Doing the work engineers hate
An interview with Mercury Interactive Corporation's President & CEO, Amnon Landan
Interviewed by Thomas Caldwell

A little more than a decade ago, Amnon Landan and some of his fellow engineers came to the realization that the growing complexity of software was mandating equally complex testing tools to ensure proper functioning. So he and his friends started a company to develop and provide them. Today, Mercury Interactive is considered one of the world leaders in software testing tools for virtually all major platforms. The Sunnyvale, California-based company started up its Japan operations some four years ago, and established its own Japanese subsidiary two years ago. On a recent trip to Japan, the founder and head of the company took time out to speak with Computing Japan's Thomas Caldwell about one of the more unpopular, yet very important, software development tasks.

(CJ) One of the first things I learned about the software business is that it is the only industry of its kind where a product is sold on the basis that there is no guarantee it will work as advertised. The customer then pays additional money to fix the original problems. Your company seems to have taken advantage of this fact. How would you describe the Mercury Interactive's business?

Amnon Landan: What we do is we sell solution products to automate software testing.

(CJ) Would it be fair to say that your company does the the job software developers should already be doing?

Landan: There are often systems that come online that should have been more thoroughly tested by the vendor or by the in-house programmer who wrote it, but that's not really the big problem we have to solve. When a company creates its own software application, all sorts of things are included in the process; the operating environment, your own unique business processes, your own infrastructure and so on. What is created is a very unique system that is duplicated no where else. It is the only one of its kind in existence. Such a unique thing is inherently complex. You just can't expect it to work properly at first. To put it simply; we help people put it all together and see how it really functions (I)before(I) a million people start using it and the success or failure of a business depends on it.

(CJ) That sounds like a great way for a company to save a lot of money, if not customers. Basically you are an interface between the people who are developing software and the people who are using it. How much money can one of your customer's expect to save?

Landan: There are several ways for measuring the return on investment in our products. Let's say you are a good organization and you already extensively test your products. Mercury Interactive automates the process. So instead of twenty people hacking on computers, you only need three test developers and our tools. Our tools cost less than one developer. The cost of less than four people as opposed to the cost of twenty. That's the least you get for you money. Secondly, there is the cost of failure. If by using our tools you drastically reduce your cost of failure, how much is it worth to you? Let's take e-commerce for example. A typical large-scale e-commerce application can cost a company six to eight million dollars an hour if it goes down. So if you use our tools, and I save you just one hour of down-time, there's a savings of at least six million dollars. Then there is the economics of time to market. Testing is the most time consuming phase of the software development process. If we can accelerate your time to market, how much is it worth to you? These are very simplistic examples. But the short answer to your question is "a lot."

A typical large-scale e-commerce application can cost a company six to eight million dollars an hour if it goes down.

(CJ) You talk about bringing software products to market. So you don't just service, for example, banks with internal systems. So testing of retail software products is also one of your strengths?

Landan: Retail, internal, everything. The higher the business criticality [of the application], the more people use us. It is a risk-based decision. The higher the risk, the more a company wants to offset that risk. Another factor that determines the risk a company has riding on a software project is the level of complexity, but it is mainly how critical it is to the health of the company. Many years ago, software was more or less self-contained. It only ran on one system inside one organization. These days business systems are often directly connected to your suppliers, customers, and other concerned parties all at the same time. There is a lot to lose if something goes wrong.

(CJ) Do your products find the easy bugs so the the humans can go after the bigger problems?

Landan: In a way we replace human testing. However, we are able to produce situations that would be difficult to impossible for human testers to simulate. How would you generate a load condition on your system that simulates ten thousand users accessing it at the same time? There is no way you can do that manually. No way! So we are able to bring software testing to a new level of sophistication that didn't exist in the past. Do you still need human testers? Absolutely! Why? Because we are providing a tool. It's only as good as the people who use it. Testing is not easy. It goes way beyond just using [our] testing tools. Knowing what you should be testing and taking action based on the results of the testing is crucial.

Even my own mother, who for years didn't quite understand what I did for a living, called me up one day after reading a story in the popular press about Y2K and said she finally knew what I meant by "software testing."

(CJ) In 1999, when someone in the computer business talks about software testing, three letters come to mind: Y-2-K. Most of the concern this year over whether or not computers will crash at midnight on December 31 concerns how well the testing of re-written software goes. How has this affected your business?

Landan: Y2K has had several effects on our company. The least important is that it accelerated our revenue growth. It [Y2K-related sales] came from nowhere to make up about one quarter of all our business. That has declined and Y2K is now responsible for about ten percent or our business. The reason for the decline is that the larger institutions believe they are already Y2K complaint. Whether they are or not is another matter. There is no question that Japan is still lagging when it comes to Y2K, so we are pushing that business here. But from a global business perspective, it is history. One very positive thing about Y2K is that it opened many new doors for our company and our technology, as well as for the entire [software testing] industry. It legitimized what we do. Even my own mother, who for years didn't quite understand what I did for a living, called me up one day after reading a story in the popular press about Y2K and said she finally knew what I meant by "software testing." In a way she is a good representation of the CEO crowd. They never understood the importance of testing. Now they do.

(CJ) How does the development of software in Japan compare to the processes used in the United States, Europe, and other countries where you sell your products?

Landan: Software testing has always been the unwanted step-child of the engineering community all over the world. Engineers hate testing. You can't expect them to love it since it is a form of criticizing their own work. Moreover, most engineers don't know how to test. Seriously! If an engineer doesn't know how an application is suppose to work from the user's perspective, there is no way they can test it. The best testers are users. If you have an application that does loan calculations you need a banker to test it. If you have a stock trading system you need a stock trader to test it. How can your average developer know how these things are suppose to work in the real world? The answer is; they can't.

(CJ) Since you get to see up close how software is developed in different countries, do foreign-based companies in Japan have any advantage over local businesses when it comes to developing applications?

Landan: The more Web-based a system is, the more of an advantage they have.

(CJ) Why is that?

Landan: Because now the issue of time-to-market [for a software application] is not measured in years or months, it is measured in weeks on days. A Web-based, online transaction system is constantly being updated and improved. Products, information and the like are being added and removed all the time, and everything has to be tested before it goes online. It is a business that moves very, very fast and never stands still.

Mercury Interactive Japan K.K.
5F TG115 Building
1-15-7 Toranomon
Minato-Ku, Tokyo 105 Japan
Tel: +81-3-3500-5161
Fax: +81-3-3500-5162
info@mercury.co.jp

Back to the Table of Contents

Comments or suggestions?
Contact cjmaster@cjmag.co.jp