In the first project, 302 consumers ranked ten cleaning concepts. Six went forward, and one technology is now with R&D partners.
The concept card shows a man in a VR headset with a spray bottle in each hand, chasing cartoon bacteria across a kitchen counter for points. Reckitt's front-end team designed the concept, called CleanQuest, for a frustration that 39% of surveyed consumers named among their top cleaning complaints: the chore feels too big to start. Even before testing, people on the team were asking whether it went too far.
In many companies, a doubt like that gets settled in a meeting room, by gut feeling or by whoever created the idea. The same forces decide the fate of front-end ideas once they reach the brands: an opinion ends some, while a sponsor who leaves or changes roles ends others. Underneath these decisions sits one weakness. Each idea rests on a guess about what consumers need, so anyone in the business can question it.
Gianluca Balzamo, a senior scientist on Reckitt's front-end team, built a process that replaces the guess with consumer evidence. Working with the business prototyping company NOBA Ventures, his team tested CleanQuest in AI-moderated interviews with 302 consumers in the US and Europe. Consumers ranked it last of ten, which settled the debate with data instead of opinion.
The same process took the brief effortless cleaning from 85 pain points to one technology platform, now briefed to R&D and on its way into Reckitt's product categories. AI made the approach practical for a small team, since AI-moderated interviews and AI ideation replaced days of workshops. The result is a front end that hands the business ideas consumers have already ranked, along with Gianluca's plan to get the business pulling for them sooner.
Where Front-End Ideas Get Lost on the Way to the Brands
Gianluca trained as a materials scientist and engineer, so consumer research sat a long way from his day job. He works in Science Platform, a Reckitt department that develops technologies five to ten years ahead of the market. Those technologies create value only once a brand picks them up and sells them, which puts the handover to the brands at the center of his work.
That handover is where his frustration started. The core business focuses on delivering fast, controlling quality and selling at volume, while Gianluca's team works on problems years further out. So when a front-end technology reaches a brand, the brand team has to decide whether it is worth the time and money. Gianluca has watched brands turn technologies down at that point, in his own projects as well as in other front-end teams.
In Gianluca's view, the deeper reason is that consumer evidence is missing. A front-end team can explain what a technology does, yet it often has only a guess about whether consumers need it. That guess hands the brand team an easy objection, because the front-end team has little proof to answer it with.
An eye-catching technology can cover the guess for a while: it can win a strong presentation, broad buy-in, even a mention from the CTO or CEO at a company forum. Later, though, the operational team usually ends the project, in Gianluca's experience, because the excitement came from the technology rather than from a consumer need. Marketing reacts the same way when a team pitches a technology before explaining why consumers would want it.
Inside the front-end team, someone newly asked to innovate often struggles to know where to start, because the process is unclear. Company challenges also tend to arrive too broad for R&D to act on. To close the gap from both sides, Gianluca built a process with two goals: give a front-end team a clear sequence to follow, then attach consumer evidence to every idea before it reaches the business. The process builds on design thinking, with AI tools added at each step, across three phases:
Discover: find the frustrations worth solving.
Define: generate the ideas that could solve them, then narrow them down.
Validate: test whether people will buy them.
AI speeds up each phase, while people make every call to continue or kill. The first project to test the process started with a brief so broad that it could mean almost anything.
Narrowing "Effortless Cleaning" Before It Reached R&D
At Reckitt, the front end's challenges come from the top: senior stakeholders agree on broad goals the company wants its future products to deliver. For the first project to run through Gianluca's process, they had agreed on one such goal, effortless cleaning. An SVP backed the project, which gave the team the resources to run it. The two words set the goal while leaving everything else open: which brand would sell the result, what kind of product it would be, and which technology it would use.
"Effortless cleaning is very subjective," Gianluca says. "What's an effort?"
