What Did the Digital You Really Mean?

Blue tits and intent

April is the cruellest month according to TS Eliot’s The Wasteland. It may also be the cruellest month in the digital world with big changes happening at Facebook and Google around assumptions of intent and behaviour.

The Facebook Why

Being British, when I read the news on April 1st that Facebook was going to share some of its algorithm selection on posts in our newsfeed with a ‘why am I seeing this option’; I wasn’t sure whether it was an April Fool. But it’s genuine, not fake news.

The change will allow users to understand what signals Facebook took to mean so that one post is more important and appears higher in your newsfeed than another or even at all. It also has the small opportunity to modify that assumption by blocking.

This is a similar feature to that already on Facebook adverts. It goes back to the fact that unlike a newspaper or magazine what and when you see something in Facebook is unique to you and selected based upon your previous digital behaviour. No two Facebook feeds are exactly the same. So if you watch a cat video; Facebook presents more as it presumes you like cat videos. Or if you always like or comment on friend A’s posts; it assumes you’re close friends and therefore want to see everything she posts. Too bad that friend B posts infrequently but always interestingly and is never seen.

Google Progress

Since the autumn of 2018 Google has been moving Adwords accounts from being based on key words to being based upon intent. From this month all Adwords will use assumed customer intent as the basis of presenting the ad. The process behind this has been happening for some time.

Consider what you now see when you use Google or Chrome search. The top entries might be adverts; then a box of suggested content followed by ‘people also ask’ with various suggested questions. Search has always assumed questions are being asked from the words that are typed. From these the algorithm deduces what might be the most relevant content on the web to answer your question. The ‘people also ask’ allows you to reframe the question if the content is wrong or to ask it in the crowd way. This enables Google to learn better the variety of intentions from actual use of words.

Google uses machine learning in order to understand the searcher better to thus increase the probability that the suggested content is more likely to provide the answer. To prove this log in to your business and home computers at the same time, with your business and domestic accounts and then do the same search. The results will not be the same. The order may be different or different entries. This is in part because in different life modes, your interests are different and your intent may also be.

The Tit, Booby, Bird Danger!

Blue footed booby avoids putting foot in it

Assume my intent is for images of colourful birds as per the image. If I type only the first two words into Google, the results are somewhat different. If I add ‘bird’ then I’m on safer ground as I’ve created a longer tailed search. The more I research birds in general and related subjects; the more likely that Google will take my intent to mean only feathered birds.

What Google can now do is to use its 7 core properties to optimise a rich user intent profile to minimise the risk of inappropriate content at the wrong time. The reason it needs to do this is because increasingly search is being done by voice. Here tone and words create intent rather than just words alone.

This massively increases the algorithm because we all talk and stress words differently. An Englishman and an Australian might write exactly the same phrase; but the Englishman won’t raise his voice at the end of a statement in the same way as some Australians which turns the statement into a question for those listening. This changes the understanding of the ‘bird’ in my search from ‘I intend to see birds’ to ‘are they birds?’.

One More Step

Both the Google and Facebook changes are further steps along the road to provide personalised relevant content to users, as those users’ needs and sophistication changes. They mark significant adjustments to focus on behaviour beyond what we think we simply write or like.

It also demonstrates the digital shift from the simple awareness of the top of the classic sales funnel to the lower down intent. Intent is an infinitely more difficult problem to solve consistently. As we know only too frequently in the real world in this cruel April.