Programmatic sampling is the automated buying and selling of sample, where sample buyers and sellers meet in a marketplace environment to conduct transactions. In programmatic sampling, APIs are used to connect and match a seller’s panel and respondent profiling data against the buyer’s survey needs. In this environment, transactions happen automatically with minimal human intervention. When technology takes over the traditional sampling process (historically carried out by human beings using spreadsheets and emails) companies benefit from large-scale efficiencies such as lower costs, reduced error rates and, ultimately, higher quality outcomes.
Despite its clear benefits, there are many myths surrounding programmatic approaches to sample. These myths add up to the persistent belief that automating the sample process will negatively impact data quality. In fact, the opposite is true.
Programmatic sampling has made massive strides over the past few years. It now allows us to not only deliver on demands for speed, but also to become more agile. Increased dependability and accuracy contribute to better data quality. We write more about the details surrounding data collection and sample supply chain automation in our new white paper, ‘Transforming data collection without paralysing your business‘.
Back in 2018, Cint’s COO JD Deitch wrote about a few of the myths surrounding sample for Research-Live, touching on some of the key processes that automation can solve including recruitment sourcing, fraud mitigation and respondent engagement. He wrote: “It is time for us to reset our understanding of good sample practices….The online experience and the technology that enables it have changed dramatically since market research moved online some 20 years ago.” Yet, three years later, some are still debating the benefits of programmatic sample. We’ve recapped some of JD’s thoughts on these myths below.
Myth #1: Programmatic sample is river sample.
In the late 2000s, as Web 2.0 applications and smooth, enjoyable, and interactive user experiences became the norm, online panel response rates began to tank. At this point we saw the rise of non-traditional sample sources from the broader advertising/marketing ecosystem. These new sample sources—commonly called “river sample”—were delivered to surveys differently: they were directed there programmatically and not by invites. Early river sample truly was crap, and people were right to be suspicious. However, sample provision has massively changed. “Programmatic” refers simply to the automated transaction. Sample quality for these sources is at least as good—and often better—than that of traditional online panels due to the sophistication of these providers.
Myth #2: Automation will increase errors.
In fact, automation reduces the mistakes that inherently come with manual processes, actually exposing weaknesses in the process that are human-driven or less precise. A programmatic approach equals better operational outcomes and accuracy across more project phases, including bidding, pricing and feasibility. Also bringing automation into the field monitoring stage, which is still painfully manual for the most part, can help proactively spot and even solve problems quickly. This kind of real-time problem detection can lead to midstream supply corrections to live insight into survey design issues at any hour of the day. This clearly reduces errors and has a huge impact on outcomes.
Myth #3: Programmatic sampling will not get me the right respondents.
In fact, in today’s ecosystem, automation is perhaps the ONLY way to get the right respondents and access representative sample. With a programmatic approach, the respondent ‘supply chain’ can be completely automated and thus optimised to the benefit of all parties. When recruitment is done right, sample suppliers source from hundreds of partners of all types, providing diversity while minimising biases that arise with concentrations of similar types of people. Sample acquisition improves as new technology and APIs allow the acquisition/recruitment of panelists at scale. Algorithms that combine anti-fraud, respondent engagement and actuarial approaches will allow new respondents, those who would not ordinarily join a panel, to participate while ensuring they remain real and engaged people and profitable for the business. Traditional, manual research processes and supply-chains are simply too inefficient to facilitate representative sample at the scale and speed needed in today’s ecosystem.
Myth #4: I don’t need automation to fight fraud.
This myth is actually as dangerous as it is outdated. In fact, researchers need fraud mitigation techniques that meet today’s sophisticated fraudsters head-on. As technology use rises across the board in the industry, individuals are also using technology advancements to up their “fraud game.” Hackers are finding the gaps in traditional fraud detection approaches. At Cint, we track a multitude of data points, so this increase in fraud is not just conjecture. Using a more programmatic and automated approach, utilising artificial intelligence and machine learning, allows the insights space to stay top of evolving fraud trends. It’s the only way to stay a step ahead of the organised, agile fraud of today.
The costs of not automating and implementing a programmatic approach to sample are mounting daily. Data quality goes down with more error-prone processes; speed to insights is negatively impacted, as is feasibility; fraud continues to mount. Programmatic sample means that insights companies can become more agile and deliver better data – at the speed that is required in a fast-paced marketplace. It’s time to stop ascribing to the myths and start looking at the results.