10 Key Themes & Learnings from Demo Days July 2024
By Insight Platforms
- article
- Artificial Intelligence
The Demo Days we held in July 2024 was another great showcase event for innovation in research tech. The event featured 16 innovative research and analytics tech solutions with very diverse capabilities.
This is a brief summary of the most interesting takeaways from the event: 5 big AI themes, and 5 things that were not about AI.
You can watch all the sessions on-demand here:
Demo Days for Research & Analytics Tools July 2024
First, the Obligatory AI Stuff
1. Conversational Everything
AI is helping to bring natural language interfaces to the whole research process: designing surveys, getting feedback from respondents, making analysis easier for non-experts.
Rival Technologies, one of the pioneers of conversational research, showed off the latest developments in their chat-based survey and feedback designs.
BoltChat AI uses autonomous chatbots to conduct interactive conversations with hundreds of participants – blending many features of qual and quant research. Their demo showed how their client Danone uses the solution for innovation in the yogurt and beverages category.
Glimpse is pioneering ‘talk to data’ or ‘chat with data’ approaches to simplify the way we analyse and query survey data.
Vurvey’s vTeam creates virtual agents – synthesised personas of research participants built on primary data – with whom you can interact using chat-based approaches.
2. Video Everywhere
The twin AI capabilities of Natural Language Processing and Computer Vision are driving huge growth (at much-reduced cost) in the use of video for research and insights.
I shared a presentation with a wide range of video use cases – it’s an extract from our AI Tools Landscape Masterclass – and then chatted with Dave Kaye, founder of Field Notes.
He explained how mobile ethnography and smartphone video research have developed, and shared insights from Field Notes projects in locations as diverse as the USA, Uganda and China
Rival Technologies uses video extensively in their chat interactions, and has been used for projects where thousands of videos have been collected from hundreds of participants.
QuestionPro demoed their new Research Suite capability to embed video feedback questions in the middle of surveys – giving respondents the option to speak or type their response.
And Vurvey – a brand name that comes from ‘Video Survey’ originally – collects video feedback to create their teams of virtual agents.
3. ‘Synthetic’ Data is Here to Stay
The use of the term ‘synthetic’ to describe generated data has been a bit unhelpful for the research and insights industry. With connotations of an ersatz, inferior product synthetic is an emotive label that has triggered some highly polarising debate.
But whether we call it augmented, enhanced or synthetic data, it’s here to stay.
Vurvey showed us how to ‘synthesise’ virtual agents that are trained on the data collected in primary video surveys. You can ask these agents questions – but if the answer is isn’t in the underlying data, they will cheerfully say, “I don’t know”.
Glimpse combines ‘talk to your data’ with synthetic data generation for new ways of interacting with survey datasets.
You can ask questions to an individual respondent or to a segment; answers will be based deterministically on data that was gathered in the survey – or modelled probabilistically by combining with an LLM. Glimpse also allows you to use synthetic / modelled data approaches to grow your dataset (by adding generated respondents); or extend your dataset but imputing the answers to unasked questions.
Learn more about the approach in this article: How Survey Research Will Be Transformed by AI.
4. Clients Still Worry About Accuracy and Security
These are still major concerns for clients when selecting AI tool partners for research and insights. Almost all the AI demos were keen to tackle these concerns head-on.
LLMs can still have a tendency to hallucinate. Solutions that use AI for analysis or synthesis need to clearly demonstrate that they can be relied on.
incling, for example, took time to build AI into their qualitative and communities toolkit methodically and carefully. They wanted to ensure the analysis would be both accurate and enhance the capabilities of researchers on the team.
Quillit’s qualitative summarisation and reporting tool places clear sourcing, citations and referral links prominently alongside its output – to build confidence with transparency.
Similarly, knowledge management solutions Market Logic DeepSights and Stravito Assistant are at pains to ensure their users can easily click between any summarised information and the original source material.
Concerns around security, privacy, and confidentiality of data are still a big deal for clients. They want to ensure that no data is fed back into training data for the big models. All the demos at this event featured some variant of “we don’t share your data – it’s never used to train the model – it’s all safe and private.”
5. The Era of of AI Realism
Boosterism, fear-mongering, media frenzy.
We’ve been living at the Peak of Inflated Expectations in the Gartner AI Hype Cycle for the last 18 months. We’re probably about to slide down into the Trough of Disillusionment.
