I am currently looking for new career opportunities on LinkedIn, and it has inspired me to write a quick blog post about structured data and its importance under the hood to keeping the wheels of digital systems turning.
IMHO LinkedIn is very impressive. It has grown within its role as a business networking site, embracing some of the better features of its social media contemporaries without losing sight of its unashamed role at the centre of – now global – business networking. It’s unashamedly commercial, which is important because social media platforms face an uncertain fate when the business model relies on VC injections and hype rather than a sound business model. The fact is, LinkedIn ads, sales packages and recruitment plans are costly but effective.
Savvy Satya and the Microsoft team decided to buy LinkedIn because it has a wealth of data, and in its core B2B market. Data is the new oil… blah blah blah. But anyone who has built a system from a modest piece of PR software to an enterprise database will quickly learn that designing and policing the database structure is essential. If you can’t categorise things with a standard set of criteria and use a consistent format, then any attempt to recall that data within an application or report on it will fail. Take this blog, for example. It’s currently using pretty standard out-the-box WordPress (I’m writing words not code ATM) so this blog is
- Categorised as Digital (meaning web design and soft tech). That makes it possible for me to list this as a post under the category Digital and create an archive of content that might be valuable to an amateur web designer or digital team. It also makes it ease to list related posts alongside, because they are also defined as Category = “Digital”
- Tagged with keywords data and database which means I can also create a whizzy Wordcloud (yes I know they went out of fashion about 7 years ago). That Wordcloud could link to a archive of posts related to the keyword, in this case the purpose might be to create a content silo for SEO purposes.
- Authored by me. Admittedly, it would be odd for anyone else to author a post on robinwolstenholme.com – I shy away from it myself as an Englishman – but it’s theoretically possible for this post to be authored by, let’s say my brother the awesome photographer, PHP developer and web extraordinaire Miles Wolstenholme. In that case, he could have his own archive page and an author profile.
This is all operational – ie. using data within the application, but a search engine illustrates how it would translate to reporting. With a little work, and possibly a good open source plugin, I can use that data to enable visitors to filter their search results. If I were LinkedIn, the my blog post example could be replaced with a job posting. When a recruiter posts a job, they are required to enter certain fields in much the same way as my WordPress post – a descriptive (not much room for catching headlines) title, a salary (can be blank), a location, seniority level, description – and author should they wish to be inundated with connection requests. I have noticed that location is not terribly good data hence an advanced search for “PR” in “United Kingdom” under “Acme Rockets” for example might miss out your dream job. That’s bad for LinkedIn, bad for the recruiter (I have space rocket PR creds, buddy) and bad for the candidate (who doesn’t love launching rockets?).
Today it’s entirely feasible that fairly straightforward fuzzy logic can help match
Object-person (Robin Wolstenholme) to
Object-post-job in LinkedIn’s vast data warehouse rather than relying on a simple query by location or skills. Indeed, one could argue that the more data I share and the most content I feed the LinkedIn data warehouse, the more likely I am to be found, because machine learning automates the discovery of additional signals from the unstructured “big data” that will improve matching in a similar manner to an Amazon product recommendation engine or Google’s much-maligned Ads. However, if I want to improve my chances I will reverse engineer my LinkedIn profile so that my hard, irrefutable structured data such as Skills make me a better match. I’m not talking about gaming the system, it’s about talking the same language as the algorithms that will match my profile so that the computer says yes and I make the recruiter’s shortlist.
I’m not grisly, but I’m a big longer in the tooth than some. How might this effect younger potential recruits with just a few years under their belts? Will they self-homogenise their profiles get their next role, when they need to come out their shells? Yes of course – describe yourself in structured data so you can get on the list then look for ways to stand out once you are there! Social media should have taught young ‘uns in my industry that.
Consider Twitter hashtags (yes, to be fair LinkedIn is also re-evaluating the value vs spam factor of hashtags in UX in 2019). In the vast unmanaged, unstructured streaming data lake of the Twittersphere the hashtag provides a means to categorise posts. It’s exactly the same as a WordPress tag in the way it works – add one or more keywords to describe one piece of content. If you homogenise your use of hashtags you can tap into a conversation or gain visibility for your message. The difference is it’s open and democratic, so anyone can rain on your parade, or contribute…
From now on I expect “Stakeholder relations” will need to become a core skill alongside media relations. Hopefully not Case Sensitive – but let me check the autocomplete to see what Keywords LinkedIn gently nudges me to use (thanks for thinking through the UX guys). There are clearly advantages to the end user of any IT system in understanding and adapting the data they provide in order to gain entry to a larger community or ecosystem – but where does it present roadblocks (AR), deliver diminishing returns (SEO?), or threaten to homogenise the very thing that makes a person or commercial entity unique?
Listen to this spot (experimental)