Disposable Head Toilet Brush

This project is a disposable head toilet brush, and RB’s attempt to enter and grow a nascent market segment in Europe and North America. We set out to design the best version of this product in the category with our “5th mover advantage.” I led the project from conception through a 200-person consumer test.

The segment is well established and worth $80 million in the US, but is tiny in Europe. For each cleaning a user clicks on a new sponge head, which is impregnated with cleaning formula, and proceeds to wipe down the inside of the toilet. After use, the head is ejected. The systems are sold as starter kits with a head-holding caddy and a wand plus some heads, and a separate SKU for the head refills. Users tend to be more germophobic than average, and love that there’s no germ covered toilet brush to store. The product’s sponge type genuinely just cleans more per swipe than a normal brush too. Though it seems like every major player but RB has launched a product in the space the vast majority of market share belongs to Clorox so their version is the one to beat. We wanted to beat them on any given aspect possible. 

From the start we knew that the biggest challenge would be that RB was late to the party and there was a large IP minefield. So we started there. There wasn’t enough internal resource at the time so we decided to outsource early mechanism design to a company called Cambridge Consultants. I wrote the specifications and the brief for the consultancy. We worked together on the process and agreed to a statement of work which I approved. The two companies collaborated closely. We went to their site for brainstorms and key milestones when ideas were filtered out. The agency prototyped and refined the mechanisms in between these check ins.

Parallel to Cambridge Consultants working on the product, RB internally conducted a thorough patent landscape exercise. The one thing that was obviously in common with all the other product’s connection mechanisms was that the male side was usually on the head, and female side on the wand. So we focused on differentiating this as a seed for our brainstorms.

Out of the first brainstorm’s many ideas two emerged as our favorites after we filtered based on cost, mechanical intuition, ease-of-use and IP clearance. One we called Harpoon Plus is involved in barbed end connecting to a female side on the sponge head. This one has the potential to be made via stamping instead of injection molding to save money. It also circumvents all of the patents by having the female side on the sponge head instead of Vice Versa as all the competitors do.

The next concept was based on the idea of a reusable wine bottle cork. The cylindrical silicone material on the wand tip is squished between the head and the wand and creates a wider interference fit on the inside of a shaft. The mechanism did need to be a bit different because we want the head to click on just by pushing the wand into it, whereas the wine cork product uses a lever/cam to compress the silicone part. 

Parallel to IP search, and the mechanism design, we also ran a series of consumer tests. Consumers each took home one of the off-the-shelf products, and then were interviewed in focus groups and surveyed. The output was a theoretical ideal toilet brush, and an understanding of attribute priorities. 

The first Harpoon Plus prototypes were made with laser-cut samples. They iterated many times on the thickness length and overall size of the mechanism. The samples improved drastically and were functional but we judged not strong enough to ensure complete confidence. Even the perception of a loose head was enough to turn people off to the idea. Nobody wants to risk a lost sponge in the toilet. We discovered there is a certain way that you can torque the heads in a toilet bowl against the rim while scrubbing very hard as a consumer would. This can twist off the mechanism much more easily than we had initially expected. The planar mechanism well functional and simple was not strong enough ultimately.

The wine bottle Concepts was also progressed to a fully functional prototype with a placeholder wand design. The downside of this design was that the heads are much taller and therefore the overall size of the packaging and caddy would need to be much greater. In addition the torque off problem and the lower friction properties caused by the lubricated soapy water meant this version was not secure either.

The budget had to run out at this point for consultancy work and the mechanism had not progressed as far as we would have liked. At the about same time RB had acquired a new Carbon M1 3D printer which vastly increased our capabilities and the materials we could use. It would allow us to quickly iterate with a polypropylene-like material. I took over personally the design and iterations of the mechanisms from this point.


