Conduct an analytics health check. If you use Google Analytics, make a copy of the Speero GA health check template and fill it out.
Validate the implementation, ensure that the data you are collecting is useful, can be trusted, and identify any issues that might affect results. There could be several problems when the configuration is broken which will lead to false results.
Check key metrics for specific dates in your analytics tool against data in other tools to validate tracking and events.
- Look at traffic volume for a defined date, for defined pages, in the analytics tool. Then, look up the volume for those same pages on other tools, like hotjar, BigQuery, fullstory. See if the numbers line up.
- Follow the same process for other key metrics, like conversion rates, button clicks, pages visited, and funnel progression.
- For conversions, compare the numbers in the analytics tool against your back-end data. Does the amount of leads or transactions actually reported in the analytics tool reflect reality?
- Sessions: Similar to Users, but a single user can have multiple sessions. Generally, it’s better to use session metrics rather than user metrics. This tells us how many sessions were observed that went to a particular page, and what metrics are associated with them.
- Bounce: User enters and exits the site on the same page, no interaction with the site, exit within 30 seconds. Keep in mind that bounce rates can be misleading. Look at how the tool is defining bounce before extrapolating insights, and make sure no interacting events are skewing it. Common culprits are live chat widgets.
- Exit page: Page from which the user exits the site. These can help you understand why people left. Context is important. It’s fine for exit rates to be high on an order complete page, because the point of the visit is finished. Similarly with blog posts, especially if the post doesn’t include a CTA. However, shopping cart and category page exit rates are typically areas for concern.
- Landing page: Page on which a user begins the journey on the site. This can help you understand how people are getting to the site. This can, and should, be segmented by source.
- Pages per session: How many pages a user views in each session. This can be helpful for understanding how engaged users are with the site. A user can go through multiple pages on a single session and that can be good – they’re engaged. OR… they can be looking at multiple pages because they haven’t found what they were looking for. You need to look at the larger context to figure out which. For example, look at the exit pages; engaged users will leave further down the marketing funnel or at the end, confused users will leave at random pages.
- Goal completion: Look at how you track and measure goal progress, and check that the analytics tool is matching your expectations.
- Search queries: If you have this enabled, it tells you what users are searching for on the site. Very helpful to understand intent if your site has a high internal search volume.
- Scroll depth metrics: The maximum amount a user scrolled on a page. This shows the average level of intent and interest in the content on a page. For example, you can give more context to a video on the home page with a low video start metric by looking at its location and the page scroll depth metrics. If it’s buried at the bottom of the page with low scroll depth, it’s possibly just not being seen; if it’s above the fold and getting low click engagement then it has high visibility, just low audience interest.
- Page value: A calculated dollar value for a page or set of pages. Typically, a higher page value means it is an important page. The critical thing about page value is making sure the metric is set up properly! It’s arguably the metric most often set up incorrectly. When it’s set up correctly, though, you can use it to prioritize optimizations on high-value pages to boost ROI.
- Desktop or mobile.
- New or repeat visitors.
- Specific categories or personas, like people shopping for backpacks or goggles, or small business or enterprise.
- Logged in users or not logged in users.
- Converters or non-converters.
- Source: paid advertising, organic, or direct traffic. It can also be worth comparing different campaigns on a specific page.
- Landing page
- Events: for example, used filter or didn’t use filter.
- Channels: check that these are tagged in a consistent way. For example, some organizations interchange source and medium, or different people use the terms differently, leading to confused data.
Source: where the traffic came from. For example, Google, Bing, Facebook, third party referrer.
While you’re here, you can flag things like Facebook campaigns aren’t driving as much revenue as Google campaigns. Include context about the type of campaign. For example, a brand campaign might understandably draw few conversions – the important metrics might be time on site or pages/session.
Additional segments (desktop vs. mobile, etc.) can be helpful in insight generation here
- Medium: category of the source. For example, paid search, or social. Further segment source by looking at the medium. Brand campaigns tend to be categorized within the medium more easily, too.
- Landing Pages view: Identify top entry pages and segment to understand who is entering where, and at what levels. Then look at what do they do after that.
- Exit Pages view: Identify top exit pages. Exclude pages that you don’t care about – for example, a thank-you page is usually an understandable and acceptable exit point. Segment to understand who is exiting, where, and after how long.
- All Pages view: Identify top trafficked pages, based on sessions and pageviews. Segment to understand who is going where. Look at bounces on pages. Segment to understand who is bouncing the most. For example, it’s common for paid search to have higher bounce rates than direct and organic traffic. Look at time on page. Segment to understand who is spending more or less time on specific pages – high time on site pages could be higher value pages.
- Site Speed view: flag pages that are slower than average to load. Segment by page, browser, and country. Test problematic pages on PageSpeed Insights to find tactical items to address. If you can’t see what’s going on, flag some slow loading pages with your development team to see if they can help figure out why those pages are loading slowly.
- Site Search view: validate that Site Search is installed correctly and tracking queries. Look at top queries, and conversion rates based on the top queries.
- Conversion goals: Focus on tracking which pages drive higher conversion goals. For B2B companies, make sure you understand what a conversion is. Is it a lead generated? Lead qualified? Make sure you understand what is being tracked, and how.
- Abandoned cart rate: look at the exit rate on the cart page, and corroborate that metric with the abandoned cart rates.
- Revenue per user or visitor: understand what revenue is, and how it’s tracked and calculated, especially for B2B businesses. There may be a backend system unrelated to analytics which is the master revenue data, and the revenue data may not be collected retroactively in google analytics. For ecommerce, make sure the revenue accounts for things like returns and upsells.
Identify pages with:
- Top traffic. See if there are any metrics which stand out, like time on site, add to cart rate, or exit rate. Pages with high traffic and bad engagement have the biggest ROI for testing and optimization. Segment to find the biggest source of traffic to those pages. Use previous and next page paths to understand how users came to this page, and where they went next.
- Lowest engagement, like time on site and high bounce rate. Segment to find the biggest offenders dragging down the average.
- Highest page value, hence conversion potential. Segment to identify which segments draw the highest value on those pages.
- Highest exit rate. Segment on traffic source and device.
- Top custom events. Segment on audiences. For ecommerce, see which products get added to cart the most, and understand what % of those convert. For example, a lot of people might add certain products to their cart but not go through with the purchase, which could provide interesting insights.
Look at high-risk accounts that are most likely to churn soon. Identify crucial actions or features that contribute to user retention, and analyze users who used it 0 or 1 times in the last 30 days.
Analyze new compared to returning users.
Pull interesting data from your analytics tool that you can use to highlight key issues to stakeholders. Add it to slides, and include commentary.
The key here is not to point out the obvious, but to highlight WHY this data is important or interesting and what you’ll do with it.
For example, if you have a landing page with a very high bounce rate, you might have investigated and found that Facebook traffic in particular is performing badly. Then reviewed Facebook ads to discover that they weren’t consistent with your site design and messaging, so visitors were immediately turned off. You could add the bounce rate data, the segmented bounce rate data, and samples of the website and Facebook ads to show how they were misaligned.