In context

It stands out as a compelling case for the targeted functionality of analytic applications. At the end of 2001, US-based catalogue retailer Lands End, which mails 269 million catalogues to consumers each year, set out to enhance the data analysis capabilities of its sales, marketing and production managers.

By implementing customer, supply chain and service applications from business intelligence vendor Business Objects, the company calculated it could reduce the number of catalogues sent out annually by one million – without reducing revenue. The upshot: it would increase its profits, as measured by earnings per share, by a full 1c.

Lands End is hardly alone. Since 2000 when analytic applications moved beyond their established base in financial departments, tens of thousands of organisations have sought out applications that address specific business activities or job functions. While these may lack the flexibility associated with traditional business intelligence tools (which can be customised for specialist users' requirements) in select business areas they match users' requirements well and, as with other packages, can be implemented in a fraction of the time than a semi-custom system.

Analysts have tipped the analytic applications market for explosive growth. Henry Morris at IDC for example, estimates that the business intelligence (BI) analytic applications market was worth $2.5 billion in 2000, but will almost treble is size by 2005 to $6 billion.

Not surprisingly, vendors have been eyeing this lucrative opportunity for some time, reckoning that such applications can be sold to a much larger pool of end users including CEOs, middle management or line managers. In contrast, "BI tools have [often] been targeted at just a small percentage of end users in the BI ghetto of organisations," argues Keith Gile, senior analyst at Giga Information Group.

As a result, a slew of BI tool vendors including Business Objects, Cognos, Hyperion, SAS Institute, and Informatica have rolled out packaged analytic applications. There software generally falls into four core business activity areas: finance, customer relationship management (CRM), operations and production (including supply chain), and business performance management (BPM), says Morris.

The only problem with this picture is that outside of traditional areas of finance and business performance, the uptake has been much slower than ever- anticipated. Particularly disappointing have been sales of analytic applications such as CRM, operations, supply chain and production.

But why are vendors struggling to convince organisations to buy — instead of build – analytic applications?

Explanations for the slower than expected sales depend on the vendor, but include the need to still tailored the analytic software to the requirements of specific vertical industries, evidence of (in some cases) a lack of return on investment, a perception by IT departments that such software leaves them out of the purchasing decision, and the high price tags of some of the software, and the fact that, despite their packaged status, nearly all of the applications still require a large amount of customisation.

Packaged promise
On paper, the benefits of packaged analytic applications are compelling. As with other business packages, applications promise to significantly overcome the time and cost involved with writing and refining customised queries, analysis functions and reports for end users. By slashing the application development cycle, organisations not only save money, but they free up developers to concentrate on other BI projects such as building cross-departmental data warehouses.

Also, the capability to deploy applications quickly is vital for organisations that need to swiftly bring an analytic capability online, whether it is assessing the success of a new product launch or gaining insight into their supply chain.

For example, Netherlands-based pension fund manager PGGM deployed SAS's performance management Balanced Scorecard application to analyse its pricing policy for pensions. Before activating the application, PGGM set targets for how much it cost the company to sell each product. This included the amount of time it took employees to run various checks on clients. SAS's application was able to show PGGM was pricing 80% of its products too low, as its employees were taking too long, and costing too much, to handle the offerings.

For such organisations, extending analytic applications out to more end users promises to broaden inter-departmental collaboration. This addresses a criticism of BI tools in which end users often can only access 'stovepipes' of data, or customised multi-dimensional cubes, tied to a specific department or job function.

In contrast, a suite of analytic applications typically contains common modules and templates to run query, analysis and reports against a relational data warehouse that spans multiple departments. This makes it easier for middle management users to identify and act on spikes and troughs in demand for products and services across multiple operations. For example, a purchasing manager using a supply chain analytics application could cross-check CRM-based sales data before deciding how many products to distribute.

It was this capability that convinced mobile device manufacturer Motorola to deploy Informatica's CRM, supply chain, and business operations analytic applications back in 2000. Motorola estimates it will get a $200 million return on investment by using Informatica's applications to analyse sales, production and inventory levels across several of its global business units, claims Matthew Goldsbrough, European marketing director at Informatica.

Application apathy
Despite these endorsements, most organisations are cautious about analytics application purchases. According to an April 2002 survey by Giga, 74% of companies said they were not going to purchase an analytic application in the next 12 months.

The fact is that most so-called 'packaged' or 'shrink wrapped' analytic applications require a substantial amount of customisation. Kevin Magee, sales director at BI vendor Information Builders says, "An out-of-the-box application may meet only 50% of the needs or at best up to 80% of an organisation's needs. This means that they do not always shorten the time and cost of implementation."

At the crux of the problem is the fact that most vertical industries require very different functionality from an analytics application. This is particularly true for CRM, and operations and production applications, says Morris at IDC. This would account for the decision by Siebel Systems, the clear leader in the CRM market, to tackle the CRM applications with packages specific to 20 industry verticals.

So any company planning to deploy a 'one size fits all' CRM analytics application should be prepared to put in the customisation groundwork, says Bob Skeens, UK country manager at BI tools vendor MicroStrategy. For example, the data model used by a bank to describe a customer would include terms such as interest, current account or loan, but this data model would be vastly different for a pharmaceutical company.

In addition, both types of company would also run very different queries, analysis and reports. "An organisation that decides to implement an analytic application for CRM would have to tear out its existing data model and replace it with a model based on the templates of a vendors' application," argues Skeens.

The need for customisation means that IT departments need to be heavily involved in the purchasing decisions of analytic applications. But such applications are typically sold directly to the business executives who will use them — and often IT is only consulted when it becomes clear how much customisation work is required. Such approaches can lead to a highly fragmented BI deployment where individual departments use very different, and very incompatible analytic applications from different vendors.

