When AI can help your team generate hundreds of product ideas in the same amount of time it took to produce a handful, all that possibility creates another problem. How do you prioritize all those product ideas and which ones deserve more attention?
Imagine you’re developing the next generation of cold beverages for a major coffee chain. Your team might explore fruit flavors, protein with existing coffee bases or new blended drinks. When you add AI-assisted ideation, it’s easy to scale the ideas to 100 or more.
But what happens when you need to narrow down all these ideas or concepts?
Traditional idea screening and innovation testing at scale (Think: 50, 100 or 300 early ideas) with human respondents simply isn’t practical for your budget. So how do you bring consumer input into the process before you’ve narrowed the field to a handful of ideas?
That’s where synthetic research comes in. Instead of recruiting real people to evaluate every early idea, synthetic research uses AI-generated respondents to simulate how consumers may react. These synthetic respondents can help teams identify which ideas show enough potential to move into deeper product concept testing with real consumers.
In this article, we’ll look at how to approach product idea prioritization, where synthetic research fits and how Zappi’s Synthetic Idea Pre-Screener can help teams narrow a pool of dozens of early ideas to the ones worth testing with real consumers.
The innovation bottleneck has shifted from coming up with enough different ideas to deciding which ones are worth pursuing.
The challenge is figuring out what to do with everything you’ve generated.
Traditional prioritization frameworks like RICE and ICE can help. RICE (Reach, Impact, Confidence and Effort) scores ideas against these four factors, while ICE (Impact, Confidence and Ease) scores them a little differently.
And an Impact vs. Effort Matrix compares an idea's potential value vs. the work involved.
Together, these frameworks put some practical criteria around the first cut. How many people could the idea reach? How much effort would it take? How confident is the team in the opportunity?
But what they don't tell you is how consumers might respond.
While the capacity for creating endless ideas is well…endless, research budgets, timelines and team capacity are finite. The next challenge is deciding where deeper research will add the most value. You don’t want to narrow the pool too quickly, and your timeline and budget can’t afford to get bogged down in the infinite possibilities.
Say an AI-assisted ideation exercise produces 100 possibilities for a new household cleaning product. The ideas include formats, ingredients, benefits and use cases. Testing every idea with human respondents wouldn’t make sense. But immediately choosing the team’s favorite five means 95 ideas disappear without any consumer reaction.
So how do you surface which ideas deserve the research investment, without losing good concepts along the way?
Before ideas reach consumers, teams often narrow ideas based on:
These criteria matter. There’s no point in pursuing an idea the business can’t deliver, or that doesn’t fit the strategy. If your operations team says a new frozen beverage concept requires new equipment across the brand, the idea may not be feasible.
However, relying on internal criteria has risks. Teams may gravitate to familiar ideas because they’ve already worked.
PepsiCo’s Director of Consumer Insights, Ryan Dirkmaat described this concern:
“Our marketing teams felt that our legacy tools required us to test too late in the innovation process when ideas had already been pared down. They thought we might be leaving great ideas on the cutting room floor.”
The first cut isn’t meant to eliminate human judgment. The question is whether the team can add useful evidence before cutting interesting ideas.
Effective product idea prioritization means using the right information at the right stage. Start broad enough to leave room for surprises and bring consumers into the process before finalizing the shortlist.
Start with a broad pool of meaningfully different ideas because some of the best ideas can come from unexpected places. Take the Post-It Note. 3M initially struggled to find a use for an adhesive that was weaker than conventional glue. But when the company put the unfamiliar product into consumers' hands, more than 90% of people who tried it said they would buy it.
That’s a useful reminder when building your initial product pool. An idea can be unfamiliar and still promising. If 30 ideas are essentially variations on the same concept, you haven’t cast a wide enough net.
For a snack brand, that could mean exploring different flavors, textures or occasions because 30 variations of barbecue chips aren’t meaningfully different.
A more varied pool gives you more opportunities to find ideas worth exploring.
Don’t create the shortlist without consumer input. When you bring in an early consumer signal alongside considerations such as strategic fit, feasibility and business potential, you can find more evidence on which ideas to move forward.
For instance, an idea stakeholders think will perform well might fall flat, while a more unconventional idea might give the team a reason to keep exploring. AI prediction models (Models trained on large databases of real human responses to ads, ideas or concepts) can help you introduce consumer input earlier before bringing a shortened list to test with human respondents.
At this stage, you’re looking for reasons to investigate an idea further. You’re not moving into production mode or betting on your next winner, just deciding if an idea is worth testing further with real consumers.
Screening early ideas with AI prediction models or synthetic respondents (AI-generated personas designed to simulate how specific demographic, attitudinal or behavioral profiles might respond) can offer insight based on real consumer behaviors to help you identify the ideas with the greatest potential that are worth moving forward to human testing.
Once you have narrowed your concepts into a reasonable amount of promising ideas, it’s time to move them into concept screening and product concept testing with real human respondents. This step is critical for deeper understanding and potential development and validation.
Now, you can explore how real consumers respond to your shortlisted concepts and how to strengthen the ideas further. This is where you can invest more time and research budget to bring your concept to the next stage with confidence.
So what does synthetic or AI idea screening actually look like?
Zappi’s Synthetic Idea Pre-Screener helps teams quickly sort through a large pool of early ideas and identify which ones consumers may find most appealing before investing in human research.
This tool uses AI-generated representations of consumers designed to simulate how consumers may respond to research questions. Zappi’s synthetic respondents are grounded in demographic and behavioral data.
Their job here is specific: help teams make an initial cut across a large pool of ideas. They aren’t intended to replace human respondents in deeper innovation testing.
If a team has 60 meaningfully different ideas for a new product line, they can put the entire pool through the Pre-Screener rather than choosing which ideas to test first without any input.
Teams can screen 30-300 ideas at once using text only or text with images. Ideas can come directly from Zappi’s Concept Creation Agents or teams can upload ideas developed through their own ideation process.
Once teams upload ideas, they select their market and category, and Zappi’s synthetic respondents evaluate the pool. Results are available in around 15 minutes for up to 100 ideas and within 45 minutes for up to 300.
The results give teams several ways to compare ideas:
From there, teams have more evidence to decide which ideas deserve deeper research with real consumers.
To learn more about Zappi’s Synthetic Idea Pre-Screener and how it can help you get more out of your early ideas, reach out to us.