His own answer changed within a year. After his son was born, his back pain got worse, so chores he once considered easy started to feel like hard work. If one person's idea of effort can shift that much in a year, a two-word brief leaves R&D guessing which problem to solve.
To narrow a brief like this, R&D needs consumer data. In many large companies, though, R&D sits apart from that data, with little time or access to study it. So Gianluca's front-end team did the narrowing first. It sorted through what consumers say about cleaning, found the frustrations that matter most, then turned them into clear problems R&D could build products for. Gianluca describes this step as turning consumer noise into R&D action.
Starting with the consumers who struggle most
The team started with the people who find cleaning hardest, such as people with arthritis, ADHD or a newborn at home. The team calls them extreme users. They feel the same frustrations as everyone else, only more strongly, so they post about them online and come up with their own workarounds. Most people put up with a milder version of the same frustration and seldom mention it. Studying extreme users first therefore showed the team problems that a product could fix for a much wider group.
The team mined what these extreme users, along with mainstream consumers, say and show about cleaning on TikTok, Instagram, YouTube, Reddit, Facebook and WikiHow. It grouped the extreme users into seven contexts, including these three:
Physical and mobility limits: arthritis, chronic pain, back injuries, pregnancy, older users and limited grip strength. Gianluca expects this group to grow as people live longer.
Neurological and cognitive load: ADHD, executive dysfunction, time-blindness, anxiety, autism-related sensory aversion and memory issues.
Caregiving and family: newborns, NICU patients, elderly parents, immunosuppressed relatives and allergy-prone children. Gianluca recently joined this group himself. He now catches himself watching the mist from a trigger spray, wondering whether it reaches his son's porridge two meters away.
A user with ADHD posted that starting chores felt overwhelming. The dishes weren’t the problem; getting up from the sofa to do them was. So he told himself he was giving the dishes a bubble bath, which made the chore feel like a game. Gianluca saw two useful things in that post. The real problem was procrastination rather than the cleaning itself. The user had also already found his own fix.
Posts like this one added up to 85 pain points and hacks, which the team clustered into 11 friction spaces. One space, which the team called the mental side of cleaning, covers feeling overwhelmed and struggling to start. The posts only showed what each frustration looked like. To learn how many consumers share each one, the team moved on to interviews.
Putting a number on every frustration
The team interviewed 300 consumers: 150 in the US and 150 across the UK, Germany and France. An AI moderator on the platform Outset.ai ran the interviews. Each consumer chose which of the 11 friction spaces were among their biggest cleaning frustrations. Hard-to-reach spots came first, picked by 48%. Cleaning that takes too long or has to be redone came second, at 45%. The mental side of cleaning came third, at 39%.
In the interview videos, participants explained what the mental side feels like. A long list of chores felt overwhelming before they started. Once they began, cleaning felt easy, even fun. For these consumers, the start was harder than the cleaning itself. Gianluca sees the high ranking of this frustration as a sign of how much has changed, since for a long time people mostly counted something as a pain only when it hurt physically. Physical strain still matters, though: one young participant in the same interviews mentioned back pain.
The ranking showed how many people share each frustration. To find the most urgent ones, the team also needed to know how often people feel each pain and how bad it is. So it broke the friction spaces into more specific pain points, then scored each one with this formula:
Measure | What it captures |
Reach | How many consumers feel the pain |
Impact | How often they feel it × how severe it is |
Opportunity | Reach × Impact |
Critical (Tier 1) | An opportunity score more than one standard deviation above the average, meaning the pain clearly outscores the typical pain point |
From pain points to jobs to be done
Nine pain points scored as critical. The team turned them into a list of jobs to be done: short statements of what consumers want to get done, such as “clean gaps without physical strain.” Each job names the goal but leaves the product open, so the team could consider any kind of solution. Every idea that came later then faced the same simple test: whether it got the job done. In a workshop, Reckitt stakeholders picked the five jobs that best fit the company’s strategy:
Clean gaps without physical strain
Get surfaces guest-ready between deep cleans
Break the mental weight of starting
Remove mineral buildup without DIY hacks
Get a 100% clean without the "finishing force"
With five jobs backed by consumer data, the team had a clear target for new ideas. The next step was to fill that target with as many ideas as possible, as fast as possible.
