Turning Common Website Analytics Issues into Measurement Opportunities

By Brian Keefe posted 09-03-2021 11:29 AM


Turning Common Website Analytics Issues into Measurement Opportunities

I am just going to come out and say it: collecting website analytics data can be intimidating. It requires a level of technical know-how to implement, training to configure correctly, and experience to analyze. Platforms like Google Analytics and Adobe Analytics collect massive amounts of data, and it is tricky to figure out not just what to focus on, but what all this data means for your bottom line. It is super common for projects to have some bumps in the road when setting up their analytics or begin analyzing their data, but not every mistake leads to catastrophe. On the contrary, they are opportunities to get better, grow out your website monitoring capabilities, and start to leverage data to improve your website.

The ICF Next Digital Program Optimization team got together and worked out some of the more common “uh-oh” website analytics issues we have seen over the past few years. These are things we have seen happen on almost every website at one point or another, so it is not the end of the world if this has happened to you. This is the chance to take an issue and turn it into an opportunity, so you can get the absolute most out of your website data. Let’s dive in…

Uh-Oh: You have a lot of direct or typed/bookmarked traffic

Explanation: Visits where an analytics platform cannot identify the source of the visit. It is not always people who bookmark your site or type the URL in directly and can be several different use cases. This includes traffic from email clients like Outlook, from a link in a mobile app, and from dark social (e.g., text messages, messenger apps, email). Having a lot of direct traffic can make analysis difficult since you do not know exactly how a user arrived at your website.


  • Examine your acquisition data to understand just how much of your traffic is coming from “direct”. If it is a huge amount of your overall traffic, there might be something amiss.
  • Look for spam or bot influence: If either of these two conditions is true, you may have a bot/spam issue.
    • Segment your direct visits by geographic location, device type, hostname, browser, landing page, and other dimensions to see if a huge chunk of your direct traffic is coming from one particular type of user. In particular, if the hostname does not match your website, you are likely being spammed.
    • Look at the engagement of your direct users (overall and by segments). If you see an incredibly high bounce rate, low pages per visit, and low average session duration, you may have a spam/bot issue.


  • Ensure all digital outreach links have tracking parameters (UTM for Google, . For non-digital promotion, create a vanity URL that redirects to a tagged link, so you can better capture offline influence.
  • Ensure you have hostname filters in place and are blocking all known bots (more on this in a moment).

Uh-Oh: Your website is being hit with spam/bots

Explanation: Visits to your site that are not legitimate. They can be from an automated bot or some other organized spamming effort. This can skew your website data.


  • You see a large spike in traffic that you cannot explain, particularly spikes in traffic coming from “direct”.
  • You meet the criteria described in the “direct” diagnosing section described in the previous point.


  • In the “View” settings of Google Analytics, check the “Exclude all hits from known bots and spiders” box. If you are using Google Analytics 4, this is done automatically.
  • Create a hostname filter to only accept hits from your website (this will exclude a lot of junk data). If you are using Google Analytics 4, talk to a member of your analytics team, as implementing this requires more technical knowledge for the new version of Google Analytics and requires Google Tag Manager.

Uh-Oh: Your analytics tracking code is not placed correctly

Explanation: Website analytics platforms work via a snippet of code that is placed on each page of your website you want to track. A lot of things can go wrong with implementing the tracking code, but the most common issues are having the code added to the website twice or your analytics are just not functioning at all (tracking code is missing or not placed correctly on the page). When the tracking code is placed twice on a page, double-counting happens, pageviews are counted twice and all your data becomes a total mess. When your tracking code is implemented incorrectly or is missing, you just will not receive any data.


  • If you have a sudden massive increase in pageviews and pages per session and your bounce rate drops to near single digits.
  • If you have just created an account and your bounce rate is below 10% and pages per session are above 3 or 4, you should check implementation.
  • You are not getting any data at all coming in.


