A couple of days ago, my Twitter timeline included several instances of a photograph of what appeared to be a small stack of floor tiles on which a cat had been sick. On closer inspection, it was a promoted tweet from an American food outlet which was offering me its product for free if I turned up and claimed it in person.
The nearest outlet is some 3,500 miles away from where I live and, leaving aside the point that I would actually rather eat carpet tiles and cat sick, the chances of my being remotely interested in the promoted Tweet was zero on geographic grounds alone. Why then did Twitter push multiple instances of this advertising tweet into my timeline, presumably charging the food-maker for doing so?
If Twitter’s algorithms had missed their target in pushing this advertisement at me, it is not the only target Twitter has missed. In reply to my tweet on the subject, someone sent me a graph showing Twitter’s share price over the last few months – you would get a steadier descent if you fell into the Grand Canyon.
When we come to look back on Twitter (and I’m afraid that we will soon be looking back on Twitter) its failure will turn almost entirely on its inability to understand and to target its core users. Under pressure from investors who seem to see value only in mere user numbers, it has been thrashing around making changes which seem designed to alienate its existing user-base without making it any more attractive to new users.
Its so-called “algorithmic timeline” (for now, at least, voluntary) seemed designed to entice users too stupid to curate their own timelines – perhaps the advertisers actually want stupid people, but not at the price of losing the rest.
Replacing the star with a heart seemed to follow from an assumption that if Facebook has a “Like” button then that must be why Facebook is successful. An entire vocabulary had been built up around Twitter’s stars, not all of which carried the “spread the love” implication of a heart. It is hard to see that dumbing down the vocabulary in this way has increased users nor, indeed, has Twitter made that claim.
The new “Moments” tab with its anodyne collection of someone else’s tripe, can at least be ignored for now, although it gives us a premonition of Twitter’s fate if an algorithmic timeline becomes the unavoidable norm.
All this is insignificant compared with the failure of Twitter’s developers to produce algorithms which accurately identify the sort of promoted tweets which users might actually be interested in – if it cannot do that, what hope has it with anything else?
Consider the carpet tiles and cat puke advertisement with which I opened. It does not need anything very advanced to deduce that I live in Oxford, England – it is a matter of record in my profile. Whether I disclose it or not publicly, my present location is usually known to Twitter. No great sophistication is needed to ensure that promoted tweets at least qualify to be relevant (or, rather, are not disqualified from being relevant) by geography.
I get this in examples less extreme than the foodstuffs shop across the Atlantic. One of my Twitter correspondents lives near Diss in Suffolk, 130 miles away. Presumably on the strength of that connection, I once received promoted tweets about the services of a Suffolk estate agent – useless to me, and useless to the estate agent, particularly if it was paying for the privilege of having its details shoved at me.
For several weeks now I have been getting a promoted tweet about a US document shredding service. It reads “Hey ATL Recall launched residential shredding services today. Got any old files to destroy?”. There is quite a lot wrong with this beyond the fact that I automatically ignore anything which begins “Hey” just as I would ignore you if you accosted me with “Hey” in the street. There is something endearingly dim about repeating daily an advertisement which includes their launch “today” as a key feature – they should call themselves “Groundhog Shredders” if every day is a new launch. But the main thing is that the possibility of my calling up a US company to pop across the Atlantic to shred my papers is zero. Why does Twitter do this?
Twitter’s use of my data could and should be much more granular and clever. I have created nearly 43,000 tweets and you would think that their content and their links and attachments, together with the sort of people I correspond with, could easily be mined for a pretty comprehensive picture of what interests me. That could be used to target me fairly precisely with promoted tweets which might, just might, be of interest.
I tweet a bit about food, for example, either about decent restaurants or about delicious things my wife cooks; how about advertisements for good restaurants or food shops near where I live or (since Twitter knows where I am all the time) close to my present location wherever that may be, rather than for junk food 3,500 miles away? That, you would think, would attract advertisers willing to pay good money to have their information pushed under my nose.
I am not in fact complaining here about the quality of the advertisements but gaping in bemusement at Twitter’s failure to use them properly. I accept advertisements as the price for the use of this marvellous free tool, and I particularly like the ones which recur often because they are instantly recognised and thus ignored. I would rather pay a fee to be rid of them but Twitter is not the only one to have shied away from that model, presumably because they reckon they can make more from advertisers by throwing junk at me even if it is usually irrelevant junk.
Ultimately, this is a fraud on advertisers. I occasionally reply to promoted tweets which have no relevance to me, for geographic or any other reason, to suggest to the advertiser that Twitter is taking them for a ride. Sooner or later, advertisers will want something better and Twitter, tied up as it is with frigging around at the edges with hearts and Moments, will not be able to deliver.