Let’s understand this through an user perspective:
A typical BI project in traditional system used to be a big challange to take up to. The budgets required to deliver the output would be very high and so would be the time needed to implement these projects. For example, a typical traditional BI product, which requires creation of OLAP cubes, would take any where between 1 - 2 years depending on the state and readiness of the Organization. Typically, a lot of business requirement change in this time frame - new products come up, old lines of business get shut down, there could be mergers / acquisitions, so on and so forth. By the time the solution comes to implementation stage, it remains far less useful than what people would have thought at start. Having said that if you can mange the requirements well, a traditional system can yield a lot of benefits.
On the other hand, products like QlikView and Tableau work on very short implementation cycles. They don’t depend on creation of OLAP cubes to retrieve data. The end user or designer is free to pull in data as per their need. These solutions are also RAM based, so the hardware costs of implementation also run relatively low.
A good analogy to understand the difference is expensive SUV vs. low cost sports car. An expensive SUV might take more time to reach a place, but it does so in robust fashion. A low cost sports car can reach the destination quickly, but might need some engine tweaking and small changes to reach the destination.
Hope this brings out the difference.