Spring Clean

Question

We are concerned that across our hundreds of users a lot of data has been created that is not being actively used. Many of these files have a tiny storage cost implication but cumulatively it is very significant.

Asking users to review their content isn't always practical, so i wonder whether periodically we should download all data from ArcGIS Online and wipe the slate clean. Users could then add back the data they actually need.

How have other large organisations dealt with this?

jcarlson · Accepted Answer

A clean slate sounds nice, but the risk of breaking something is pretty high. Even if you add things back, everything will get a new ItemID, and any prior links or map configurations will not work.

We're not a big org, but here's how we approach it. You can use the built-in Administrative Reports, or else use Python, but every month or so, get a list of all your portal's content, along with how many views the item has had. Keep generating these reports, and after several months, you can combine reports into a single dataset.

At that point, you've got your items and their view count per month. You can calculate how much an item is being used month-over-month and get a derived percentage.

For items that are not being viewed (e.g., the growth percent is 0, no new views are coming in), you can more safely assume that these items are "stale". I would suggest establishing a formal policy in your organization that items which remain "stale" for more than, say, 6 consecutive months, will be archived and removed from the site.

This gives your users a clear metric and a policy to back up any decisions to remove content from the site. You can fine-tune the idea however you like.