Google Discover is a personalised and “queryless” search where cards appear rather than the standard search results. On the whole, it works pretty well and it can be a great source of erm, discovering new content sources.
Well, it used to be anyway. At the start of March – or thereabouts – something changed and it has been a downhill slide into oblivion irrelevance ever since.
The main issues centre around irrelevant cards, empty feeds and Google Discover not registering a user marking a topic as “not interested”. Publishers, especially small and medium-sized ones, have suffered terribly as a result.
Now, before this starts descending into the ramblings of bitterness let’s have full disclosure. Yes, we have been severely impacted by whatever is going on. The effect has decimated all the growth of the last 12 months. If you’d like to discuss this in more detail, you are welcome to contact us directly, but that is not what this post is about.
Earlier this week, Google made some interesting changes to its official Discover documentation. This might or might not relate to some of user-reported issues over the last few weeks – who knows.
On Monday, a Tweet from Neil Gaiman (yes, that one) appeared in my Twitter timeline by chance. Turn out Neil himself (couldn’t resist, sorry) has been having one of the same issues that we have seen and reported countless times recently.
Gaiman’s issue is largely around Google Discover not understanding that he is not interested in sport – having set that preference on previous occasions.
My fascination and desire to figure out how Discover works started long before the feeds went rogue though and have been trying to extract information about what influence the appearance of things for months – sometimes with some interesting results.
Despite extensive testing, though, determining what the data means is still a challenge, because it seems utterly random. But, as it turns out, there is a better way to do things.
Currently, this will only really offer insights into your own behaviours with Discover (but here is hoping there is a bit of goodwill and open-source camaraderie to try and learn more about these feeds), but it does offer a way to see what influenced the appearance of certain cards in your Discover Feed.
It details things like how many cards appeared, which topics they were on, how the user interacted with the cards and, somewhat creepily, details about when a card might have surfaced based on your location history.
The screenshot above shows a partial view raw data set (my own personal one) that can be downloaded for deeper insights into your Discover Feed’s behaviour.
The desire to find this information was borne mostly out of poking around to see if there are any further signs about the Discover feed coming to Chrome Desktop (spoiler alert: it almost certainly is).
This data set will be invaluable for analysing and tracing the “zero cards in feed” bug though. While it has been largely fixed, it still rears its head at least once a day.
How to access your Google Discover Data Feed history
This is done by downloading your Google data and, since they don’t mind using other people’s content to answer questions, we assume they won’t mind us doing a quick copy and paste job right here. Discover Data is found under “My Activity” and can be downloaded in either HTML or JSON format.
- Go to the Download your data page. Google products that have your data are automatically selected.
- If you don’t want to download data from a product, uncheck the box beside it.
- If you only want to download some of your data from a product, you may have the option to select the hamburger menu to refine it. Clear all selections and then, you can uncheck the box next to data you don’t want to include.
Google Discover is found under the My Activity section – which you can select by clicking the hamburger menu again (do this or you will download everything – well, unless you want to download everything, then you do you, boo).
Off to the rabbit hole we go.