NOTE: these are some thoughts I wrote a year ago. I publish them here “as is”.
Why would you think about having statistical to analyse the audience of your Digital Humanities website? There are several reasons:
- A demand from funding bodies (answering the boring but sometimes pertinent question: is this website used by someone?) In my experience, the funding bodies were asking indicators related to user’s loyalty, to the number of consulted documents (rather than pages), and are sometimes interested into advanced webometrics/altmetrics indicators;
- Because you need to know who is using your website / webservice / etc.
An indicator is usefull if you can have an influence on it, otherwise it does not indicate anything. Statistical indicators, in the case of the web, are based on the number of people who are going to your site and based on your ability to have an idea of who are those people.
This has two consequences:
- First, you must know your users. Which means, in the case of digital humanities, that you’d better know who are your potential users from the beginning of your project. Once it is launched, you need to use your Google statistics, logs, piwik statistics, etc. to try to understand if your supposed users are really your actual users;
- You must havec clicks, you must be well known, that implies a communication policy, of course, but also search engine optimization (SEO).
SEO is… painfull.
In my previous experience, a satisfying search engine optimization lasted a year, notably because we did not integrate SEO in the development process from the beginning. Furthermore, results of SEO are not immediatly obvious in statistics. SEO is a long term task.
SEO is also a painfull process, because a part of your editorial policy will be handed over to Google:
- the design of your URL (forget putting in the URL the DOIs or handles, not enough “human” for google) – so you need to have persistent URL and human readable URL;
- need for Canonical URLs;
- the use of titles (<h1>, h2, etc tags) within the page, etc
It might not seem that dramatic, but we are talking about Humanities, where the editorial process is of particular importance.
The most important point remains the links that are pointing to your website and the quality of those links. This is more a (very) long-term communication task than a SEO one.
The definition of the indicators is crucial for two reasons:
- If you set up indicators in response to a demand from your funding bodies, they will judge your work partly in regard to those indicators;
- Because indicators must also be usefull to you, to the way you are elaborating your website editorial policy.
There are plenty of types of indicators that you can implement. Choose them carefully: they must be meaningfull for your funding bodies and for you. The funding bodies will look at your capacity to increase the number of your users, to increase their loyalty, to increase the number of viewed pages. So you must choose an indicator that you can influence. To know if you can have this influence (ie if the way you are managing your website can change the results of your indicator), you must carefully dig into your statistical data.
Dangers of indicators
The problem is that the statistical needs of the funding bodies are not necessarily matching your needs. You might be more interested in a part of your audience – the most loyal users or researchers rather than students, etc. So, it is probably better to separate the funding bodies’ indicators and yours. But that’s more time, more energy, more funding.
The other danger is to change your website’s editorial policy to fit it into your indicators. Your indicators are supposed to be helpful. But, like SEO, they can transform an interesting poject into a meaningless website that is made to attract users whatever the cost is.