--- title: "Analytics and User Feedback" date: 2019-06-05 weight: 7 description: > Add Google Analytics tracking to your site, use the "was this page helpful?" widget data, disable the widget on a single page or all pages, and change the response text. --- ## Adding Analytics The Docsy theme contains built-in support for [Google Analytics](https://analytics.google.com/analytics/web/) via Hugo's [internal template](https://gohugo.io/templates/internal/#google-analytics), which is included in the theme. Once you set Analytics up as described below, usage information for your site (such as page views) is sent to your Google Analytics account. ### Setup 1. Ensure you have [set up a Google Analytics property](https://support.google.com/analytics/answer/1042508) for your site: this gives you an Analytics ID to add to your config, which Docsy in turn adds to all your site's pages. 1. Open `config.toml`. 1. Enable Google Analytics by setting the Tracking ID property to your site's Analytics ID. [services.googleAnalytics] id = "UA-00000000-0" 1. Save and close `config.toml`. 1. Ensure that your site is built with `HUGO_ENV="production"`, as Docsy only adds Analytics tracking to production-ready sites. You can specify this variable as a command line flag to Hugo: ``` $ env HUGO_ENV="production" hugo ``` Alternatively, if you're using Netlify, you can specify it as a Netlify [deployment setting](https://www.netlify.com/docs/continuous-deployment/#build-environment-variables) in `netlify.toml` or the Netlify UI, along with the Hugo version. ## User Feedback By default Docsy puts a "was this page helpful?" feedback widget at the bottom of every documentation page, as shown in Figure 1.
The user is presented with the text 'Was this page helpful?' followed
            by 'Yes' and 'No' buttons.
Figure 1. The feedback widget, outlined in red
After clicking **Yes** the user should see a response like Figure 2. You can configure the response text in `config.toml`.
After clicking 'Yes' the widget responds with 'Glad to hear it!
            Please tell us how we can improve.' and the second sentence is a link which,
            when clicked, opens GitHub and lets the user create an issue on the
            documentation repository.
Figure 2. An example Yes response
### How is this data useful? When you have a lot of documentation, and not enough time to update it all, you can use the "was this page helpful?" feedback data to help you decide which pages to prioritize. In general, start with the pages with a lot of pageviews and low ratings. "Low ratings" in this context means the pages where users are clicking **No** --- the page wasn't helpful --- more often than **Yes** --- the page was helpful. You can also study your highly-rated pages to develop hypotheses around why your users find them helpful. In general, you can develop more certainty around what patterns your users find helpful or unhelpful if you introduce isolated changes in your documentation whenever possible. For example, suppose that you find a tutorial that no longer matches the product. You update the instructions, check back in a month, and the score has improved. You now have a correlation between up-to-date instructions and higher ratings. Or, suppose you study your highly-rated pages and discover that they all start with code samples. You find 10 other pages with their code samples at the bottom, move the samples to the top, and discover that each page's score has improved. Since this was the only change you introduced on each page, it's more reasonable to believe that your users find code samples at the top of pages helpful. The scientific method, applied to technical writing, in other words! ### Setup 1. Open `config.toml`. 1. Ensure that Google Analytics is enabled, as described [above](#setup). 1. Set the response text that users see after clicking **Yes** or **No**. [params.ui.feedback] enable = true yes = 'Glad to hear it! Please tell us how we can improve.' no = 'Sorry to hear that. Please tell us how we can improve.' 1. Save and close `config.toml`. ### Access the feedback data This section assumes basic familiarity with Google Analytics. For example, you should know how to check pageviews over a certain time range and navigate between accounts if you have access to multiple documentation sites. 1. Open Google Analytics. 1. Open **Behavior** > **Events** > **Overview**. 1. In the **Event Category** table click the **Helpful** row. Click **view full report** if you don't see the **Helpful** row. 1. Click **Event Label**. You now have a page-by-page breakdown of ratings. Here's what the 4 columns represent: * **Total Events** is the total number of times that users clicked *either* **Yes** or **No**. * **Unique Events** provides a rough indication of how frequenly users are rating your pages per session. For example, suppose your **Total Events** is 5000, and **Unique Events** is 2500. This means that you have 2500 users who are rating 2 pages per session. * **Event Value** isn't that useful. * **Avg. Value** is the aggregated rating for that page. The value is always between 0 and 1. When users click **No** a value of 0 is sent to Google Analytics. When users click **Yes** a value of 1 is sent. You can think of it as a percentage. If a page has an **Avg. Value** of 0.67, it means that 67% of users clicked **Yes** and 33% clicked **No**. [events]: https://developers.google.com/analytics/devguides/collection/analyticsjs/events [PR]: https://github.com/google/docsy/pull/1/files The underlying Google Analytics infrastructure that stores the "was this page helpful?" data is called [Events][events]. See [docsy pull request #1][PR] to see exactly what happens when a user clicks **Yes** or **No**. It's just a `click` event listener that fires the Google Analytics JavaScript function for logging an Event, disables the **Yes** and **No** buttons, and shows the response text. ### Disable feedback on a single page Add `hide_feedback: true` to the page's front matter. ### Disable feedback on all pages Set `params.ui.feedback.enable` to `false` in `config.toml`: [params.ui.feedback] enable = false