BusinessWeek declared 2007 as the year of the mashup. But that year came and went and then some, so looking at 2010 the question is “What is a mashup and are people interested in using them?”
Last week, Allan Wille joined industry leaders to discuss the idea of The Last Mile: Data Visualization in a Mashed-Up World. An MP3 of their discussion can be found on the DM Radio website.
According to Wikipedia, a mashup is an “…application that uses or combines data or functionality from two or many more external sources to create a new service”. However, that definition could be used for nearly everything. And the term becomes a bit more grey when you consider what can be constituted as part of a mashup – must it be totally dissimilar data, or can it also include historical data to put things into context?
Technically any location where data is co-located can be considered a mashup. You iGoogle page is a mashup of RSS feeds from different sources. A website that includes a company’s twitter feed and a calendar to display upcoming events could be considered a mashup.
In business intelligence (BI) terms, a mashup is something that combines information – enhancing context – to provide understanding. As with any BI system everything needs to be placed into context. Mashing up sales revenue and temperature would not provide relevant data, unless you were running an ice cream store. But mashing up something like revenue and hours of overtime can provide some very relevant contextual information.
Mashups have traditionally promised two benefits. The first is to provide accessibility, which has been achieved through introduction of visualization. The idea is that more users can take advantage of the data because they can understand what is being portrayed through the graphics. The second point, and one that mashups have only had limited success at delivering upon, is ease of implementation. This is the idea that any non-technical user could create their own mashup. Today, however, this is mostly limited to RSS feeds which a user can set up, such as an iGoogle page.
While people are not necessarily using the term “mashup” many are implementing the concept whether they realize it or not. More companies are combining data from not only their own data sources, but also 3rd party systems, especially as more SaaS systems are deployed.
There are dangers associated with mashups. As more data is available, it can be overwhelming to determine what information, and from what source, is relevant to the key performance indicators (KPIs) that a company should track. The idea is to really think about what matters as opposed to what can be mashed together. Data quality can also be a concern as information is pulled from multiple sources, which may or may not be qualified and cleansed. The final issue to consider is when drawing relationships. Is the data correlated, as in the case of sales and temperature, or causal, which means that there is an actual relationship between the data points.
The future of mashups will include adding more relationships in clearly displayed formats. The new relationships will include things like sentiment and social data to understand the impact of these activities. And improved visualization will enable more users to benefit from BI without a technical background or in depth understanding of how it works.