The key trade off on the harpoon plus mechanism that was holding it back was this: a conical barb inserted into a planar female clip mechanism cannot have a much stronger tear 
off force than on force. These two forces are related to the deflection of of the teeth. With the teeth on a plane, the deflection to get onto and off the barb was always the same, weather the teeth bend up or down doesn’t matter. With a planar mechanism it would never be easy to put the head on, and also to difficult to tear it off unintentionally.

In the course of the next two weeks I learned how to use our new 3D printer and iterated about 20 more versions. With the teeth now bending down it meant that the teeth did not have to bend very much to be attached (so it was easy to attach), but to fail it would now fail via buckling instead of bending. It was also discovered that distortion along the main plane of the disc was actually significant so a large rib was added to isolate the clipping teeth from the cleaning forces.

I spent a lot of time tweaking these numbers. This is the tooth profile shown against the barb of the wand.

A more close to final version of the mechanism profile.

I was also able to improve the wine bottle solution. The crux of the concept was that the mechanism relied on friction of the interference fit to hold on. But in a toilet with soapy water there’s not much friction to be had. The next iterations of the concept had a flange which constrained the silicone part when it was compressed. For various reasons the team preferred the Harpoon plus concept but were not 100% sure that the consumers would prefer it as well.

Around the same time we also kicked off industrial design of the wand and caddy. RB partners frequently with a design agency called VanBerlo. They worked on the wand design in parallel to me refining the mechanism so that we could run a consumer test with 3D printed prototypes in the minimum time.

We mostly left the outer wand design work to VB. The internal mechanism however I knew would be challenge. Inside the wand there needs to be a way to transfer a force from hand to the sponge head’s disk to eject it. We began by examining the competition.

Clorox’s product has a similar mechanism that uses a thumb push to extend a rod and release the head. It uses four parts: push rod, collet, spring, gasket. Our consumer test showed that people prefer a trigger based mechanism because they were worried about accidentally ejecting the head in toilet. When you’re pushing on the wand to clean you can never accidentally pull a trigger. This also presents a mechanism challenge: how can we turn a pulling action into a pushing action down the wand?

VB proposed a rack and pinion mechanism and also a linkage mechanism. Both of these would return with a spring. The spring would be in the back of the wand so that I wouldn’t rust. A metal spring in water with acid formula would quickly rust. Clorox solves this by making the wand watertight with a gasket that seals around the shaft of the head.

I felt these mechanisms would be hard to assemble and an unnecessary number of parts. I proposed a one piece trigger shaft spring mechanism. I was sure we could crack it with enough iterations, but I knew from experience it might take a lot of iterations to refine it. so I built a jig so we could quickly test new versions without printing new parts. 

The jig has movable retaining studs so that we could easily adjust the position and pre-load of the part, and an adjustable fulcrum point. The part could also be 3D printed with slightly different thickness curvature Etc. the test jig quickly demonstrated what I suspected: allowing the mechanism to bend but not buckle on such a long shaft was key. Once the shaft had buckled it could not push any harder on the disk, regardless of the trigger’s travel. However if the shaft was 100% constrained it obviously would be completely stuck,or just very hard to pull. We were able to slowly add studs to reduce the length that could bend such that the trigger would bend properly. We integrated this learning into CAD.

The wand designs from VB eventually arrived at a few versions pictured below which were consumer tested. 

In parallel to all of this development we also were sourcing sponge-head suppliers. Our procurement groups took on the challenge but returned with few positive results. The R&D team took on the challenge themselves, cold calling and searching for major sponge manufacturers. We solicited samples from series of sponge manufacturers and we narrowed the pool based on their confidence in being able to produce the samples and their quality of samples. We knew eventually we would want to run a consumer test in homes with a functional product based on pilot tool samples before we would fully commit to spending the investment to create the heads. However we also knew that no supplier would help us produce the samples without some level of payment or commitment. So we moved to establish a joint development agreement with one supplier based in Europe. We agreed to pay for sampling fees and Engineering fees of pilot equipment. Basically we were using them as a sponge head consultancy. The supplier was helpful in prototyping and selecting the exact sponge size and creating pilot Machinery to impregnate the formula into the heads. They also help scope the overall cost of the project for financial vetting. 

