Thursday, November 24, 2005

Directionless.gov rides again

Tom had the nerve to post that directionlessgov.com produces the correct answer when given the word "think" (as in road safety - see below) to hunt for. The "less" site (which is not to say the other is the "more" site) shows both direct.gov's search output as well as google's (which it rebadges and rebrands very nicely). I hadn't planned to defend direct.gov here, but it will look like I am. The google output is 9 links to the thinkroadsafety site and one to the hedgehog site (likely winner of the award for most unusual .gov website ... www.hedgehogs.gov.uk. If you care to visit, you'll find you have to "hog in" - I'm sure the bad jokes don't stop there but I didn't dare go any further). Direct.gov demotes road safety to number 7 as I posted before, but it does at least give 10 different sites, just in case when you typed "think" you really did want to visit rethink or RU thinking about it or whatever. Google probably don't mind if you hook into their site for the odd search, maybe even a few hundred (although they'd probably be a bit upset if they looked and found that directionless.gov strips out the ads which drive their revenues). But, there were 589,039 searches on direct.gov in October 2005 (from the EDT report) and I can't imagine google being too happy about that amount of freeloading on their servers. Indeed, when we talked to google (in my old job) about using them directly for government search (we even registered www.search.gov.uk I think), they wouldn't allow a direct hook into them but, instead, wanted to sell a google black box for close to $1mm. Not the best spend of taxpayer money I think - when most users are going to try google first anyway. Interestingly, if you use google and don't restrict the search to just .gov sites (which is what directionless does), and you search for "think" - where does the road safety campaign site come out? 7th. The other point, reinforced by Jason's comment below, is that there are still just too many domain names. There are probably at least 5,000 maybe 5,500 which reduce down to perhaps 3,500 individual websites (allowing for, say, www.ir.gov.uk pointing to hmrc.gov.uk - maybe that's a plan, we can just rationalise government departments to reduce the domain count?). I think that's 3,250 too many and if I'm really nasty, it's probably 3,490 too many. I've often stood at conferences and said that the average person has 10 bookmarks for key sites. The odds of one of them being government are pretty low which is perhaps why direct.gov whilst doing well with 1,500,000 visitors isn't getting anything like the 20% of the population visiting that it probably should get. Not an easy set of problems to solve - but it needs more than strategy, it needs action.

13 comments:

  1. Hmmm. The direct.gov search results are in fact weighted towards keywords found in the title, with less emphasis given to keywords found in the body - a sensible strategy, I'd argue. (I think the body-copy gets something like 40% of the weighting of a page's title.)

    Meanwhile, I'm struggling to see the point of directionlessgov.com. It's comparing google results from *.gov.uk (as Alan says, a very inaccurate representation of the UK government) to those from a single site (d.gov). The adage that "google's results are right" is a scary one, particularly in such a subjective field.

    Also, I find it somewhat frustrating that all bar one of the top ten google results are from the same site. What if you were looking for something else? I'd rather have a broader set of results, personally.

    d.gov should put a recommended link in place for the "think" keyword" (akin the one you get when you search for "learn"). The fact that this doesn't exist, along with the fact that neither the Motoring section nor the Travel Safety page within d.gov has a link off to the "think" site, suggests that joined up government is hard - the motoring bodies in government were a key part of the initial d.gov launch.

    How about a pan-government ad server?

    ReplyDelete
  2. So much did direct.gov cost the taxpayer really?

    ReplyDelete
  3. Well, Directionlessgov uses the Google API, so the rebranding is completely above board (though there is a limit to the total number of searches per day). The first version, of course, just redirected to Google, adding "site:gov.uk" to the end of the query (with one minor exception), so wasn't subject to any limits on usage.

    ReplyDelete
  4. 'Nerve'? I'm not sure why you use that word, Alan. You know as well as I do that Directionlessgov was a hack to demonstrate how poor the directgov search was 12 months ago. What precisely is wrong with concerned citizens showing that they're not satisfied with public services in a constructive way?

    Also, some observations on your critique.

    First, Directionless uses a free API key provided by Google, which provides licenced search responses without adverts. So suggesting they wouldn't like this hack when it is both freely licensed and makes them look really good doesn't really follow. We could invite someone from Google to comment if you want.

    Second, the fact that Google gets the right link first for your search term is obviously what's of importance to someone searching. The observation that direct.gov produces loads of *diverse* but almost totally irrelevent links, wheras Google just repeats the correct one really sounds like clutching at straws.

    Last, the cost/benefit value. If they really were going to charge £1m (and I have heard contrary stories on this) I reckon it would still very much have been worth it, given that a) the yearly budget for the site was £4.4m in 2004 and b) your half million search users a month were (are?) getting substantially below par results, and many of them will give up before they get the services they were looking for.

    Tom Steinberg

    ReplyDelete
  5. Dear Alan,

    Directionlessgov records the link that people click on, and the search engine that gave them that link. Users of directionless pick google results two thirds of the time - Direct.gov.uk gets chosen one third of the time.

