A one-stop enterprise search shop?

Imagine this scenario: a sales person at a paper manufacturer needs to access a variety of information to ensure that they will meet an ambitious target for the current quarter.

Imagine this scenario: a sales person at a paper manufacturer needs to access a variety of information to ensure that they will meet an ambitious target for the current quarter.

Some of that information is external, some of it resides in the company’s databases, and some of it is on file systems scattered around the organisation. Instead of logging into several different systems to retrieve the information, they access just one: www.google.com.

By clicking “include secure content” on the familiar browser interface, which sits alongside the usual images, groups and news categories above the Google search box, the executive can retrieve that information straight away.

By typing a prospect’s name into the search box, for example, the Google interface provides them with information direct from the company’s Siebel customer relationship management system. That includes contact details and lists recent communications with that prospect. They do not even need to spell the name correctly.

By typing in “sales in the South West”, Google brings back a real-time sales figures graphic from the company’s Cognos business intelligence system. The sales executive does not need to know the exact name of the report and the graphic appears directly in the browser interface, so there is no need to click on a link to view it. A search under “South West sales” would deliver exactly the same authoritative and accurate result.

By typing in a purchase order number, Google recognises the number and retrieves the appropriate records from the company’s Oracle financials system, as well as finding and retrieving a document from a file server and a contact name from an Outlook contact list. “Google is not just for web searchers any more,” says Jim Murphy, an analyst at AMR Research. “It increasingly becomes the human interface for enterprise information.”

Rather than a last resort when a user cannot navigate to what they are looking for, he says, Google becomes the starting point for enterprise information access and retrieval.

With that goal in mind, Google released version 4.6 of the Google Search Appliance in April 2006, which incorporates its Onebox for enterprise architecture.

The Google Search Appliance is two years old, but this latest announcement represents a step forward in its enterprise capabilities. Although Google Search Appliance already enabled companies to add search to their enterprise networks for structured and unstructured information, Onebox extends that search to dynamic, real-time data held in enterprise applications, financial records and business intelligence repositories through a single search request.

This requires close partnerships with enterprise systems suppliers, and Google is relying heavily on such support to develop modules that will allow its products to interact with Onebox. To date, Google has lined up about 60 partners and systems integration companies to do so.

“With eager software suppliers clamouring for attention, Google has the luxury to choose the best fits, the complements that meet common demands,” says Murphy. These partners will build and offer connectors that allow access to information stored in their systems, as well as provide business logic that goes some way toward interpreting user queries and determining relevancy.

One such area is business intelligence. “Cognos and SAS Institute have both developed demonstrable integrations with Google,” says Murphy. Another is CRM. “Many CRM systems have failed, or at least struggled severely, for lack of accessibility and usability for a wide range of workers,” he says.

Google has already announced partnerships with software-as-a-service suppliers NetSuite and Salesforce.com. Other Google Onebox partners include Oracle and networking firm Cisco, which plans to open up its Meetingplace conferencing system to the Google Onebox interface.

Analysts have largely praised Google’s strategy for the Onebox architecture. “If you provide applications or data that enterprise personnel may need to search, evaluate Google’s API programme immediately,” says Whit Andrews, research vice-president at analyst firm Gartner.

But, he warns, companies that take this advice should carefully examine the depth and sophistication of the connectors that Google’s partner suppliers build. “Do not simply assume that a connection is deep and productive,” he says.

However, the ability to search across multiple data types is nothing new to most companies: they have been able to buy it for years from enterprise search suppliers such as Fast Search and Transfer and Autonomy.

For that reason, the incumbents in enterprise search will not be driven away. High-end systems that cost hundreds of thousands of pounds but which are engineered to produce more relevant results across a wider range of sources are likely to continue to be the choice for large enterprises with specific business needs.

In some cases, these are more sophisticated at categorisation: sorting keyword searches by concept and context. Others surpass Google in text mining, or deriving meaning from unstructured data. These companies and their systems integration partners have been working for years with large enterprises to build taxonomies specific to their own requirements and the information they hold and need to search for.

In February 2006, for example, Fast signed a contract with legal and professional publishing company LexisNexis, which will see Fast reselling and integrating its Enterprise Search Platform with LexisNexis’s existing human-built taxonomies. The plug-in application will be integrated with 24 industry-specific taxonomies, as well as daily general business and news updates.

Rather than rely on third parties to provide business logic that goes some way toward interpreting user queries, as Google does, companies such as Fast and Autonomy have already built these capabilities into their own products.

These companies can point to some sophisticated applications of their technologies: the rapid retrieval of clinical trials data at a major pharmaceuticals company; real-time access to customer data by call centre operatives; the construction at one of the world’s leading automotive companies of a “learning network” that delivers content to employees from corporate applications and legacy information systems; as well as third-party research libraries and news websites.

That has led many search specialists to position Google’s Onebox as an entry-level offering for small companies that want to provide search at a departmental level.

Larger enterprises may need more sophisticated technologies. One approach is that taken by enterprise search company Autonomy, which uses advanced Bayesian techniques and background tracking capabilities to take account of the context of a search query.

This works because information workers typically search for whole ideas, while Google’s approach is limited to searching for keywords.

“The search engine in a large company needs to understand the meaning of the information, who is authorised to see it, to alert others when it is searched for, and to ensure it is accessed and used in strict compliance with legal and regulatory mandates,” says Mike Lynch, chief executive of enterprise search company Autonomy.

An emerging method for helping knowledge workers to access information is to “search the search”, whereby the search engine uses information about previous searches in the same area, both by the current searcher and by their colleagues, to deliver a more tailored result. Autonomy’s Idol product pools search information, so that searches made by one team member can narrow the search made by another.

Google Onebox is a combination of hardware and software designed to index all forms of content on intranets and websites. Google says its approach to enterprise search provides a robust, scalable and cost-effective way to deploy businesses enterprise search. David Chalmers, European programme manager at Google, says, from a technical perspective, the Onebox approach is to simplify the deployment and licensing of enterprise search. “We want to shield IT from the complexity of hardware and software integration,” he says.

“There is no need for a big-bang approach to search,” says Chalmers, which is a strategy often used to deploy enterprise search across large companies.

Users can start with a smaller box, suitable for the number of documents they wish to index.

The starting price for Google Onebox is £21,000, including two years’ support. This rises incrementally depending on the amount of documents categorised.

“Users frequently complain that enterprise applications are hard to learn, especially when they are needed infrequently,” says Brown. “That leads to employee frustration, quantifiable training costs and lost time associated with finding, accessing, and using information stored deep inside CRM systems, document repositories, corporate directories, portals and intranet sites.

“The idea that, just like on Google.com, two keywords and a search box will let them retrieve a preformatted live purchase order from an enterprise system without leaving the familiar search interface, is very compelling.”

And that idea is already attracting some interest. Google does not reveal exactly how big its enterprise search business is, other than to say it is about 1% of revenue, which suggests sales of about £11m in its latest quarter. It does claim more than 3,000 customers and enterprise sales growth in excess of 100% last year.

Gartner values the enterprise search market at £200m this year, up 10% from last year.

So, although there is a place for high-end enterprise search, it looks as though businesses will adopt Google. “At its core, Onebox makes searchable information useful. It breaks from the traditional notion that search is just a list of links and lets IT departments deliver highly relevant, up-to-date, useful information directly to users based on what they ask for: not what information designers think they want,” says Brown.

Read more on Business applications