Establishing a living corporate knowledge base

The lack of a systematic means for employees to retrieve and access the information they need to carry out their jobs effectively is a frequent oversight, even among the largest and most successful organisations. Susan Gordon and Gloria G. Kinrot describe how Comverse worked to develop a collection of tools that have helped address this failing, in turn creating within the company a living corporate knowledge base.

You work in a large company and need to answer customer queries quickly and efficiently. Where do you find the product information you need in a hurry? Locating just the right piece of company or product information can send many into flurries of searching through the tangle of computer document files, subdirectories and corporate intranet websites, not to mention the piles of old proposals, product descriptions, technical bulletins, features lists, user manuals and whatever else is stashed in your own ‘library’, organised according to your own idiosyncratic system (or non-system).

Once you’ve located information that appears relevant, do you know if it’s up to date? What if the person you used to rely on for answers has moved to another division? Who can help with a quick answer now? Was what you wrote in your last proposal to that big client consistent with what your colleague detailed in his most recent proposal for the same product to another client? Is everyone selling the same features as part of the same release? Have these features been developed or is what you’re promising still on the development roadmap? On top of all this, if you’re new to the company, where can you learn quickly about all the company’s products, their various capabilities, services, components, and operation and management requirements?

Even some of the largest, most successful organisations continue to struggle with the glut of information at their disposal. The lack of a systematic means of retrieving and accessing valid, authorised information is, unfortunately, all too common. At Comverse, a supplier of messaging software, we have combined concepts from knowledge management with the techniques of requirements management to develop a ‘living’ knowledge base and three user-friendly tools that seek to address this problem.

The off-the-shelf application and database we use is Requirements Traceability Management (RTM) by Integrated Chipware, based on Oracle. First introduced at the company four years ago, RTM is the basis for the corporate knowledge base that we created in Comverse. The knowledge management applications discussed in this article have been created to tailor the RTM tool to Comverse employee needs and business objectives.

A collection of tools based on a taxonomy we developed, along with the offline ability of users to access the entire knowledge base, offers a practical solution to the difficulties of sharing information broadly among proposal and document writers throughout the company. For those who need this information to sell products or write proposals (answers to customer tenders), using online tools is impractical because network connections to company servers are often too slow or non-existent. We needed a solution for those who work on aeroplanes, in hotel rooms or at home.

With the tools we have developed, anyone with a laptop can carry a copy of the full knowledge base without requiring a modem or network connection to access information.

The solution was developed over a four-year period (with many struggles) through the efforts of a small team. The knowledge base eliminates information redundancy and inconsistency since there is just one source for update and retrieval of all text and auxiliary information, which was formerly found in multiple documents.

Three tools based on the centralised knowledge base have been created at Comverse:

RFP Assistant – providing a way to search for and insert answers to customer query documents;
Doc Assistant – enabling the quick generation of topic-related documents ‘on the fly’;
AskHQ – a visual e-learning interface, used online to access all information in the knowledge base.

Taxonomy

First, let’s look at the taxonomy used to classify chunks of information that were captured from legacy documents at the establishment phase of the process. This is handled through the RTM by means of a MS Word interface tool.

Every product is ‘established’ and classified by information analysts. When this task is completed (at Comverse, this usually takes between four and six weeks per product), the knowledge base becomes the information source and the original documents are no longer required. Document creation will ideally be done periodically as outputs of the knowledge base. The knowledge base should become the living source of information for the organisation.

To establish the knowledge base, legacy documents – including product descriptions, product boilerplates and well written proposals with generic product information as sent to prospective customers – are captured in chunks (often paragraphs or whole sections) into the RTM. Document logic is transformed into requirement logic as each document no longer depends on the narrative flow in which it was originally written. Each chunk receives a unique name. After naming each chunk, the analyst determines and assigns a general area of interest (equivalent to a ‘heading one’). We then define cross-product categories (‘heading two’) and further break these down into specific and related topics (‘heading three’). Individual requirements thus have four levels of categorisation: area of interest; domain (or related) category; requirement topic; and, requirement name, the most specific label for each piece of information, each knowledge base ‘object’. There are two ways to visualize this hierarchy, either as a pyramid, with the first level – the broadest – at the bottom, or as a table of contents outline, with the general headings and their corresponding sub-headings underneath (see figures 1 and 2).

Figure 1.

Figure 2.

After categorisation is complete, process metadata can be added, including such attributes as product availability, product release, date of last review by the product manager and so on. The knowledge is now based on a single corporate vocabulary and taxonomy.