'Justifiable scepticism'
Gile at Giga, says IT departments are justifiably sceptical about analytic applications. "A key reason is that BI vendors have failed to prove a return on investment for analytic applications."

The bottom line is that organisations are being asked to spend on average between $250,000 and $750,000 for applications that still require a large degree of customisation, adds Gile. This is obviously a step many organisations are unprepared to take.

In fact, several BI vendors have already withdrawn from, or baulked at entering, the packaged analytics applications market after discovering slow demand. For example, MicroStrategy released three applications for CRM, supply chain and web analytics during the first half of 2000, but stopped marketing them as packaged applications at the end of 2001, says Bob Skeens, UK country manager at MicroStrategy.

The functionality of MicroStrategy's applications is now bundled as modules with the company's core BI tools suite, MicroStrategy 7i Platform. This enables developers to build customised applications, but at the same time use generic application templates for CRM or supply chain applications.

Gile at Giga concurs that applications are priced too high. However, he adds, "analytic applications are a great idea, but vendors just need to adopt a different [more honest] approach to selling them". For example, if vendors marketed the clear benefits of analytic applications, but also mentioned substantial customisation work is still required – their sales would be much stronger, says Gile.

Even executives at vendors such as Informatica and Business Objects admit sales of analytic applications still only account for a small portion of their revenues. In its 2001 annual financial filing, Informatica singled out the "uncertain acceptance" of its analytic applications. Meanwhile, Business Objects – one of the few vendors to break out its analytic application figures – stated that analytic applications only contributed $8 million to its total revenues of $415.8 million for 2001.

In defence of vendors, the analytics application market is still immature. In fact, Goldsbrough at Informatica says, "The analytics applications market is at a stage where the enterprise resource planning market was 10 years ago."

However, if the market is to realise its true potential, vendors will have to provide a more convincing argument than they have so far demonstrated.


Turning web metrics into web insight

Web site analytics software has become an indispensable tool for organisations with an active online presence. By capturing and analysing the massive volumes of data generated by customer interaction on a web site – from click stream patterns to transactional data – they can make informed decisions about how to, for example, drive up site traffic, target customers more accurately, optimise their web site designs, measure the success of marketing campaigns, and ensure more customers find and buy what they want.

But web analytics is still in its infancy. Many basic tools provide facilities for logging raw web site data, but only a handful provide highly-sophisticated business intelligence facilities for using that data to optimise the relationship with the web site user. Even fewer enable companies to integrate that web data with customer interaction information gathered from other points of customer contact such as the call centre.

US retailer Macys, for instance, has tried to improve merchandising and operational decisions relating to its site using Accrue's G2 analytics tool. "It helps us gauge customer behavior and assists us in matching their needs with appropriate content and sales," said Gary Beberman, director of technical research at G2 tool is used to analyse conversion ratios, including content-to-purchase and browse-to-buy ratios, the overall streamlining of the sales process, abandonment rates and conversion metrics for product lines.

The starting point for all tools is raw metrics. There are multiple ways of collect data on web interaction, including extracting data from web logs, using packet sniffing technologies to examine packets travelling between the web client and web server, timing how long users spend on one page or how long it takes to complete a transaction, their click stream movements, and so on. Most of these are server-side technologies, but many more advanced techniques download and embed JavaScript code on a web page and feed that information back to the server. This code will then collect all the data about page interactions directly from the user's browser.

However, all of this information-gathering exercise is pointless if the tools are not calibrated correctly.

John Woods, CEO of analytics vendor Site Intelligence, recounts selling to one online retailer who was already using a basic web analytics tool. "They were tracking the conversion rate on their site, but they had wrongly configured the tool. As a result all their key performance indications were wrong," he says. In fact, the company's conversion rates were out by a factor of seven, so it believed it was converting very few customers and thought it would have to redesign its site completely. In fact, the conversion rates were actually relatively high and the real problem was that the site simply lacked visitors.

Accuracy is paramount. "The more precise a company can be about their metrics the more effective it [the analytical software] will be," says Martyn Reeves, vice president of CustomerCentric's European operations. Part of this may involve considering how an organisation's web analytics software fits in with its overall customer relationship management (CRM) initiatives.

As that suggests, web analytics is inseparable from customer relationship analytics, and the goal for many organisations is to fully integrate data about web interactions with other channels such as call centres and physical stores. This enables companies to see, for example, how the same customer behaves in both the online and offline worlds, and to achieve a complete view of behaviour.

But the 'soft' nature of such web data – especially data such as click stream analysis – means it does not integrate easily with wider CRM systems. While most vendors are working towards this goal, its realisation is some way off. "There's no way in real time to sync up that data [between call centres and live website activity]," says Rand Schulman, senior director of product management at web analytics software company WebTrends.

For example, it is difficult, if not impossible, to accurately align data from sales that begin on a web site (where a customer may have discovered or researched a potential purchase), but which are completed in the offline world (either through call centres or in shops).

Web analytics software is not cheap, with sophisticated systems costing up to six figures, but it promises important business benefits in terms of both operational cost reductions and revenue generation. Leading vendors in the market include Accrue, Site Intelligence, WebTrends, CustomerCentric and Speed-trap.

Many companies still make decisions on their web sites based on what they think customers want and simple observations of how they behave. But doing so without web analytics software is flying blind.



Avatar photo

Ben Rossi

Ben was Vitesse Media's editorial director, leading content creation and editorial strategy across all Vitesse products, including its market-leading B2B and consumer magazines, websites, research and...

Related Topics