  • Install the Google Tag Assistant Chrome Browser Plug-in. This plug-in can tell you if Google Analytics (as well as Google Tag Manager, Google Pixels, and other tools) are implemented on your site and implemented correctly. If you are using Adobe Analytics, there is a similar Adobe Analytics de-bugger plug-in that you can use for the same purpose.
  • If you do not have it installed on your site already, install Google Tag Manager and remove all hard coded Google Analytics code (similar suggestion for Adobe’s tag manager product and Adobe Analytics tracking code). Tag management systems allow you to control most of your analytics without the need to bother a developer or adjust your website code.

Uh-Oh: You lose focus on the data that matters

Explanation: There is a TON of data available in most website analytics platforms, so it is easy to get lost or carried away. Instead of reporting on data that can inform optimization opportunities, you end up just reporting all the data you can find.


  • You spend more time reporting data than analyzing it.
  • You never get any feedback from your client on your analytics reports.
  • You do not have time to do any ad hoc analysis or testing.
  • Your regular reporting is looooooooooooong.


  • The backbone of optimization analysis is identifying key performance indicators, data that is tied to the user goals on the website. Work with your clients to work out in plain language the goals of their website, what they hope users will do, and examine what users are currently doing on their site (what the client wants them to do and what they do can sometimes not match). Tying specific data points or user actions to those goals will focus your analysis and reporting.
  • There is “health of a website” data measures you will always likely include in your reports and analysis (e.g., visits, device, geographic location, acquisition), but leverage them in conjunction with your KPIs. For example, if one of the goals of your website is to sign users up for a text messaging program, look at completions in conjunction with acquisition source, device type, engaged vs. non-engaged users, new/return users, and other factors. Try to identify the factors that went into users completing that action and be sure you are strategically segmenting the data.
  • If possible, move your reporting and analysis out of excel sheets and word documents and into some kind of data visualization program like Google Data Studio or Tableau. This will help to automate reporting, reduce data input errors, give clients a more visually appealing report, and free your time to analyze the data instead of just reporting it.

Uh-Oh: You spend all your time on small impact tactics

Explanation: Spending a ton of time and resources optimizing outreach that has a low return on investment.


  • Look at your website acquisition data to really understand what is driving people to your site (and what is driving the important actions on your site). I would bet almost any amount of money, if you are not running any kind of paid media, that organic social or email will be way below organic search when it comes to driving website traffic. Organic social and email are important, and you should work to optimize them for sure, but they are typically not a huge driver of website visits as compared to organic search.


  • Extol the benefits of search engine optimization to your clients using your analytics data and via an SEO audit. Many clients over-focus on email or social media because those are things they can control easily without needing to engage developers or content teams. SEO takes work, but the investment is worth the effort.

Uh-Oh: You are seeing more website users than website sessions

Explanation: Most website analytics platforms break down website traffic into some variation of pageviews, sessions/visits, and users/visitors. Each user can have multiple website sessions and each website session can have multiple pageviews. When you look at your website data though, you somehow see more website users than sessions, which should not be possible.


  • There are several reasons why this might be happening, but a few of the more common reasons are:
    • You have created a custom report and are matching up dimensions (data categories) with the wrong data points. The most common example is trying to view page/page titles with sessions. Page-level data, except for the landing/entry page, does not mesh well with sessions.
    • You are looking at sessions and users by the hour of the day. The way most analytics platforms work, sessions are defined by the length of time a user is engaged with your website. If that session drifts from one hour into the next hour (e.g., 1:50 pm to 2:05 pm), the session may only be counted in the first hour (1 pm) and not the second (2 pm). The user will be counted in both hours though, so users will end up being higher than sessions.
    • You have custom non-interaction events set up in an iframe or have a third-party tool sending non-interaction events to your analytics platform. Non-interaction events are data on user actions that do not start a new website session and do not impact your bounce rate measurement. However, these actions still cause your analytics platform to add to the user metric. This is the trickiest of the use cases, so consult your analytics expert when you suspect this might be the cause of the issue.