Letting AI Write the First 156 Ideas
In Gianluca's experience, the first hours of an ideation workshop go to the most obvious ideas.
"Often people come into a workshop already with some idea in mind," he says.
To get past that bias, the team had AI write the first list, so a broad set of ideas was on the table before anyone's favorite could steer the discussion.
The team gave the AI two things to work from. The first was a set of checks based on what Reckitt can make and sell: its R&D limits, strategy, business case, sustainability goals and technical feasibility. These checks kept the AI’s ideas realistic. The second was the list of DIY tricks consumers had shared online. Gianluca’s thinking: if people have already built their own fix, Reckitt might only need to sell a ready-made version of it.
Using both inputs, the AI produced 156 ideas for the five jobs. The jobs with the most potential got the most ideas. People took over from there. Reckitt experts brought 22 of the ideas into a workshop, deciding for each one whether to proceed, pivot or kill. They filtered on fit with the job, clarity, plausibility and uniqueness, combining some ideas while reworking others.
The more rounds an idea went through, the more people changed it. Some concepts ended up looking very different from what the AI first wrote. AI then turned the shortlist into visual concept cards for consumers to judge in AI-moderated interviews. The define phase ended with ten concepts for 302 consumers in the US and Europe, including one concept the team itself already doubted.
CleanQuest: The Concept Consumers Voted Down
In the interview videos, some consumers said playfulness made chores feel lighter. So for job 3, the team took that idea to its most extreme form: turn cleaning into a game. CleanQuest combined home sensors with a mobile app plus an AR overlay that turned chores into quests, with points and rewards for progress. The bet was that if cleaning felt like play, people would start sooner and keep going.
CleanQuest was the most extreme of the ten concepts. The team included it on purpose. People on the project were already asking whether it went too far, yet the team let consumers settle the question. At this stage, Gianluca says, "We are defining the solution space." The test had two jobs: to show whether gamification was worth pursuing at all, then to show whether the process could kill a weak idea quickly and cheaply.
The 302 consumers gave a clear answer:
CleanQuest scored 2.3 out of 5 for desirability, the lowest of the ten concepts (2.3 in the US, 2.4 in Europe). The top concept scored 4.0.
63% of consumers rated it 1 or 2 out of 5.
25% said they would consider buying it, against 73% for the top concept.
One participant explained the reaction: "Gamification of tasks like this [is] interesting, particularly if you want to get your kids to help, but it also feels extremely dystopian. So it's creepy."
Those results led the team to retire CleanQuest as a standalone product, with consumer value, viability and feasibility all rated red. Gianluca stresses that consumers made this call, rather than the team. He counts the result as a miss in execution, yet a success for the process. "Killing fast, it's also a KPI of a front-end innovation team that has got limited resources," he says. A team with a small budget has to explore many options, so it can only afford to carry the strongest ones forward.
Even as a retired concept, CleanQuest taught the team three things:
Consumers filter gimmicks instantly. Adults judged the concept on one question: does this make my cleaning better, or is it only novelty?
Rejection can become a brand risk. Consumers used "creepy" and "dystopian" unprompted, which signals more than low interest. Words like those next to a Reckitt product would put brand equity at risk.
The need survives the idea. People still struggle to start cleaning, so job 3 stays open. The next idea could motivate people with a satisfying, visibly finished clean instead of a game.
Every concept faced the same gate, which the team called "guilty until proven innocent": each had to show evidence of consumer value, viability and feasibility before it received more investment. Six of the ten passed. The other four, CleanQuest among them, went back to discovery or ideation, carrying what consumers had taught the team about cleaning behavior and gamification. Those lessons went beyond any single concept, because consumers judged all ten by the same rules.