The strategy on the head design was to somehow be differentiated from Clorox’s hexagon shape, while still tessellating, and allowing the heads to better clean into the nooks and crannies of a toilet. We set tessellating heads as a requirement because we knew manufacturing these heads would happen by a stamping operation and if there were any scrap pieces a hand separation procedure would be required which would increase the cost significantly. These are the options that we put on the table after a few rounds of sketching and cutting sponges by hand.

At this stage all three streams of work came together and we felt it would be appropriate to run a central location test in our offices help choose the best direction on a few attributes: Wand shape, which of the two mechanisms, and Head shape. I set out to produce functional 3D printed samples with enough disposable sponges that consumers could test various variations of the product with different heads different mechanisms at our offices and then discuss the pros and cons in a focus group afterwards. 

As an aside this level of consumer interaction with prototypes is very unusual and was only made possible buy a high level of prototyping and our new high resolution 3D printer. Myself and our summer intern produced all the samples, cutting sponges, 3D printing polishing and assembling the mechanisms. This was a major effort over the course of a month refining the fabrication process and polishing samples to be looks like real product level.

Consumers agreed with the team that the Harpoon Plus mechanism was superior to the wine bottle mechanism in its robustness and feeling of confidence. The click on feedback was also more noticeable and it did not have a spongy feel to the connection. The wine cork design had and inherent type of ball and socket flexibility which we thought consumers may prefer but in the end they did not.

Consumers were quite opinionated on head design too. Although each consumer had their favorite shape, ultimately the most well-liked on average was the circular head. We included a circular head as a control but we knew that a circular head would be too expensive to make. So after the test we decided to change to a curved hexagonal head. This new design was functionally not much different than a circle but does have a more powerful look, and has some small corners for the nooks and crannies. It was also novel and did not look exactly like the competition which would be important for a design patent. 

The wand was adjusted based on consumer’s favorite, and their feedback that the wand was too long and too bulky. We shrank it a bit as seen below.

At this stage we would normally run another consumer test at a prototype level to confirm that the design we chose was the best. However the company decided to take a risk and move towards pilot tooling based on this version. We kicked off pilot tooling with our supplier based in China.

VB  also produced this caddy design which holds six heads and holds the wand upright next to the toilet. Although this element of the product is not hugely novel compared to the competition it is superior in a few ways. It is much smaller and less plastic. It is aesthetically Superior, I think, having a premium matte clear style and a patterned texture which calls back to the head’s shape. We designed it so that the consumer could see how many heads they had left as a reminder to buy more but also as an unobtrusive, beautiful, neutral element in the bathroom.

There’s one notable feature that I specified for.  That the diameter of the caddy was exactly wide enough to hold the heads plus the wrapper that it comes in. I envisioned that the consumers would unwrap the heads only at the top similar to a bag of chips and drop the whole sleeve of heads into the caddy. This way the consumer never needs to touch the heads, which do have some blue, acidic formula on them. The perfect diameter of the caddy holds the sleeve just the right amount such that only one head would come out at a time.  When pulling the last head out the sleeve would come with it so it can be easily grabbed and thrown in the trash.

A few months later when pilot tooling was ready to be sampled I visited the Chinese factory to witness and approve of the samples.

The first samples were functional but we made the tough call and judged the discs were not quite strong enough. They could be forced off with a strong scrubbing motion. We had deliberately undersized the teeth thickness slightly so that it would be a metal safe modification to increase the tear-off force. The tear-off force was on the borderline so it was a tough call but it was decided to go ahead and modify the pilot tool in a couple of iterations making the disc more rigid and making the teeth more robust. We ended up adding a very large rib to isolate the connection mechanism teeth from the forces on the rim of the disk due to scrubbing. This large rib actually violates normal injection molding best practice because it is a variable wall thickness section and did have some voids and sink marks because of it. However we judged on this disposable part that aesthetics were not as important as the functionality and that trust in the robustness, especially at this pilot stage was paramount. Here’s the pilot tool iterations that happened within a couple weeks while I was in China. 