But we’re a pragmatic bunch in the research industry. It seems we’ll treat those two imposters just the same. Many of the week’s demos had a thoroughly sensible and realistic take on the practical limits of AI.
MX8 Labs is a fully AI-native solution that automates much of the end-to-end survey research process. AI features are deployed in perhaps eight or ten different stages of that workflow. But the demo has a refreshingly honest view about where it’s not especially impressive. It may still save time – but don’t expect it to replace a capable research analyst just yet.
Similarly, MarketLogic DeepSights has a feature called AI Caveats. It draws the user’s attention to areas where the AI may not have synthesised its answers reliably, or may be missing key data to answer the question with confidence.
incling’s AI tools for analysis of qualitative research and communities were tested alongside human researchers – helping to flag where people still clearly hold the upper hand.
Similarly, Aurelius, a product and UX research analysis tool, built its AI Assist feature into its existing tagging and coding workflows – rather than creating something entirely new.
But It’s Not All About AI
Despite the astonishing pace of innovation in AI, many of these demos showcased innovations that don’t really have anything to do with AI at all.
1. Great User Experience
Research tech was late to the UX party. Mercifully, it’s now catching up and adopting good product design principles with gusto.
Knowledge management platform Stravito showed us how they grounded their whole value proposition in ease of use, simplicity of design and driving adoption with stakeholders. In their telling, these attributes are at least as critical as effective search and summarisation tools.
Upsiide has built an innovation testing solution with a unique mobile-first participant experience that uses simple swipe gestures to measure concept preference. It’s a clean, intuitive interface that maintains respondent engagement and shortens the survey experience.
AddMaple is a survey analysis and reporting solution with a founding team whose origins are firmly in product and UX research – and a clean, simple interface for publishing charts and dashboards.
2. Integration is Critical
Market Logic DeepSights has more than 100 integrations with different research and data suppliers, partners, publishers and tools.
incling – in a move clearly copied by Apple Intelligence – offers both native AI capabilities and the option to use specialist third party AI tools such as CoLoop.
Delineate’s real-time brand tracking and market intelligence data gets piped directly into some large customers in a continuous data stream via API. This means that research data effectively becomes a programmatic measurement source – rather than a series of disconnected research projects.
The research & insights tech and data ecosystem is maturing – and integrated solutions like these will be more and more in demand.
3. Research Rigour
The old rules still apply.
Core research principles of sound design, effective sampling and rigorous analysis often get overshadowed by the shiny new stuff. But these elements played a key part in many of the demos.
Innovation platform Upsiide – despite its apparent UI simplicity – has sophisticated capabilities under the hood for techniques such as portfolio modelling, incrementality and market simulation.
Advanced stats and analytics capabilities are built into the InfoTools Harmony solution – one of the few ‘research native’ BI platforms out there.
MX8 Labs, a recently launched agile research platform, has all the stat testing you’d expect from more established, full-suite survey tools.
4. Real-Time Data
Glimpse, MX8 Labs, Rival and BoltChatAI are all ‘agile’ research solutions. You can get feedback from hundreds of people in under an hour with these tools.
Delineate’s value proposition is delivering real-time measurement into organisations.
Upsiide supports brand and innovation teams with iterative product and comms testing research – getting answers in hours and then generating fresh new concepts based on that feedback for another round of testing.
5. Communicating & Embedding Insights
The last of our big non-AI trends is about being commercially useful: helping stakeholders make decisions or take actions based on sound data and insights.
Several demos went large on this.
Aurelius is all about surfacing user insights from qualitative research so that product and UX teams can make the right changes.
InfoTools is about helping stakeholders do their own analytics and data visualisation by tailoring their platform to the specific needs of an organisation.
AddMaple is about making survey analysis easy for anyone.
QuestionProbe BI automatically generates dashboards from survey data that can then be .
MarketLogic DeepSights, QuestionPro Insights Hub and Stravito Assistant all exist to embed insights more deeply into organisational decision-making.
This collection of demos clearly show how important AI already is for the research and insights technology; but it also revealed how many other components that make for a successful research tool and product.
We’re grateful to all 16 companies who shared their solutions and innovations with the Insight Platforms audience in this event. You can also watch dozens more on-demand demos from previous events here: