Turning 2025 Learnings into Client Impact

This year’s conferences and client collaborations have reinforced a clear message: putting the consumer first is still the driving force behind product innovation.

Blue Yonder’s response is to focus even more on our 3 innovation pillars: Human Powered AI, Agile Obsession, and Real-World Moments.

Having spent 9 months this year on the front line of innovation, with our clients and our team, we’re reminded of Chief Innovation Officer Jonathan Million’s industry predictions for 2025, in which he said ‘smart agile methodologies will dominate’, combined with a human expertise partnership.

In this article, we will outline how we’re applying these priorities, and what that means for your projects, your teams, and the impact you can achieve.

1. Agility is a mindset before it is a method

Speed is non-negotiable, but meaningful speed comes from reframing the brief, not just pressing a faster button.

  • Start earlier with lighter designs. Smaller, focused samples and tighter protocols deliver direction in days, not weeks, when decisions are binary or directional

  • Co-create the “acceptable risk” bar. Agree margins of error and trade-offs up front so teams are confident acting at pace

  • SprintAI in practice. SprintAI resonates as a flexible wrap around AI and automation, not a black box. The win is configurable design plus next-day visibility to make a go/no-go call

Takeaway: Speed sticks when the organisation aligns to a lighter test that is good enough for the decision actually being made, and as early as possible in the process.

2. Human-powered AI beats platform promises

The conference sessions that landed paired transparent AI with human steering.

  • Two kinds of AI matter:

    1. Efficiency AI to automate the grunt work, accelerate desk research and free budget
    2. Added-value AI to enable analyses that were previously impractical
  • Cascaid is our added-value example. With a small amount of primary data, Cascaid infers the missing matrix to model preference drivers when classic mapping is not feasible

  • Guardrails are essential. No black boxes, show your working, and stress-test outputs so the AI cannot convince you of nonsense

Takeaway: Use AI to raise the starting line and to unlock the previously out-of-reach, while keeping experts in the loop.

3. Beyond liking: measure what changes behaviour

“Do you like it?” is only a sliver of the commercial picture. Behavioural change needs capability, motivation and opportunity. Whatever the method, think mindfully about the KPIs beyond Liking.

  • Ask replacement and repertoire, not just preference. Preference can be high without substitution intent

  • Design metrics for the decision. If the goal is in-market knock-out, mirror shelf, price and context

  • Emotions as a second layer. Growing momentum to blend emotion with liking to get closer to real choice, without treating emotion as a silver bullet

Takeaway: Choose metrics that predict action in the real-world, not just sentiment in a test.

4. Face-to-face qual is back!

There is a consistent theme: robust, skilled qual is scarce, but still so important.

  • Experience mapping wins attention. A tight, attribute-level language bridges consumer talk and formulation tasks

  • Ethnography and imagery travel further than means. A single expressive moment often mobilises a business faster than a page of scores

  • Netnography is useful, not sufficient. Online scraping gives a strong first pass; depth still comes from moderated, crafted conversations

Takeaway: Double down on qual craft. It is where language becomes specifications and where impact begins.

5. Granularity at scale: bots, databasing and the right jobs for each

The most interesting AI-qual stories were specific, not catch-all.

  • AI IDI bots can reduce interviewer bias and unlock scale for sensitive topics or very large sample size. They do not fit every study but belong in the toolbox

  • Databasing past testing changes the budget. With well-structured historical data, predictive layers triage what to re-test versus monitor, shifting spend to genuine unknowns

  • Data quality stays human. Automated checks help; “good enough for purpose” remains a researcher’s call

Takeaway: Match method to moment. Use bots where bias or scale demands it, and mine existing data before commissioning the next million.

6. In-the-moment still matters

Real context keeps us honest. From packaging handling to first-use rituals, in-the-moment tasks reveal frictions and delights that lab-like settings miss. Our Clickscape® approach also showed how simple, live signalling can turn passive attendance into measurable ROI for stakeholders.

Takeaway: Keep research close to the usage moment and give stakeholders hooks they can feel and share.

7. Market shifts you cannot ignore

A few currents with immediate implications for R&D and claims:

  • Holistic wellbeing and “little treats”. Consumers are buying moments of calm and competence amid uncertainty

  • Naturals and new ingredients. Curiosity is high but language is muddy; clear, consumer-grounded semantics matter

  • GLP-1 adoption in the US. Appetite-modifying medications change purchase drivers in food and drink. Taste is necessary but no longer sufficient; satiety, portioning and occasion are rising claims

How are we applying our learnings?

These insights are valuable, but what do they mean for you? Here are the key actions the Blue Yonder team are already putting into practice on client projects.

  1. Open every agile brief with a decision map that sets acceptable risk, sample and timing before method.
  2. Deploy a vetted AI IDI bot partner for sensitive, large sample size exploratory phases.
  3. Offer a beyond-liking module by default with replacement intent, repertoire and an emotion layer.
  4. Prioritise qual craft resourcing for experience-to-attribute translation on all R&D programmes.
  5. Audit client data estates to identify where predictive layers can reduce routine re-testing.

Conclusion: from conference notes to commercial advantage

Our 2025 learnings point to three essentials for winning in R&D: achieving confidence to proceed faster, using human-powered AI to push boundaries, and delivering insights in formats stakeholders can act on immediately.

There is a limit to what can be said in a general article like this. Whilst I hope you have found it useful, the truth is every manufacturing business, every client, every category and brief is different. So great work always starts with a conversation. It’s always about applying the right concepts to the specifics of a challenge.

Contact [email protected] to explore how SprintAI, Cascaid, or our experience mapping can support your next project.

Ready to talk?

Tell us about your goals. Whatever the stage, whatever the horizon, let’s find a way to get your business an edge.

2025-10-01T13:19:50+00:00
Go to Top