What are analytics?
Web analytics provides statistics and basic tools to help you understand how people are using your online service (i.e. website or online application) and how the service is performing. More importantly, you can use this data to make improvements to the service.
It’s equally important to remember that data is part of a bigger picture. Analytics will not provide you with conclusive answers about your online service - it can help you understand how your service is performing and evaluate how to further improve and invest in an online service. This data can and should be complemented by user research and usability testing.
Why gather analytics?
Site analytics can help you understand how to improve your service. If you can ask specific, discrete and data-focused questions of the data you can gain important insights into your platform.
Some questions you might have about your site include:
I have a lot of outdated content and updating will be a big job - Where should I start?
I get constant email requests for information that is already included in the web site - Why are people having trouble finding that information?
The head of my department is pushing for a new feature on the site - but, I’m not sure it’s a good idea. How do I support this claim?
Related blog: Setting SMART Website Analytics Goals for Success
Understanding the Big Picture
Users and Audience
Google analytics is only one part of a larger puzzle. The best way to know if your service is useful and valuable is to ask the people you are trying to serve.
Don’t be intimidated by user research - you can start small. Identify a group of 5 people (site users, colleagues and/or strangers). Ask them basic questions such as “How do you typically get to the site? What is the most common reason you have for visiting the site?”
Be neutral and open to user feedback. User opinions may be opposite of yours (ex. Users like the blue button, you like the red button. Users want feature X, you want feature Y).
Product Purpose
Every online site has a different purpose and is targeted to specific audiences. Some sites (e.g. http://unocha.org ) are for advocacy and the purpose is to share the information with a wide number of people. Other sites are operational - they fill a specific need for a certain audience.
Different purposes means the way we measure and understand users will be different. Use your product goals to guide your analytics.
Type of Site | What you might measure | How you might measure |
|---|
Public advocacy (e.g. http://unocha.org ) | How many people are visiting the site? | Site visits (trends over time) and user engagement - CTA (click to action): subscribe, downloads |
Specific service - Humanitarian Data Exchange ReliefWeb Response | How many datasets are being downloaded? How many users are coming from countries with active humanitarian operations? | Engagement rate/clicks on links Site visitors (location) |
Humanitarian Context
It is important to consider what is happening in the larger humanitarian context in order to make sense of patterns you might see in the analytics data. Questions to consider include:
Is there a new emergency or rapid scale up?
Is there an advocacy campaign that may be encouraging more page views or new search terms?
Has there been an interview with a senior leader?
Has a major news organization linked to a report or page on the site?
Sampling in Google Analytics
When dealing with large datasets in analytics tools like Google Analytics, data sampling becomes necessary for performance reasons. However, it’s essential to recognize that sampled figures are not precise representations of the entire dataset. Instead, they provide a statistical estimate based on a subset of the data. These estimates are valuable for identifying trends, making comparisons, and understanding overall patterns, but they should not be reported as exact values. For specific details refer to the Google Analytics documentation.
Proposed Disclaimer: When sharing sampled data, consider including the following disclaimer: “The figures presented here are based on a sample of the data and may not reflect the complete dataset. Use them for trend analysis and relative comparisons, but exercise caution when interpreting specific values.”