4 IT self-service benefits made possible by a well-designed search architecture

Donny is an account manager in a large bank. About two months ago he received a newsletter from his company’s IT. It mentioned a new service offered to employees.—Cloud Drive. Donny is an enthusiastic Dropbox fan, ever disappointed from not being able to use Dropbox at work (“Our IT has security concerns” he says…). As such, this new service caught his eye. It promised to enable sharing multiple file types across people, PC’s and mobile devices. “Finding the Google-Play download link for the Cloud-Drive mobile app on the same news post was truly a hopeful moment for me” Donny says.


And so I clicked it”. Donny’s short celebration ended quite quickly as he discovered that he cannot download the mobile application to his (quite new) Android smartphone. “As I really wanted to try this service I decided to pursue answers”. “My first stop was the company portal”. Searching for “Cloud-Drive” in his company’s portal yielded one result. A News item from IT which “shockingly resembled” the item in the newsletter he had received just the other day. It had the same promise with the same dysfunctional link to Google-play.


Still not willing to give up Donny went into his organizational Facebook-like social site. Could it be that others have come across the same issue? He ran yet another search. Once again only one result. This time it did seem like a valid lead. Donny found a single user post about this exact issue. It mentioned that in order to get started with Cloud-Drive one needs to “contact ITand register as a beta tester”. Donny: “Offering a link to the subscription site could have been very helpful but at least I made some headway”.


Donny went on with his investigation. This time he logged a self-service IT support request asking to register as a Cloud-Drive beta user. “This was almost my breaking point” he said. Donny’s request was received by Dave, an IT support agent. It took Dave a few hours to talk with another IT support agent which in turn talked with the Cloud-Drive service owner just to learn that only few hours after the service was launched, it failed due to a huge interest. “The service owner further said that they decided to slow the stream of new users by requiring to register. They finally registered me and I am happy ever since”. At this point Donny sighs and adds: “I could not help but wonder… Are other users in my company happy? Is my IT happy? Consider that multiple employees, my friends, are receiving the same newsletter, try but cannot download the app, open multiple tickets to be handled by multiple support engineers which in turn all go through the process of looking for a service owner which in turn needs to manually register each user to the new service”. Ouch!


#1 -- You don’t need to know what you are looking for

By now we are all trained investigators. We come to expect that on the top of every decent application there should be a magnifying-glass icon and a search bar from which our investigation starts. Donny’s first search was “Cloud Drive”. The first lead in his investigation. What would the following “leads” look like?

  1. A post someone added in a forum?
  2. A news item published by IT?
  3. A product manual?
  4. A step by step getting-started guide or
  5. A catalog that offers service registration?

I vote for “all of the above”. An organization providing multiple IT services to its employees can no longer afford to separate these critical self-service information sources. One should not be aware of what they look for in order to find it



#2 -- You don’t need to know who you are looking for

The owner of the Cloud-Drive service, the guy that added the post about Cloud-Drive beta program or the one that already uses the service and formed an opinion about it–these are all potential leads for Donny’s investigation. For an ITSM system designer, it might be helpful to look at people as yet another information source. Building user-profiles out of the input these users provide in various interactions with the system and modeling it a similar way to other information records enables your users to find the right person when they are needed most…  




#3 -- Your next investigation lead is there already using unstructured-referencing

So we search without knowing what is it or who it is we are searching for. While presenting a list of results is relatively straightforward, anticipating the search result one will select from the list is a different challenge. The system should be able to leverage the user’s selection and timely present additional related “leads” based on that selection. Creating such intelligent connections between unstructured pieces of data requires not only the ability to index and store the text but also an ability to understand what it means. Our investigation may advance one lead forward if by selecting the cloud-drive news item from the search results, a getting-started-with-cloud-drive user post will appear as an additional suggested reading.


#4 -- Your IT can spot the big pains

When encountering an IT issue, some users prefer a free search interface, some may try asking their friends and others may opt for an official response from a qualified IT support engineer. A search query, a question, an IT support request or even a comment one posted, all are valid self-service channels modern users have grown to expect from their self-service system. All of these channels share one common characteristic: The input is an unstructured chunk of text that describes a user problem. Now wouldn’t it be cool if we could throw all of this text into one big processing machine and in turn get the most conspicuous pains and problems people are reporting.


Imagine the ability to correlate these textual inputs from multiple people and/or sources, automatically classifying it and use it to suggest to IT new informative articles to write or missing offerings to add in the catalog. How about using these questions or support calls to generate a new FAQ article for the Cloud-Drive or even a new automated support offering to enable users registration?!