We didn’t have a formal test method to approve the strength of the mechanisms because we didn’t have time to develop a custom jig. Ultimately we approved it using an informal test method: When we were  unable to intentionally remove the head by pressing very hard on it, inside a ceramic dish with toilet bowl cleaner and water inside.

Other aspects of the wand worked very well. the ejection mechanism work even better than the prototypes because of the polypropylene trigger shaft. We were worried that the wand’s barbed tip would wear out so had designed it as a separate POM part. But it did not show any wear even when molded in ABS and cycled hundreds of times. Overall the pilot tooling was very successful and only had normal, minor challenges.

The caddy came out beautifully when we mold it with ABS. The texture was noticeable but understated. It has a low key, premium elegance, especially placed side by side with Clorox’s.

The team and I also visited our European sponge supplier to review the overall heads manufacturing process, visit the Factory, see operations and discuss risks in more detail. The supplier designed a handmade process and designed a dosing machine that would make one head at a time. At this stage we refined the heads the heads material in various ways. The amount of dye, dosage of formula and glues used for all parts needed iterating. There was some times when the heads materials would delaminate from each other and then half of the sponge would fall off. It was actually the glue that binds the fibers of the non-woven together that was failing. We changed the design to have thinner layers to reduce the shearing on this glue. 

The product finally worked well and was at a like-final-product level. It was such a satisfying moment when it all comes together and works as we envisioned it.

The consumer test went ahead with 200 samples being produced. It was a small logistical feat to arrange samples made in China shipped to the EU sponge supplier and assembled under RB supervision with formula made in New Jersey. For the next two weeks, 200 French consumers, rated their liking of the written concept,  used the product in home, and then rated their usage experience based on various metrics. It is a thorough test run by Nielsen. Results came back six weeks later.

Overall consumer results showed that product was very effective and easy-to-use. There were almost zero complaints about the product’s functionality. It was considered very innovative and even worked better than expectations.

However, bad news emerged. The overall purchase intent was 60% of consumers rating the product as probably or definitely would buy. According to the testing agency this is low compared to other products in the category. With this level of results RB policy is to not proceed with the product. When digging into the results it’s obvious that the main dislike of consumers was the cost. The cost of the product is effectively four times that of liquid toilet bowl cleaners. This is true even in the successful US market but apparently this is a larger turn off for French consumers.

The project as-is was officially rejected at our strategy meeting by the business. Because cost was the major limiting factor we decided to adjust the design to squeeze cost out and quote on it one last time. 

I made changes to minimize impact on the consumer experience while maximizing the impact on the cost. A few changes were made: the area of the heads was reduced by 25% this correlates to a reduction in diameter of only 14%, not very noticeable by consumers. The design of the starter kit box was also reduced so that the product could be nested in transit more effectively. The overall size and weight of the wand and caddy were also reduced. The cost was cut by 25% total. Despite this improvement the business decided to move on, and the project is now over.

Looking back, I’m proud of the product that emerged. I genuinely believe it’s better than any other system on the market. Unfortunately, there are meaningful differences between European and American consumers that cannot be ignored but only arise in quantitative, in-use studies that take a long time to arrive at in the product development process.   In this giant post, I haven’t even mentioned any of the packaging design work, or the formulation work, or the myriad other tiny challenges we overcame. I wore many hats in this development: Design engineer, packaging engineer, project manager, formulation support, QA,  and even finance. So for me it was an amazing learning experience and I enjoyed the journey. 

We filed a utility patent for the system and the mechanism, and also a design patent. The GB patent application number is 1714687.9.  It is currently unpublished but the patents will be filed regardless of project cancellation. At the very least we have made the job of copying the idea one version more difficult.