    Our adage that "google results are better" comes from the above numbers. According to our users, they are.

    While you may find it frustrating that all bar one of the top ten google results are from one site, that is probably because the one site happens to be the most useful for whatever you looked for. If you were looking for something different, you'd obviously get different results.

    ReplyDelete
  6. I knew I'd get it wrong. I thought by saying that I wasn't trying to defend directgov that I'd covered myself. Still, it got a good debate.

    Tom - I wasn't using "nerve" in the pejorative sense, more with a wink and a smile. I remember when you press released the launch of directionless and all the fun that was had. You may have heard conflicting stories on the $, I'll leave you to google the authoritative source. Not clutching at straws ... but obviously touched a nerve on your usually calm exterior.

    I know directionless uses the google API, I'd be interested in google's view on the ads though - after all, it's what they do to make $. I love the idea of using google to search government - but I also agree with Dan that it may not give the best results. Directgov searches only that site, google searches everything - they're doing 2 different things. But I've been through all of those arguments before and lost many of them, so I won't try again.

    I'm intrigued by the 2/3 1/3 stats from directionless that sam smith quotes ... what's the volume of searches carried out?

    ReplyDelete
  7. For information, I'm the person who built the directionlessgov.com comparison search.

    What direct.gov.uk chooses to search for is their decision. That direct.gov.uk suggests that it is the place that you should go means that the comparison is valid. While your post criticised DLG for showing 90% of it's results for one site, why do you suggest that direct.gov.uk showing only it's own results isn't a flaw?

    It's also worth pointing out that the "new improved search" on direct.gov.uk meant the direct.gov.uk clickthrough rate went up by around 2%.

    DLG has shown 2119 results. We should probably get round to telling someone about it again. Usage is pretty much completely flat (other than a spike this morning from this) - the site has not been promoted since January when the comparison stuff went up. I'm not sure how much Direct.gov.uk has had spent on it this year. Do you happen to know?

    ReplyDelete
  8. I'd argue that in DLG, your audience base is a tad skeptical already, otherwise they wouldn't be rummaging around on such an emotively named site.

    I'm impressed that direct.gov.uk gets 33% of the clicks. I don't have the numbers to support the case, but I think Google has put more investment into its search algorithm than direct.gov.uk has.

    ReplyDelete
  9. There's a key point here that everyone's missing... To me the pointed joke that directionlessgov is making is not that direct.gov should be licensing Google's search.

    It is that direct.gov should not exist at all - that in practice everybody types what they want to do into Google.

    With the budget saved by axeing direct.gov, instead optimse text and titles on government websites. i.e. Do some Search Engine Optimisation. Run user tests to find the terms that people search for when wanting to do things that government can help them with. Arrange that Google, Yahoo and MSN searches for those terms take them to the correct site.

    ReplyDelete
  10. I half agree with Francis' point - I fully agree that there is a key point there. I think directgov should absolutely exist, but that a government search should not. Without directgov we'll be left searching 3,500+ government websites, many of which are contradictory, all of which work in different ways, few of which are accessible and most of which are just plain not relevant to us. Directgov's job is to bring joined up government to us by working with departments to help them write things that make sense to the masses, without government double dutch and without the boundaries between every department getting in the way.

    Some great points here - Sam's point about dg being flawed for only looking at itself versus dlg looking everywhere, for instance. I'd argue though that the self-selecting (as Dan put it) dlg userbase of 2000+ is not a good enough sample yet - and would agree that perhaps some more PR is due, although I'd much rather see that PR put to use for e.g. theyworkforyou or publicwhip which I still find stunningly brilliant.

    ReplyDelete
  11. Forgetting the tedium of search discussion, and going back to the customer, does anyone think that the Hedgehog site is any good?

    It it what govt. funded website designers think talks to kids?

    How many children actually visit this site via Direct and would use it?

    Public money being thrown after bad.

    These people should stick to the back office stop pushing paper.

    ReplyDelete
  12. Charley, talking about hedgehogs.gov, says "Public money being thrown after bad."

    I guess that says one thing, that he things there's some money being spent well somewhere. Hedgehogs is a site that has baffled me since I first came across it. I first mentioned it in Jan 2003, but I think I'd seen it before (http://www.diverdiver.com/2003_01_02_diverdiver_archive.html)

    As to how much directgov cost (was that madge who asked?) or how much it spends on promotion, I have no idea. But it's a lot less than it cost to build 3,500 sites or promote the same number ;)

    ReplyDelete
  13. Charley, talking about hedgehogs.gov, says "Public money being thrown after bad."

    I guess that says one thing, that he things there's some money being spent well somewhere. Hedgehogs is a site that has baffled me since I first came across it. I first mentioned it in Jan 2003, but I think I'd seen it before (http://www.diverdiver.com/2003_01_02_diverdiver_archive.html)

    As to how much directgov cost (was that madge who asked?) or how much it spends on promotion, I have no idea. But it's a lot less than it cost to build 3,500 sites or promote the same number ;)

    ReplyDelete