Of course I’ve given this article a deliberately clickbaity title because every fourth recommended post on LinkedIn is talking about leveraging the power of AI to do something, and since in the LocalGovDigital world we’re barely out of the Web 2.0 era of discourse I thought I’d throw in a crowdsourcing reference too.
Context
A few weeks ago I happened to be on our website on a Saturday wanting to find out a certain piece of information, so the first thing I did was a search from the website home page. Nothing. I like to think I’m good at search queries but I’m not infallible, so I tried a different query. Still nothing. Hmm. ‘Online payments’ is my usual go-to test query, and that produced nothing as well. Further checks using different browsers on my phone and on my laptop proved that the site search was down. Because I’ve got Teams on my phone I made a remark about it, more for the historical record than with any expectation that it would be picked up by anybody over the weekend, on the internal channel for discussing website issues and outages. My counterpart in the UX and Content Design Team spotted it and between us we agreed we’d keep a watching eye on the situation so that we could give knowledgable updates to relevant people on Monday morning. By Sunday the search was still down and my colleague remembered a mitigation from a few years ago when search was down and instigated that.
Search on a website being down is generally something of a big deal - especially on a website which doesn’t incorporate topic-driven hierarchical navigation - at the best of times, but given it was the weekend of Storm Darragh myself and my colleague made an assumption there would be a lot of people trying to search for information about reporting fallen trees and other issues relating to storm damage, so we agreed to alert relevant senior managers and have the issue raised as a major incident on Monday morning. And sure enough, a check of Google Analytics revealed that a large number of attempted-but-failed searches during the weekend were related to fallen trees and flooding on the roads.
As I’m writing this article, the date is 28 December, the period now known as Twixtmas, when folks get confused as to what day of the week it is. And more importantly, the period when householders up and down the land are asking ‘when is bin day?’
I’m going to make something of a guess that up and down the land many people will be going to their council websites to find out this information.
Of course in the modern world the experience of long-serving website editors and managers about what people might be interested in needs to be backed up with solid data, so in that spirit I give you the Google Analytics results on our Search Engine Results Page for the period 23 December to 28 December:
The top query (after users just clicking return in the search box) is Bin collection with a space, the fourth query is Bin collection without a space, the 11th query is Bins, the 19th query is Rubbish collection, and so on. The other queries there are ones which might typically be being searched for throughout the year.
Surprisingly, a quick glance around a few key council sites and I found relatively few of them calling attention on the website home page to the fact of bin day potentially changing for people; many sites had an anonymous link amongst the other links on the page labelled something like 'seasonal changes to council services' in our best council-speak, at least one mentioned it in the form of a link to a press release in the news section which had a quote from the council leader wishing residents a happy Christmas and giving some facts about how many bin collections from how many households are done every year before they got to the nitty gritty of when your rubbish will be collected these two weeks.
The best I found was Edinburgh City Council with a big banner of Father Christmas pushing a wheelie bin with a link to the page with the dates.
The takeaway
I’ve long advocated that council website home pages - still statistically one of the most accessed pages on the site even if most website visitors are entering the site directly from an external search result for the service they’re interested in - should not be static pages which barely change from year to year, never mind month to month, rather they should dynamically change in a manner which is relevant to the specific user who is accessing the site right now.
My sample home page for this site includes near the top a personalised area, where the user would be given configurable key information they want to see most often, links to favourite pages, and updates on recent requests, all without having to log in to a separate account page.
But I contend in this article that as people in charge of the user experience of council websites there are things we ought to be able to predict which are likely to be of relevance to whole swathes of our users on a week to week, even day to day basis. Storm Darragh was a predictable event, and its consequences for large numbers of citizens who use council services was therefore predictable, so therefore council website home pages should have included on them direct links to the key services users might have needed access to during the weekend, regardless of the fact of search having been broken. Christmas and New Year bin collection changes are similarly predictable events. On the latter, since those changes literally affect almost every household in the borough, why make people click through to a separate page when the key information could actually be conveyed in the space of Edinburgh’s funky banner graphic on the home page itself?
Of course, we can’t predict all of our users’ changing user needs, especially not on a day to day basis.
But guess who can? Our users themselves! Which brings me to the clickbaitey title of this article.
When a user enters the site from an external search query, the query they used is logged. Similarly, when they use the site search, their query is logged. As website runners we can easily see on a day to day basis what our users are searching for, that they may be searching for because they can’t immediately see a link to it on whatever page they’ve landed on the site on. We know the website home page is one of the most accessed pages on the site, and we can reasonably deduce that the reason it’s one of the most accessed pages is either because that’s the page the user entered on, or that’s the page they jumped to when their entry page failed to meet their user need.
We will obviously want to continue to design our home pages in the expectation that links to certain pages and services will be persistent over the course of the year; we know that whatever time of year one of the largest page accesses by our residents will be for the School Term Dates page, so we know we’ll always be on a good bet by including a link to it in the absence of developing a clever personalised home page for logged in users. Similarly, we know that even if few - fortunately - people need to access content pertaining to vulnerable people at risk, that content and any accompanying forms need to be prominently linked to from the home page because it’s a critically important life and death service we’re responsible for which a user in need should not have to poke around looking for it.
But as well as our persistent and personalised home pages, in order to make them useful to our users a user-centred designed home page will also have a region on it which updates on a regular basis according to what we know users are searching for this week, in a world where user needs change on a seasonal basis. I would suggest splitting such a region into two - a block which is automatically self-updating in real time showing the top, say five, search queries over the last week as a link to the results page for those queries, and a block which is manually curated links to the top, again say five, service main pages during a longer period of time, say, the last month, that a clever analysis of our search data shows us people are searching for as distinct from the raw uncontextualised pageview analytics. On the manual curation section care would obviously need to be taken not to create a Survivorship Bias / Confirmation Bias loop where you give prominence to one thing which makes people more likely to search for something else, so you then give prominence to that which makes people more likely to search for something else. Etc. But that's the great thing about a website - you've got a canvas you can experiment with that you can change on the fly according to how you see people using it, it's not a CD ROM which is glass mastered and duplicated 20,000 times where getting something wrong can be very expensive.
I've mocked something up on the BigTown Council home page:
If you're somebody who maintains any of the main Content Management Systems used in local government or independently of a vendor makes plugins and widgets for them, then do feel free to take this idea of a trending searches widget and make it for your CMS, and let me know if you do so I can update this post to big you up for it. I'll see if I can make time to make something for the Jadu platform and the LocalGovDrupal platform but my personal development time is limited, so feel free to beat me to it if it interests you.
And in the meantime, Happy Twixtmas, 'to all those who celebrate it'!