Instead of manually putting a snippet on every page where you want to trigger an event or a conversion, you can install it one time in the tag management system and roll it out to all the pages where you want it.

Doing that manually obviously takes a lot more time and development resources and is prone to errors. Even if you are implementing GA through a marketing automation platform such as HubSpot or Pardot, you still need a tag manager.

There is much more to a tag management system than just installing Google Analytics. Many of those automation platforms will have a section to install your base page view pixel, but that’s it. There are tons of other tags that need to be installed on a website and you want them all in one place.

There are also a lot of cool things you can do in the TMS, such as build custom conversion actions that you can then deploy to any marketing platform — LinkedIn, Facebook, Google Ads, etc.

Another very cool use case: You can use JavaScript APIs from many different platforms in conjunction with the TMS. For example, I was working on email sequencing for a SaaS company that offers a free trial. There’s a welcome sequence, and after that is complete, we wanted to be able to send different messages based on a person’s activity in the tool. By assigning each user an ID in the email platform and on the tool, you can create a tag in the email platform to indicate whether they’re using the tool or not and automatically send them the relevant email.

This is an advanced use case that requires a little development work, but it gets to the power of what you can do with a tag manager. It’s really good for marketers to manage basic tracking across your whole ecosystem; it can help break down data silos across platforms, and once you’ve got that under your belt, it has really advanced capabilities that developers can use.

Mistake #4: Counting Duplicate Transactions

This is very common in e-commerce businesses. If you’re looking at product sales in GA — and you should be — you know your sales in GA are usually a couple of percentage points lower than actual because you can’t track absolutely everybody. But you should be able to reconcile the two within a small margin of error.

What you can do to combat this is build a custom report to look at your transactions by transaction ID. If you see multiple transactions for the same ID, you’re probably overstating your sales in Google Analytics. That means you’re going to make investment decisions based on overstated data.

Mistake #5: Not Having Discrepancy Reporting

If you have transactions recorded in a database, in GA, and in other platforms, and you have marketers making decisions off of these different data sources, your implementation should include automated discrepancy reporting. This is a report that pulls from the database and all these channels and monitors whether there’s a reasonable and consistent discrepancy level — within plus or minus five percent is a good rule of thumb. The report should be configured to send automated alerts when it gets outside of that, so you don’t accidentally optimize using the wrong data for three months because nobody got around to reading the report in their inbox.

This does require custom development in Python and SQL, but with API connections it’s not that hard. It’s a four-digit project that could pay for itself very quickly. I always think of it as a protection plan — if you’re spending the money to get your analytics in shape, this will quickly alert you if something’s gone awry.

Mistake #6: Not Excluding Internal IP Addresses

Just about everybody knows you need to exclude your internal, partner and vendor IP addresses from your traffic counts, but few companies have a solid implementation.

The challenge is collecting and maintaining the data, which is an ongoing effort especially if you’re a growing business, adding new partners, vendors, and offices, and now, having a lot of people working remotely. You have to develop a process for collecting the data, keeping it up to date and making sure it gets updated regularly in GA.

Not doing this correctly can hurt you in a lot of ways. If you’re building remarketing lists, you’re going to have a lot of extra people on them. If you’re doing any advertising using CPM (cost per impression) bidding, you’re going to be wasting money serving ads to your own people. You’re going to get low click through rates, which hurts your quality score and raises your advertising costs. Internally, it gives you the wrong picture of the campaign, and you’re optimizing using bad data.

Mistake #7: Not Testing Your Tracking

Any time you set up an event or conversion, be sure to send a couple of people through the funnel and make sure all your pixels are firing and you’re counting correctly. One very common mistake people make is putting the conversion pixel on a landing page, instead of the thank you page — the page they reach after they’ve paid or signed up or registered. I’ve done this myself. There’s nothing worse than telling your boss you’ve gotten 103 conversions when what you’ve really gotten are 103 people hitting the landing page. Other pixel placement errors can result in double counting or not counting, so always be sure to test your setup anytime you deploy new code.

If you’re making any of these mistakes with GA, or all of them, you’re definitely not alone. We’re all trying to be more data driven, but it’s not easy getting there. It can be really overwhelming to start thinking about things such as discrepancy monitoring or building JavaScript APIs when you’re still trying to figure out what a conversion pixel is and where it goes. It seems so unattainable, but the best thing to do is get started.

There’s so much you can do with today’s tools, but you have to have good data, and you have to have complete data. It doesn’t help to have great data coming out of your CRM if you can’t go back and troubleshoot the top of the funnel, and that means getting your GA implementation in order.

If you think your GA implementation might be messed up, now is the time to fix it. It’s similar to personal health. We all know that if you’re healthier and feel good, you’ll be better at everything you do. It’s the same thing with data. Fix it now, and six months from now, you’ll be making decisions with full confidence that you’re working from good, clean data.