In the past, multi-sourcing or aggregation was synonymous with bad respondents and experience. Legacy online data collection was conducted via double opt-in proprietary panellists. They were recruited from SEO-type campaigns and member-get-member programmes and in the early ’00s hit their peak with tag lines like ‘XXXL’ and ‘one size fits all’.
1) Direct commissioning with a research agency with a proprietary panel
Market research agencies with their own proprietary panels create their own sample supply chain networks via APIs to deliver a seamless end-to-end solution. (Private marketplaces)
Some of these still exist in certain highly saturated and competitive markets and for niche audiences. The challenge of these legacy models is cost and consistent engagement of the audience. Building proprietary online panels has always been tied to CPA/CPCs for advertising. But now this is becoming too expensive for research buyers to ingest into their budget planning and it is also why standalone ‘panels’ often have the most expensive cost per interview and incentives.
Today, online data collection is sourced, whether knowingly or unknowingly, from multi-sourced marketplace models. This practice has now become commonplace and light needs to be shined on it in the interest of transparency.
Understanding modern multi-sourcing
It has never been the case that ‘one size fits all’. No panel, proprietary or otherwise, can ever be big enough to support every research study all of the time. The utilisation of third party; or ‘top-up’ partners has always been part of this process; however, with automation and APIs in full flow, this has now become seamless and sightless to the everyday buyer.
Regardless of whether you are engaging a market research agency, a ‘traditional’ panel provider, or a marketplace direct, the chances are high that you are multi-sourcing. The important questions are: Do you know you are? Who are the sources? What measures are in place to prevent bias, quality issues, deduplication, and ensure better representativity?
As there are potentially endless scenarios for how many steps there are in a sample supply chain for a research project, for illustration purposes, we have chosen three common examples. All of them illustrate respondents entering and completing research projects. But it has to be asked what is more efficient? Which prevents potential data challenges? And which of these scenarios gives the researcher visibility of the journey itself?
2) Engaging with a traditional panel provider
Traditional panel providers bolster their proprietary offerings with API partnerships as well as consistent, manual buying’ through vendor networks.
Marketplace technology has evolved and is now able to combine panel member ID and device fingerprinting to allow the restriction of multi-panel memberships from entering surveys multiple times. This is not necessarily the case in the more traditional models, as this may be being undertaken at different stages/through different owners. The sample supply chains in images 1 and 2 illustrate respondents navigating an endless maze of vendors and channels to reach the ultimate destination of the customer survey. This struggle can also negatively impact the panellist’s experience.
Assuming these methods in 1 and 2 are employed, there needs to be tight cadence and management to avoid overlap, contain potential bias, and be mindful of the panellist journey. The same needs to be the case with quality controls: every source is different and is aligned to its own standards, so it is imperative that this is in place at a respondent level, not at the source level, otherwise you may not be seeing a consistent approach.
Additionally, when these more traditional ‘add-on’ models are utilised, the buyer is often unaware of the sample/supplier composition and whether controls are in place to limit single-source bias.
The marketplace model has become mainstream because of the efficiencies and benefits it can provide over traditional models. Speed and cost have become more important over time for data collection, with budgets and timings being constrained (even more so post-COVID). Multi-sourcing must be recognised as both acceptable and also the primary route of data collection today.
3) Engaging with a purpose-built multi-source
True marketplace models that are fit for multi-source purposes avoid supply-on-supply overlap and provide a single source of truth through a management layer.
What marketplace multi-sourcing enables
There are advantages to engaging a marketplace platform directly. Along with cost savings (by removing a step in the chain), you also receive transparency. Transparency of supply channels, individual supplier costs down to CPI level, and sample allocation to avoid potential source biases.
Researchers conducting online studies have increasingly adopted marketplaces that utilise multi-sourcing because the methodology delivers:
- Faster audience and quota delivery where and when required
- New channels for previously untapped/unavailable respondent sources (by engaging publishing networks, loyalty programmes, or brand-specific audiences)
- Increased inclusion through wider sources (vs. legacy SEO campaigns preaching cash for questions)
Given the number of routes that sample can fulfill survey requirements, it is important to understand the origins and the ultimate control measures in place. When combining data from different sources for analysis, it is vital to have common denominators or reference points to maintain integrity. Shouldn’t this be the case for the sources they come from as well?
PureSpectrum offers a complete end-to-end market research Marketplace and Insights platform, helping insights professionals make decisions more efficiently, and faster than ever before. Awarded MR Supplier of the Year at the 2021 Marketing Research and Insight Excellence Awards, PureSpectrum is recognised for its industry-leading data quality. PureSpectrum has developed the industry’s first and only respondent-level scoring system, PureScore™, creating a new standard of data quality for the industry.