Assessing the value of business intelligence

One of the most difficult issues companies face is evaluating new technologies such as business intelligence (BI) software. Often IT employs a total cost of ownership (TCO) model to evaluate how the technology affects the bottom line. These models, designed initially for PC deployments, are extremely complex and may not take into account your organisation`s specific BI requirements.

Alternatively, IT may engage a consulting firm to recommend a short list of favourite vendors and perhaps even the criteria upon which to evaluate the vendors. Unfortunately, there is no guarantee that the vendors on the shortlist will provide the best solution for both the IT groups and the business users.

Unless you fully consider both sides of the equation - key costs and end user value - you will deliver an unnecessarily expensive "solution" that users ultimately reject. So how do you effectively evaluate a business intelligence solution? It is critical to identify those areas that constitute the majority of a BI deployment`s costs. Questions such as: "How quickly can it be implemented?" and "How many IT resources will be required to maintain it?" help you get to the heart of a BI solution`s TCO.

Additionally, you must assess the value the BI solution provides the end-user community. Is it easy to use? Can users answer their own questions without involving IT? Does the BI solution meet the various needs of each type of end-user from the power user to the report consumer?

Studies on technology deployments identify dozens of cost elements IT should consider when calculating TCO. However, only a handful of variables determine nearly the entire cost of a BI deployment. They can be placed within three major categories: purchase costs, implementation costs and maintenance costs.

By focusing on these critical cost factors, you can confidently understand nearly the entire cost of a BI solution without becoming bogged down in the complex TCO models that are not suited to analysing BI deployments.

Primary considerations:

* Accurately matching analytic power with user needs - don`t over-equip.

* The versatility of the BI solution - from ad hoc query and drillable reports to dashboards.

* Tight integration between all functions within the solution.

Because purchase costs typically consume one-third of the total cost of BI, you must carefully evaluate your business users` needs and purchase the solution that best meets those needs - no more, no less. Vendors too often convince customers to buy more BI power than they need, which unnecessarily increases the cost per user and drains budgets.

Consequently, companies must resist the temptation to provide ad hoc capability to everyone. Although ad hoc querying capability is a potent weapon in the hands of a skilled analyst, it places more demands on clients and servers, which of course adds costs if new servers must be purchased or clients upgraded. In its own TCO research, Gartner promotes the practice of understanding user requirements. "It is almost as inappropriate to provide a worker with too much functionality as not enough. Functionality adds complexity; complexity adds cost."

In the end, carefully segmenting the user community by need enables IT to properly equip - at a reasonable price - each user in the organisation who requires BI.

The versatility of the solution is another purchase consideration. Vendors must provide a variety of tools and delivery mechanisms to meet the disparate needs within a business. Offerings should at a minimum include a consistent, easy-to-use interface, powerful ad hoc querying capability, drillable reports, and dashboards. Organisations that require flexible report distribution should seek some sort of adaptive reporting capability. That is, the ability to build and distribute a report once while allowing each user viewing the reports to see only what she is authorised to see.

In fact, a powerful, flexible reporting solution will satisfy the vast majority of business users. Meta Group concludes, "...as the number of users expands, business intelligence tools are becoming available to people who are not business analysts in any sense, and do not need, or intend to do, complex analysis, but, instead, need to look at reports." In fact, nearly 80% of all business users need to view parameterised reports only. To give them more is to waste money and over-burden resources.

Any BI solution should enable easy dashboard creation. A rapidly growing trend is the use of dashboards to deliver key performance indicators in a graphical format. The dashboards are highly intuitive because the graphics are drillable, allowing even the least technical users to make discoveries quickly and easily. A final purchase consideration is the level of integration between the various products as well as the different delivery mechanisms. For instance, does the BI solution enable users to easily distribute reports from within the query tool or must they open a separate tool with a different user interface? Can Web and non-Web users share reports with each other or does the Web product create different file types than the client/server version? These capabilities are critical to successfully rolling out a BI solution. If it cannot perform such basic functions, it is not worth any purchase price, no matter how inexpensive.

Understanding the basic capabilities of a BI solution enables IT to deliver the most cost-effective, applicable analytics to each user group within the organisation. With that in mind, you should ask BI vendors the following key questions to determine whether they will provide the best solution for the cost:

* How flexible is the reporting solution? Does it provide adaptive reporting?

* How easy is it to create dashboards?

* How well integrated are your query and reporting tools?

* Are the user interfaces identical for each type of tool and for each delivery mechanism? Are they easy to use?

