Friday, 6 March 2009

Enterprise search revisited - updated

Although I am a bit slow catching up with the conversation (hey, I have been busy....) I wanted to respond to Pie who has made his first posting specifically around search, as he say's search is cropping up more and more often in his world of ECM work.

Pie's posting his here: "Traditional Enterprise Search meets Enterprise 2.0"

Pie's posting is largely in response to and reference's a recent post by Bex on "Why Google will never be good enough at enterprise search"

Now I have given you the links, I don't need to regurgitate their posts in full, when you go and read Bex, ensure you read the comment thread too.

So, enterprise search, its a tough nut to crack thats for sure. For a start, like 'knowledge management' the term 'enterprise search' can mean slightly different things to different people, but the sake of arguement lets define it as the ability to search across all the various content repositories and information systems within an organisations internal boundary (aka "inside the firewall").

Personally I have to say that like Bex, I am not a Google Search Appliance fan. However this is not particularily because of the hoary old "page ranking does not work on an intranet" arguement - well all know that, and if you believe what Google tells us Page Rank has been tweaked and joined by other algorithms designed specifically for intranet search in the GSA. No its the twaddle about 'plug and play' and a 'simple approach' that annoys me the most. I have one client who has been a poster child for Google in this respect, and yet when I interview their intranet users I still hear "such sucks"........

Enterprise search is absolutely not a simple problem with a simple solution. Alright, actually let me caveat that straight away, all organisations are different, and the context could possibly mean that in any particular organisation the problem might be simple, and a simple answer might indeed be good enough, but lets return to this later.

So, the fact that things are difficult to tweak with a GSA (although it gets better with each release) and the fact that organisations decide to plug in the simple appliance and cut the funding to the 'search team' (beause Google say they are not needed) has an overall negative effect.

But if we return to the concepts of 'enterprise search' per se it is what 'systems' academics call a messy problem. It is a space populated with interlated problems and issues, a myriad of dependencies and complexities and so we should consider that solving these issues may take time, a phased approach, a realistic approach too, and of course money and effort.

As Bex notes in his original posting and Pie refers to, the move to web 2.0 technologies in their enteprise 2.0 guise does nothing to ease the situation. Now as well as federating a query against a document and records management system, the intranet and your CRM you may want to query everyones 'personal profiles' (in your corporate Facebook analog), your Yammer micro-blogging system, hundreds of 'workspaces' (think eRoom, Quickr or SharePoint) that inlcude discussion forums, blogs, and yes and dont forget the wikis (oh my!).

Oh yeah, and RSS feeds, don't forget the RSS feeds.........

Now Bex suggests people don't really want to find the documents or the web pages, they just want to find someone to give them the answer. I think this is an oversimplification, and indeed it speaks to the differences between 'findability' of information and actually undertaking some 'research' - but lets just say we do consider 'social search' to be the panacea and we all put up highly detailed personal profiles detailing our expertise and what we are working on - how many times are do you want to be interupted by phone calls, IM's or email asking for your opinions, help etc before you decided to remove your profile ???

I am not criticising social search per se, I just think that because technology can enable it, it is not necessarily the right option to persue, depending on the organisation. Relying purely on social search would bring all those thorny old 'km' issues back to the fore; trust, who defines 'expert' within the context of the organisation, how do you get people to contribute content (and therefore expertise) etc etc and so I tend to fall back on my trustee methodology and my usual maxims, think about the human element and don't be seduced by the technology !

  • Who are your users - my colleague Martin White suggests using persona development techniques when undertaking requirements analysis for search
  • Your users needs and requirements - what are they looking for, what are the pain points
  • Content findability - whether its documents in an ECMS or comments on a blog, consider both metadata and full text indexing capabilities - but mostly consider your enterprise metadata policies, taxonomies and controlled vocabularies (because you do have them, right ?)
  • The strategic context - where does your enterprise search initiative fit into your other information management or knowledge management priorities and your enterprise architecture (because you definately have one of those...... don't you ?)
  • Your approach to the technology and usability - are you happy with a pragmatic approach which may see many tools in use, or are you looking for the 'one ring to rule them all' ?
So, perhaps even more so than other large enterprise wide initiatives based on the implementation of technology, the need for extensive business and user requirements analysis is a must, and upfront planning work will pay dividends later.

You can go for a single tool / single interface approach, single tool / multiple specialised interfaces approach or a mutliple toolset approach (the right tool for the right job?) - although if you build a specific search portal you can even have multiple tools / single interface ! However most importantly be prepared to invest in your 'search team' - you need humans to sys admin and manage your search technology, to review the searches being issued, to tweak the relevance ranking accordingly, to manually configure 'best bets' and to provide assistance and training.

I have a set of search related RSS feeds pulled into my Pageflakes ECM site here:

I would recommend looking at the AIIM Information Organisation and Access training program here:

And if your looking to evaluate specific search technology platforms, you should take a look at the new CMS Watch 'Search and Information Access Report 2009" because there are still an awful lot of products available in this space, despite what the marketing departments might spit out this is far from a two hourse race between Autonomy at the high and Google for 'simple' search !

Oh yes, dont forget to do some web 2.0 search and use search or 'enteprise search' as your keyword search term on :-)

And finally..... as they used to say on the UK television evening news, you could go here and buy a copy of a very good book on the subject, (I know its good, because I was one of the proof readers) Successful Enterprise Search Management

Updated: Found a really nice post over at the Search 2.0 blog - Enterprise search, redefining scope


Kas Thomas said...

Thank you for mentioning the Search and Information Access Report (of which I'm a coauthor). You make all good points. In addition to the blogs by Bex and Pie, I recommend people take a look at some of Lee Romero's blogs, as well as some of the blogs. I also tend to devote a lot of Tweets to search-related items: (

I would add two thoughts. First: Decide whether you need document-level security (role-appropriate search-results filtering) in your search solution, because not all products do this, or do it well, and it can be challenging to set up. You may not need this capability at all if 100 percent of your intended users are people browsing for merchandise on a retail site, say. But for corporate intranets, you probably will need support for document-level security.

Secondly, consider whether you might need federated-search capability (now or in the future). Just as organizations often end up with more than one content-repository technology, sometimes organizations end up with more than one search technology, and there can be a need to federate the technologies in some way. Again, there are search products that can handle this sort of thing readily, and others that can't.

Jed Cawthorne said...

Hey Kas

Thanks very much for your comments, much appreciated :-)

As someone who follows you on twitter and hits the CMS Watch blog feeds everyday I should really have mentioned them eh !

Good point ref 'federated' search, as I beleive this is where it starts getting complex very quickly, often stepping beyond the organisations experience and understanding.