At first, the process of naming and classifying was long and arduous because it evolved from the material rather than being applied as a preconceived structure. However, now that the structure exists, we have successfully used it in all subsequent establishments of other Comverse products and have even tested it informally with information from other industries, and confidently offer it to others as a means of supporting their own attempts to organise information.

The important thing about this hierarchy is that it is associative and flexible. We believe it can be tailored to fit many kinds of industries and that adapting it requires serious thought, but the logic seems to hold up in the various test cases we experimented with, including automotive and home appliances industries.

This taxonomy is the basis for all the tools we offer to our corporate sales staff throughout the customer-facing groups worldwide. The following sections focus on each of the tools.

Answering customer queries: RFP Assistant

Answering customer questions and meeting their product requirements is a complex and time-consuming job. For those writing the associated, and often extensive, documents, finding or creating accurate information to answer specific queries usually involves searching through a number of existing sources (ie, previously written – and successful – proposals, product descriptions, boilerplates and user manuals etc) or trying to locate a product expert. Cutting and pasting has become a routine activity for those who need to write quick answers. The life and death of a company’s sales figures depends on the confident presentation of adequate, accurate and authorised information that successfully answers the client’s requirements and concerns about the system or product they are looking to purchase. Add to these pressures the limited time sales and marketing personnel spend in their home offices with access to the company’s computer-based resources. How much can be downloaded to a laptop and, then, where does the person begin to search for the information necessary to answer specific questions? With all the documents generated within a company, how can anyone be sure what is the most up-to-date material? Who is the correct product expert to help with questions that cannot be easily answered? The document chaos that exists is cause for confusion, frustration and, ultimately, sometimes even the creation of fictitious proposals. Without a centralised source of accurate, up-to-date, authorised information, each sales representative must rely on their own resources – a personal database of sorts that is not available to anyone else except through informal sharing arrangements.

A solution to this challenge has been the development and introduction of an offline tool currently in use at Comverse. The tool is an interface between the online, centralised RTM knowledge base (described above), which contains product marketing requirements for every product being sold.

A macro-based application with an interface developed in Visual Basic sits on each user’s C drive. As an offline tool, the RFP Assistant contains a mirrored flat file version of the knowledge base stored in the RTM Oracle database. Monthly updates are sent to users incorporating any new changes made in the knowledge base, including textual or graphical changes to the data. The taxonomy described in the previous section is visually represented in the RFP Assistant and enables searches to be carried out via the hierarchy or in ‘free search’ form.

The tool enables the user to load in the RFP received from the client and then conduct searches, preview the possible answers to decide which chunk best fits the issue in question, click on a ‘comply’ button and insert the relevant piece of text into the proposal document. Figures and tables, by means of embedded macros in the RTM, will pop into place where marked.

The proposal writer proceeds through the RFP systematically, answering all the questions. If he doesn’t find the information for a question, the writer needs to find the information from other sources, for instance from a product expert who dictates something over the phone. That new information is typed into the appropriate space and, for internal purposes of the draft only, will appear in another colour (‘track changes’ is enabled at this stage). In addition, the user is able to edit the material, customising it for his client as needs dictate. At the end of the process this material (modified or newly written) will be sent back to the knowledge base team to update the RTM. The changes and additions will appear in the next update of the batch files sent out in the next snapshot of the database to all users of the RFP Assistant.

Using such a search and insert application provides ideally 75 to 80 per cent of generic (non-customer specific) information to the user and cuts many hours, sometimes days, from the proposal writing process.

Quickie documents: Doc Assistant

Having the ability to create a quick document can come in handy when preparing for a meeting, a presentation or even a more formal document such as a marketing brochure or specialised product description. Another tool, the Doc Assistant, enables the user to instantly generate documents of this kind. The tool is built on the same hierarchical organisation as the RFP Assistant, and the user searches according to the hierarchy, retrieving whole sections of information. Each section of the newly generated document can be labelled separately and accumulated section by section, if so desired. Editing and making narrative transitions are completely left to the user’s discretion. What he has retrieved via the tool are the chunks of information and any associated figures and tables embedded in the given objects.

Putting the Doc Assistant in the hands of the technical and marketing documentation people in the organisation is the real process solution to ensuring that information is continuously updated and written at the highest level.

Visual e-learning online knowledge base: Ask HQ

In addition to the two offline tools used for creating either proposals or other product marketing documents, we developed an online tool that sits on the company intranet and is available internally to all those connected to the company network. This tool facilitates the dispersal of information to the general back-office corporate staff to look up new terms or to learn about Comverse products using a visual search engine, which is based on the product taxonomy described above.

