Crowd-based Requirements Engineering (CrowdRE)

Today, every product has a “crowd”; a very large, heterogeneous and physically distributed set of users interacting online. This is true for software applications (especially mobile apps), but also for any kind of hardware and even everyday objects. Anyone who knows how to deal with this crowd can channel the power of literally everyone and thus gain a market advantage over competitors who do not recognize this potential.

A crowd does not mean the queue at the supermarket checkout or the crowds on the morning commuter train. The crowd goes far beyond that. Think for example of crowdfunding. Here, the crowd voluntarily pays money to support an initiative they like. In this way, start-ups receive seed capital, small businesses use it to finance the development costs of innovative products, and campaigns for good causes reach interest groups to collect donations for their projects.

Companies are also increasingly outsourcing tasks to the crowd (crowdsourcing). These range from simple tasks such as categorizing images or filling out a questionnaire, to creatively complex tasks such as suggesting solutions to problem scenarios or submitting visual designs. Either everyone receives a reward or only the one who has the best idea, or there is no reward at all (e.g. Wikipedia).

The crowd can also make valuable contributions when using a product. The members of a crowd usually express their wishes, ideas, criticism and demands very actively. This so-called user feedback or crowd experience is often shared on review portals. This knowledge is therefore no secret; it is available online and it would be a waste not to use it to understand what the crowd likes or dislikes.

But a crowd is different than the expression of an individual’s opinion! A crowd is a group of hundreds to millions of people with a common interest. Yet all these people are different. They use your product differently and communicate differently. Some express their opinions loudly, some can be persuaded to express their opinions, and some say nothing. As a result, what a crowd produces is not the product of single individuals, but the product of many interactions that are more useful to your business than you would expect!

Why conventional survey techniques are not enough

A product will never be successful if the consumer market is not considered. Who will buy and use your product? How will your product help the user to achieve his goals? Does the product meet the expectations of your customers? These and other questions must be answered before even the simplest product is developed. Without a broad and deep understanding of the product’s target group, the risk of failing in the market increases.

Interviews, questionnaires and workshops are some of the most popular and proven techniques in product management, requirements engineering and market research to understand what the market wants and needs. In combination with creativity techniques, they can be used in many environments to gather detailed opinions, wishes and requirements. However, these techniques are only suitable for a few stakeholders and do not scale in large and distributed user groups.

But if you want to understand all your users worldwide, traditional methods of requirements engineering (RE) quickly reach their limits. But how can you then obtain important information from hundreds, thousands or even millions of people? And how can you filter out those whose valuable comments contain the solution to a problem that could give you a significant advantage?

This is exactly where Crowd-based Requirements Engineering (CrowdRE) comes in!

What CrowdRE exactly does

Perhaps you will recognize yourself in the following example: A successful development company has several apps in its portfolio. Since they offer very effective and intuitive solutions for everyday problems, most of them have become very popular. The number of downloads is overwhelming, as is the number of reviews in which users praise or criticize the apps or ask for currently unavailable functionality. All this information about crowd experience provides important input for management decisions on some of the most pressing issues.

Once the number of people expressing their opinions about your product or service reaches thousands and more, you have access to a vast amount of potentially extremely valuable text-based data. This data is available to you to help you develop your product or service. But how do you deal with this feedback? When this data is processed manually, it is difficult to draw parallels and objectively prioritize the issues that should be addressed. Without tool support, collecting, aggregating and analyzing this natural language data provided by the crowd is impossible.

With an automated method, all this knowledge could be used in no time at all and new and improved functionalities could be brought to market even before the competition. Reports on bugs, complaints, feature requests, praise and quality requirements could be analyzed automatically.

Crowd experience analysis is suitable for such situations. Automated, recurring review analyses enable management to make informed, high-quality decisions in a timely and cost-effective manner, based on statistical evidence and enabling them to prioritize key needs and trends. This not only supports the company’s growth, but can also help it to strategically diversify its portfolio, specialize on its strengths and personalize its services to its users.

If you want to do it, then do it right!

Big Data analysis for social media is nothing new. However, most data collection and analysis tools only provide partial answers to the question of what the crowd wants and only cover certain types of feedback from specific parts of the crowd. Because they are not able to integrate information from different sources, it is unclear whether the results complement or contradict each other. Furthermore, the results are usually not validated, but their interpretation is left to the analyst.

Our CrowdRE not only offers a methodological approach that supports traditional RE techniques by extracting knowledge and information from large groups of stakeholders, but also provides a variety of tool-based services. We follow a “crowd-based” approach, where the crowd provides the basis for answering your questions in real time, at an acceptable cost and even with a quality suitable for important management decisions.

How does this work? First, you need to select the sources from which the data will be extracted. In the above example of an app development company, the relevant crowd experience data is mainly found in app stores. Our sources are continuously updated so that we can tailor our services to your wishes and needs. There is still enough room for customization and future enhancements of our already very powerful tools!

The data is then analyzed using “mining” techniques. These techniques use patterns that discover explicit and implicit problems, needs and ideas. Text Mining involves pre-processing crowd experience texts and scanning them using our finely tuned language patterns. For example, you can identify positive, negative and demanding statements about features and quality aspects of your product. We also optimize the identification and classification of statements for your context.

You receive all results either as a report created by our experts or via a web-based dashboard that allows you to navigate through the data independently.

Note: This article is based on a Fraunhofer IESE whitepaper. Fraunhofer IESE is a leader in the field of CrowdRE and was developing the approach we offer.

‍