Four Rules Consumers Applied to Every Concept
In the interviews, consumers liked or rejected each of the ten concepts for the same four reasons:
Remove a step. Consumers got much more excited about a product that does away with a step than one that only makes a step easier. CleanQuest did the opposite, because its AR layer added steps.
Show when it's done. Consumers dislike guessing when a product has finished working. A clear signal, such as a color change, makes them trust it.
Keep it to one step. People clean often and want it over fast, so every extra step or choice puts them off.
Prove it works. Consumers have heard big promises before, so they doubt new products. The biggest reason they held back from buying was a lack of proof that the product works.
Six concepts passed the gate. The next phase tested whether people would actually pay for them.
Asking for a Credit Card Before Asking R&D for a Budget
Consumers who like a concept in an interview may still leave it on the shelf. So the validate phase set out to learn whether people would actually buy, before R&D spent money building anything. The team also had to plan around one open question: which Reckitt brand would eventually sell each idea. Validation ran in four steps:
Platformize. Because the brand was still open, the team looked at the technology underneath each concept, such as an indicator that shows when cleaning is done. A technology like that can go into several different products, so the team could adjust each opportunity along the way instead of killing it. Reckitt could then choose the final product once a brand came on board.
Smoke test. Live ads and landing pages presented the product as ready to buy, which measured real purchase intent along with the minimum set of features consumers expected. In Gianluca's words, "a swipe of a credit card is one of the strongest signals of purchase intent, compared to showing an image and asking, 'Would you buy it?'"
Test features one by one. The team broke each platform into individual features, then tested each one on its own, including hands-on tests with physical prototypes in consumers' own cleaning settings. These tests separated what a first version truly needs from over-engineering that only adds cost.
Hand over. Once a brand endorsed the concept, the team sent a well-defined technical brief to external R&D partners.
As a result, R&D received a brief based on what consumers showed they would buy: which features they will pay for, and which ones they can live without. The four steps also kept mistakes cheap. If a concept failed its smoke test, Reckitt lost the cost of the ads rather than the cost of developing a product.
The Results: From Two Words to a Technology Platform
Seven months after senior stakeholders agreed on the brief, effortless cleaning had become a technology platform with brand endorsement and a technical brief at external R&D partners. Consumer evidence narrowed the work at every step:
85 pain points from mainstream and extreme users
11 friction spaces, sized with 300 consumers
5 jobs to be done
156 AI-generated ideas
10 concepts tested with 302 consumers
6 concepts through the gate
1 technology platform briefed to R&D
Part of those seven months went to waiting while the team worked out each next step for the first time. With the steps now set, Gianluca thinks a dedicated team could run the process in three to four months. Other Reckitt teams are learning a simplified version in a short workshop, and the consumer findings have gone to the insights, analytics and marketing teams for cleaning products.
What Gianluca would change next time
“Getting people inside the company on board is still the biggest challenge in front-end innovation”, Gianluca says. This project had an SVP behind it from day one, but the category teams that would one day sell the product joined only after consumers had reacted well to the concepts. By then, the front-end team still had to convince them.
Next time, Gianluca would run an ambassador program alongside the project. Someone from each category that might sell the product would follow the work from the start and see the consumer evidence as it arrives. That person, his “mole in the system,” would then pull the idea into the category, so the front-end team would spend less time pushing it.
He would also look for a shorter route to market. Taking a product through the core business usually adds about three years. A lean team that puts small pilots on the market could get real consumer feedback sooner and spend less time on ideas that aren’t promising.
Gianluca sums up the approach in one line: “Let consumers do the killing. Of course, we kill too, but we let consumers do the macro work.”
His team cut 156 ideas to ten on its own. Consumers then decided which ideas got more money, which features were worth building and which went back for more work. An idea that reaches the brands with that verdict attached is much harder to dismiss as a guess.