  • When creating a custom report, use the “unique pageviews” metric instead of “sessions” when examining data by page/page title. “Unique pageviews” is a good proximal measure of sessions by page.
  • Take care when looking at users and sessions by the time of day and note the limitation in any analyses you conduct.
  • Consult with your analytics expert when it comes to setting up custom events, especially events coming from outside your immediate website environment (e.g., iframes or CRM), as they will be able to make the call on having them be non-interaction or interaction events, as well as how to treat them in any analyses.

Uh-Oh: When I apply segments or look at data over a long period of time, it starts to look weird and some of the results don’t make sense.

Explanation: Sometimes if an analytics program is trying to display a large amount of data (over 500,000 sessions for Google Analytics) or you are applying a lot of different segments or secondary dimensions (basically, breaking down the data by multiple factors), it will only look at a subset of the data and extrapolate out results. This is called sampling.


  • In Google Analytics, you can tell when sampling is taking place by looking at the colored checkmark next to the title of the dashboard page. If it is green, you are looking at 100% of the data. If it is yellow, hover over the checkmark to see what percentage of the data is being considered.
  • You can also tell if data is being sampled if you see a lot of the same numbers across different rows of data. If your data is looking kind of same-y, sampling might be occurring.


  • If you encounter sampling in your data, try reducing the date range for the data to see if that leads to 100% of the data being considered.
  • You can also try removing segments or secondary dimensions, if possible, to mitigate the effects of sampling.
  • If you are using Google Analytics 360, you can download unsampled reports.
  • You can connect Google Analytics 4 and Google Analytics 360 to Big Query, which will allow you to do analyses on your own with the raw data. This can come at a cost though and requires some technical know-how to accomplish.
  • If you really need a free analytics solution that can handle a massive amount of data coming in (and don’t want to pay for Adobe Analytics or Google Analytics 360), there are open-source solutions like Matamo available. Matamo allows you to own all of your data (as it is on your own servers) but requires a lot of developer and analytics know-how to set up correctly.

Uh-Oh: You are reporting demographic data from Google Analytics

Explanation: You have the option to turn on Demographics and Interest reports in the older version of Google Analytics to get data on gender, age, and interest categories. In Google Analytics 4, this is called Google Signals. In a nutshell, this is an additional layer of data about some of your users. The key word here is “some”, as it is only getting data on a smaller sub-set of your users who have opted into or are signed into certain Google features/products. Reporting this data as representative of all your users can be dangerous and lead to misinterpreting your website audience.


  • When looking at demographic and interest report data in Google Analytics, pay very close attention to the % of website users Google Analytics is reporting are included in the report. It is most likely way less than half of your total users.


  • Despite its limitations, turning on Demographics and Interest reports or Google Signals is still probably the right move. Especially for Google Analytics 4, as turning on Google Signals opens a wider array of features and creates a more precise measure of users (since it can link up users who are signed into Google across devices, so if someone visits your website from their phone and then their desktop device, it will recognize that it is the same person).
  • As a rule, do not report demographic and interest report data in formal reports or using exact numbers. Since it is only a subset of your users, it could be misleading. It can be useful though for very high-level insights into your audience, as one piece of information in conjunction with other pieces of data collected by your website analytics platform or outside of your automated analytics (e.g., user testing, surveys).

This post was borne out of discussions with persons who do not do analytics work every day, so it is mainly intended for non-analytics persons who for whatever reason need to do analytics work and analytics professionals who want to share some common pitfalls with their project teams in a consice manner. 


Most Recent Blogs

  • #BreakTheStigma - May is Mental Health Awareness Month Introduction: Mental health in Tech Fostering mental ...

    1 person likes this.
  • Posted in: Member Open Forum

    This article is written to serve as a documentation of the implementation journey our members have been on and serves as ...

    2 people like this.
  • Posted in: Member Open Forum

    There is never a dull moment for digital natives. This ever-changing landscape of seismic market shifts, new ...

    2 people like this.
  • As technology has continued to advance, many classic marketing methods have started to go the way of the dodo. We’re here ...

    1 person likes this.
  • Happy New Year! It is 2022, and here we are - ringing in the new year and taking a few minutes to reflect on the old. At ...

    1 person likes this.