Company-wide dissemination

What we have described are the tools and the knowledge base taxonomy, but what makes this a living knowledge base? The tools and taxonomy are themselves static, so let’s turn to the process itself to understand what is innovative about it, how it ultimately transforms the static to the dynamic, and how users themselves are transformed from passive information seekers to active participants responsible for feeding the knowledge base and keeping it ‘alive’.

Methods for updating content

Without periodic updates, a knowledge base can quickly become an historical archive of obsolete material. Having a knowledge base that sales and marketing staff can rely on means keeping it up to date. In order to accomplish this, we rely on three basic processes:

Busy people cannot devote time to a cumbersome review process. To aid our product managers, we developed a ‘PM review’ macro, which enables them to quickly reconfirm or modify and confirm knowledge base objects. They can also indicate which ones should be deleted. In our experience, product managers are more responsive and amenable to reviewing short requests than when presented with large quantities of information in daunting formats. The knowledge base team makes their changes in the knowledge base, and a batch change is made to the ‘date of last review’ attribute, visible in the RFP Assistant preview screen;
We also decided to attack the problem from the user’s end. Those who most need the information are the ones most interested in seeing it in an accurate and up-to-date form: the sales and marketing staff themselves, who answer customer queries and need to know that they are giving out the correct information. The solution was to build this process into the proposal writing task. By creating a mechanism in the RFP Assistant, as well as activating a comments button inside the Ask HQ tool, we are able to get the necessary response. If the RFP writer cannot find the answers to his questions in the knowledge base, he must do what he used to do before when he couldn’t find an answer in his stash of documents, ie, call a product expert and get the answer over the phone or via e-mail. As soon as that information is written into the RFP, it shows up in red. The tool has automatically turned on the track changes macro in Word. The writer then clicks the ‘save proposal’ button. He cannot leave the program without doing this. An automatic message pops up with an Outlook e-mail and the proposal already attached. The writer is then ready to send the proposal to the customer; the tracked changes disappear, all new material is accepted and the proposal is ready for delivery. When the saved proposal arrives back to the knowledge base team, the document still has the track changes macro enabled so that the new material is clear to the analyst updating the knowledge base. This material is checked against the appropriate product. It is quite possible, of course, that the product manager was actually the source of the new information;
The last way of having users participate in the updating process is via the Ask HQ tool. Since this tool is available to everyone in the company via the intranet, anyone can comment on the information. When each comment is saved, it is automatically linked to the object in the knowledge base. A knowledge base team member can run a script and periodically check for new comments, analyse the material, verify the suggestion with a product manager and update the knowledge base.

By following this process, the knowledge base becomes a dynamic resource rather than a static storehouse, enhanced by active participation rather than mere passive usage.

The challenges associated with setting up a knowledge base

We have found that professionals from fields such as library science and other humanities disciplines that utilise archiving techniques do well at the job of building and maintaining a knowledge base. A technical background may be helpful but is not the principal criterion for analysing and categorising material. A knowledge analyst requires a strong ability to abstract the essence of a text, compose an appropriate and concise label for that text, and categorise it according to the hierarchy. In addition, the analyst should be meticulous, patient and have strong editorial skills. Bear in mind that the establishment of a knowledge base is a major and time-consuming job. A reasonable time estimate is six weeks per product, depending on the amount of material that needs to be included.

Perhaps the most difficult challenge in institutionalising such new ways of working is in generating the necessary paradigm shift within the organisation itself. Changing the organisational mindset and processes in order to instil a sense of responsibility for entering information into the knowledge base and for using the tools in the first place is a huge hurdle. Change management methodologies, targeting the right people and securing the appropriate levels of commitment and support are all crucial success factors.

Acknowledgements

We wish to acknowledge the help of those who participated in this project: Roy Amitzur, the manager who enabled it all to be put to the test and really happen; tool developers Alex Josefsberg, Zvika Rom, Dan Ranon, Guy Aharon, and Shai Aharon; and, Steve Menear, an initial inspiration for using offline tools that mirrored the RTM knowledge base. Additional thanks to Dr Lawrence James from Integrated Chipware, Leah Goldin, who worked in the early stages of introducing requirements management at Comverse, Paul Hirschhorn, who headed the first RFP Centre in the company, and of course the other knowledge analysts, Lia Disatnik and Reuven Genn.

Susan Gordon is senior information analyst, Corporate Knowledge Services, at Comverse. She can be contacted at sumik@inter.net.il

Gloria G. Kinrot is marketing group manager at Comverse. She can be contacted at gloria.kinrot@comverse.com