The HP Service Anywhere search architecture and unified unstructured data model

HP Service Anywhere search architecture is built from the ground up with all of these use cases in mind. Let’s review what this “marketing” slogan really means:


Searching across multiple sources. You may argue that it’s only a matter of putting together a simple interface to merge the results coming from the social site, the IT knowledge base, the service catalog and the news portal. Most likely this is what you’ll find our competitors are doing. This approach while it looks the same from the outside is flawed in many ways. Consider how to compare the relevancy of results coming from different search sources. Articles (from the wiki?), news (from the portal?), user contributions (from the forum plugin?) and service offerings (from the catalog) each graded by its own search source. The issue of sorting the results also applies to implementing a decent as-you-type query completion (AKA auto-complete). In the absence of a good way to merge the results of multiple queries how can the system select the top five suggestions to put in front of the user? The only way to ensure relevancy-sorted results is to use a single search algorithm that runs on top of a single search database (AKA INDEX) in which the data is stored according to a common normalized data model.


Many ITSM vendors are adding “Social IT” to their marketing slogans. Leveraging the “knowledge of the crowds” using social input channels is definitely a good idea. But being “social” forces additional costs on the architecture. The data we search becomes ever dynamic. It’s no longer just IT that adds knowledge. It’s everyone. And it’s continuous. Integrating with a remote social system while periodically loading a list of questions is not fast enough. The search index needs to be updated in near real time upon any change to any of the relevant items it stores. Any new comment provided by an end user, any answer to a question, any update to an IT article should be updated in the search engine. The wave of users trying out the new Cloud-Drive service is coming TODAY. It will be very much useless if all their great social contributions will only be indexed tomorrow… HP Service Anywhere is built with a unique event system in place that feeds any entity update to the search index in near real time.


While non-HP vendors use standard search libraries such as Lucene, SOLR or elastic search, HP Service Anywhere leverages the industry leading Autonomy IDOL search engine. IDOL continually identifies patterns that naturally occur in a growing body of indexed text. It learns new vocabulary and terms as it’s introduced, or as it’s meaning changes instead of relying on grammar and linguistic rules. While the data each Service Anywhere customer stores on the IDOL index remains completely separate from that of other customers, IDOL is able to intelligently learn new meaningful patterns from all customers’ data.


This powerful and unique approach account for HP Service Anywhere’s ability to:

  1. Create smart meaning-based referencing between different records based solely on unstructured text and provide reading suggestions based on the user selections.
  2. Focus IT on the main user pains (Hot Topics) by detecting closely-related record groups that contain recurrent textual patterns.
  3. Support textual search in multiple languages as IDOL do not rely on grammar rules.
  4. Provide exceedingly better search results by leveraging the huge body of text gathered from all customers in order to learn meaningful patterns.


To summarize:

  1. HP Service Anywhere uses a common and normalized model for unstructured data. This enables a unique search experience across multiple information sources and record types.
  2. Near real time indexing of new records or record updates enables relevant search in continually updated social activity.
  3. Smart unstructured meaning based virtual referencing (Let’s see you say that quickly three times in a row) enables context-sensitive reading suggestions and hot topic analysis.
  4. Usage of the industry leading IDOL search engine results in better search results and multi lingual support.


We understand IT. IT Understands You!

HP Service Anywhere - Try it now!

chuck_darst | ‎06-04-2014 03:51 PM

This is a great post. I participated in a roundtable discussion a couple of weeks ago on self-service related topics and presenting current and relevant knowledge back to IT users/customers was a key stumbling block - once you got past password reset and dealing with common office applications.


I have to add a comment for two reasons


1. I like the concluding summary points! And maybe by posting a comment, it draws attention to the end of the article and then re-reading the summary


2. A shameless plug on the related topic of knowledge management! Register for an upcoming Relevant Knowledge Management - talking self-service to the next level brighttalk webinar on June 25th. George Spalding from Pink Elephant will be co-presenting.


Chuck Darst



| ‎07-06-2014 03:52 AM

dkalian - I am the last to say Google search applience isnt a very good tool and in fact its considered in the first list with IDOL in terms of enterprise search tools. Few differences you need to look at:

1. The article above is about an embedding of such a search tool into an ITSM product such that takes the search power and harness it within the ITSM product. Thus making all the product UI automatically search and context aware. There is absolutly no possible comparisson between the level of usability you get within the ITSM product in this way vs. an external integration otherwise.

2. Price. GSA takes a starting price of $20K while in SAW case you get a cloud-based, installation-free search with the possibility to index documents from on-premise databases included in the price of the ITSM tool (per sit).


With regards to the comment about SM - the post describes the HP Service Anywhere search architecture. Not Service Manager.


| ‎07-08-2014 01:04 AM

I can say that IDOL for sure has better language support, multiple datasource connectors (much more then SOLR has) and many more file-type indexing support - for more info Start here:http://one.hp.com/_layouts/OneHP/Group.aspx?gid=364


Leave a Comment

We encourage you to share your comments on this post. Comments are moderated and will be reviewed
and posted as promptly as possible during regular business hours

To ensure your comment is published, be sure to follow the Community Guidelines.

Be sure to enter a unique name. You can't reuse a name that's already in use.
Be sure to enter a unique email address. You can't reuse an email address that's already in use.
Type the characters you see in the picture above.Type the words you hear.
Showing results for 
Search instead for 
Do you mean 
About the Author

Follow Us
The opinions expressed above are the personal opinions of the authors, not of HP. By using this site, you accept the Terms of Use and Rules of Participation