The global pandemic has further accelerated the digitalization of market research, making its democratization a reality. Driven by new technologies and new consumer approaches, research strategies have fundamentally evolved. Today, market research in Do-It-Yourself (DIY) mode is more and more popular for strategic decision-making. This transformation is beneficial for all market players, beyond the obvious budget and time savings. Do-It-Yourself has become a leading research tool and for Askia this has been a great opportunity: our clients can build their platforms around our core technologies and utilize the automation and extensibility we have been building into our development roadmaps for the past few years.

Platforms enabling DIY are here to stay and many more will be appearing on the market. But what makes a good DIY platform and what will these look like in the future? Here we summarise some of our insights from building self-service platforms with our clients and share some of the related new features we have already begun working on.

DIY – Data reliability?

Reputedly the poor relation of traditional methods, DIY has long suffered from negative connotations in an online world, where research expertise and respondent quality have been unfairly forgotten in favour of price and speed. But that’s where data reliability starts – with the relevance of the questions and the quality of the answers.

Given that the target market of DIY is among people who do not come from the world of research, the user-experience must be understandable to all, allowing inexperienced users to launch the studies quickly and easily. It is in this context that “tailor-made” support by research experts is essential. This support should be offered at all stages of a study (choice and structure of the sample, construction and programming of the questionnaire, analysis of the results, etc.). The platforms must also be complete with analysis tools and comparison options. The robustness and validity of the proposed research solutions are also crucial elements for strategic decision-making. The reliability demanded from traditional research is also expected from DIY research.

Towards a collaborative world

The DIY market is constantly evolving and will be accompanied by more technological innovations from Askia and others like us. However, DIY is misnamed because it is more and more a question of accompanied or assisted DIY that we could call Do-It-Together (DIT). This human and technological assistance will allow a rise in power for DIY.

The emergence of users from various non-research backgrounds means that research managers within companies have to play an “educational” role. For the market research agencies, this new DIY is a real opportunity because it allows them to attract new research buyers. On the platform side, Askia is working on the development of user interfaces that adapt to the level of skill and experience of each research client, tailoring both the functionality offered and the support provided.

The growing need to support multiple collaborators on the same study, so that everyone can contribute to the project, will require us to develop more multi-user interfaces.

“One stop shop” – an all-in-one platform

Today’s DIY platforms are mainly limited to online quantitative studies. We need to go further to give clients direct access to a wide range of solutions provided on one platform – qualitative, community, syndicated, telephone, F2F. The customer experience and efficiency gains would be enhanced by using the features of an integrated platform.

Statistical tools are not (yet) all automatic. Customers are invited to program their questions, but what about evaluating the results? Some platforms already offer access to their databases and standards, allowing clients to evaluate their results in relation to their market. An added value for such platforms would be to allow clients to store their research data, to build up a history and to use it as a basis for comparison & benchmarking. We also want to conduct more complex statistical analyses; the ideal would be to offer a whole range of statistical tools on a self-service basis, where the most appropriate statistical test is automatically used, based on the type of data that has been collected and business question that needs to be solved. In the meantime, agencies will have research experts there to support their clients.

Bringing the “real” is also one of the promises of Artificial Intelligence (AI)

Artificial Intelligence (AI) has a role to play in enriching the answers given to open-ended questions. AI will partly adopt the role of a moderator. Depending on the respondent’s answer, smart probing will automatically try to find out more information from this person by asking them about their response.

AI can help with closed questions too. For example, when Ipsos built their FastFacts system with Askia, they added an AI-enabled question library, which allows users to tap into a database of 250,000 questions. This empowers users to find question wordings and answer categories that have already been vetted and fielded by Ipsos researchers.

At the analysis level, AI via machine learning1 systems will not only detect statistical differences, errors or behavioural anomalies, but it will also help us identify and sort out what is truly salient. Today, only humans can reliably make this selection and find perceptive insights. 

The future is in data hybridization

Those who predicted that social listening, behavioural metrics and open data2 would replace the so-called standard studies have surely been disappointed; there has never been as many studies conducted by questionnaire as there are now. But the future is indeed in data hybridization, mixing genres by linking data associated with a loyalty card or an online behaviour to the sampling engine, for example. The idea is to constantly enrich the information available on the respondents (whilst still adhering as strictly as necessary to data protection laws and industry codes). What we are interested in is not only a static picture at a given moment in time, but also an understanding of change; for example: why did you decide to change your car brand?

Another way to maximize the use of available heterogeneous data is to build probabilistic models. By using open data, we can extrapolate local estimates from global data much more precisely. The holy grail is to measure trends over time and to superimpose them, to “deseasonailze” them and to correlate them by means of an enormous storage and restitution capacity. Surveys are one of the components adding to a ‘data lake’3 that allows for better modelling of behaviours. And at a time when data mining4 is becoming commonplace, including among advertisers, the role of platforms will no longer be to simply provide PowerPoint presentations or even dashboards, but rather data flows via APIs5.

It is clear that DIY technology platforms have a bright future. They will continue to grow and when it comes to our development roadmap, Askia is aiming to offer more solutions, more features, more usability, and greater simplicity. We will continue to leverage AI and various sciences to make them easier to use and gain better insights. And to complete their evolution, these platforms will have to be collaborative and agile – true Do-It-Together (DIT) platforms. Having said all this, these platforms will nevertheless remain complementary to traditional methods, where the relationship and human intelligence remain indispensable, mainly for the most complex study or sample designs.


  1. Machine learning: a field of study in artificial intelligence that uses mathematical and statistical approaches to give computers the ability to “learn” by themselves without explicit programming.
  2. Open data: digital data with free access.
  3. Data lake: a storage place (usually in the cloud) for massive, heterogeneous data (structured or not, from different sources).
  4. Data mining: a data analysis technique that extracts the most relevant information and indicates the relationships between the variables detected.
  5. API: Application Programming Interface (or Application Process Interface) is a catalogue of functionalities that can be used by other applications / services and facilitate the exchange between them.

This post is based on an original article published on the website (in French), written by Frédérique Alexandre, Global Activation Leader of Ipsos Digital, and Patrick Lassale, CEO of Askia.

Photo by Tyler Maddigan on Unsplash