Primary considerations:

* Time to deploy.

* Ability to leverage existing environments - no proprietary obstacles.

* Minimal training requirements.

Although consulting costs and user downtime add significantly to the overall costs of implementation, the lion`s share of implementation costs is the time of the IT personnel required to deploy the BI solution. The basic formula is the value of their time multiplied by the length of the implementation. Generally, three types of IT professionals are necessary to deploy (and maintain) a BI solution:

* Database administrators to manage the connection of the BI tools to the data stores.

* Programmer analysts to build metrics and develop reports.

* Systems administrators to manage user accounts and resolve network issues.

Some BI vendors require more resources, depending on the complexity of their products. Because an IT professional`s time is so valuable - anywhere from R300 to R800 an hour - it is imperative that the BI solution be quick to implement. One of the most important issues that affect time to deploy is whether the solution can leverage the existing environment or if IT must create a proprietary semantic layer through which the users will access the data stores. Many of the leading vendors require that customers build these complex, proprietary layers before any business users can begin to use their tools.

Implementation costs skyrocket - and time to deploy expands to months - as IT staff struggle to create the semantic layers for each data source. It is essential that IT gain an understanding of the time to deploy of any BI solution under consideration. Several businesses have learned a painful and expensive lesson by not exploring each vendor`s time to deploy claims. Another implementation cost variable is training time. If the solution is simple to learn, with intuitive and consistent interfaces, training costs are minimised. Researching third party sources for ease-of-use metrics will help identify the BI solutions that your users will learn quickly. All vendors claim their tools are easy to use. It pays to seek confirmation from objective sources.

Implementation costs can consume a large portion of an IT budget. In order to identify those vendors that will offer the lowest costs and the highest value for BI, ask the following questions:

* What is your average time to deploy for a company our size?

* Can your BI solution leverage our existing environment or does it require proprietary semantic layers be established?

* Is your solution recognised by any objective sources as easy to use?

Primary considerations:

* Ease of metadata maintenance.

* Value to all levels of users.

* Simplicity of upgrade process.

Since maintenance costs can consume anywhere from one-third to over half of all BI costs, the ongoing expenses of BI should be weighted more heavily than implementation costs and must be carefully examined before choosing a BI solution. As with implementation costs, maintenance cost factors are a function of time and IT personnel costs, but are incurred indefinitely. These costs can get out of hand if the wrong BI solution is implemented. Asking the right questions surrounding maintenance will help you quickly assess the major portion of a BI solution`s TCO.

The ease or difficulty of metadata maintenance is a critical cost consideration, especially in today`s fast-changing climate. Query and reporting needs are dynamic in any business. New data sources are added and users` analytical requirements change constantly. IT resources spend most of their time updating changes in metadata to reflect the most recent modifications. Unfortunately, just as some vendors` proprietary semantic layers add complexity and expense to the implementation process, they also complicate metadata maintenance. Each time metadata changes are necessary, IT must reconfigure the metadata in two places - their own and the vendor`s proprietary metadata.

It is also important to consider whether a BI solution enables your best analysts to make important discoveries while also allowing other users to gain valuable insights from the data.

A more cost-conscious strategy for IT would be to investigate BI solutions that offer a simple semantic layer for non-technical users, but that also enable power users to drill anywhere within the data. Power users make discoveries only when they have the freedom to navigate where their questions take them. It is critical to every business that their best analysts remain unrestrained by their tools.

A third major factor to consider when evaluating maintenance costs is the simplicity of the upgrade process. As more organisations demand Web-based delivery of BI to ease maintenance costs, it is important to understand how different vendors enable upgrades. So-called zero administration clients are an effective way to transparently upgrade a user`s BI tool via a plug-in or an applet that automatically detects a client`s software version. In most cases, it is a simple process. However, some vendors do not permit incremental upgrades and require customers to uninstall the old version before installing the new version. The same concerns apply to server upgrades. Taking servers offline in order to upgrade software is a risky proposition. Clearly, this kind of upgrade requirement is inefficient and time-consuming for IT departments.

The above issues should be the main criteria when calculating a vendor`s maintenance costs. To take a short-term view of a BI tool is an expensive mistake. As a guideline, ask your BI vendors to answer the following questions:

* How is metadata managed with your solution?

* Do power users have the freedom to make their own discoveries or are IT resources required to respond to their needs each time they change?

* What is the process for upgrading your software?

Wise buyers investigate the upgrade process of every BI vendor they are evaluating.