Up and Running, er Crawling

With our funding secured and solid interest from the media, things looked pretty good.  The one less-than-good item was sales.  We needed to establish a relationship between marketing and sales where, if we spent X dollars on marketing, we’d be able to count on Y dollars in sales—fundamental business stuff.  But unfortunately, marketing and sales seemed oddly estranged.  

So we experimented.  We tried different ads.  We put ads on different platforms.  We engaged influencers and got them to write about Tertill.  We attended gardening shows and other events where we demonstrated the robot.  We ran sales and holiday specials.   But disconcertingly, our sincere marketing efforts had negligible effect on sales. 

The price of the robot started at $350.  This was about the price we needed if we were to earn a profit on the first generation of robots.  But, we thought, it might be too high for our market.  So, we tried lower prices.  Losing money temporarily would be OK we reasoned, because later we would reduce our costs.  We’d do that by redesigning the robot and increasing the size of our factory order.  We also conducted surveys trying to discover the price customers would expect to pay for a product like Tertill.  It turned out that the number was about $200.  So we tried selling at that price.  But sales didn’t respond.  

The only thing that had any appreciable effect on sales was national exposure.  When Tertill appeared on Ask This Old House and on a few other high-profile venues, sales lit up for a while.  Then they retreated back to their too-low numbers.  

So we sought to inspire stories about Tertill on shows and in publications with a wide reach.  The problem was that after a news outlet has done one story on your product, it’s very hard to get them to do a second unless there’s something significantly new to report.  But with sales at an unsustainably low level we couldn’t generate the cash we needed to develop and release a new robot model with new features. 

Tertill-Related Robots

To be healthy, a company typically needs more than a single product or “SKU.”   So, we explored many other products we might develop.  If an interesting prototype captured a reporter’s attention, we’d get a story about our coming-soon product—that might also boost Tertill sales.  Or the idea might attract new money from an investor.    

Tertilnator (See Tertilnator) was one such attempt.  Tertilnator was a robot designed to chase bunnies, ground hogs, chipmunks, and other produce-hungry critters away from crop plants.  Over several months we built prototypes and I tested them in my garden.  I thought the robot showed promise, but my teammates were not convinced that it would be a slam dunk.  

Tertill customers sometimes requested new products or changes to the current model that they’d like to see.  A popular desire was for a robot that would remove weeds from patios.  When patios are constructed by laying bricks or pavers on the ground, weeds may sprout in the joints.  Some customers spontaneously used their robots for patio patrol.  But this tended to wear out Tertill’s designed-for-dirt wheels.  So we strongly considered building  a version of Tertill specialized for this purpose.

Leaves and other debris often settle on backyard decks.  So another idea we had was to replace Tertill’s weed whacker with a small blower.  Renamed Clear Deck, this robot would spend its days roaming the owner’s deck blowing debris over the edge.  Customers occasionally complained that Tertill wasn’t systematic—like the original Roomba it just bounced around.  Adding an accurate localization system would correct that problem but, we felt, would make the robot too expensive.  So, we found another way.  If gardeners or, more likely, small farmers planted their crops in rows connected in the right way, we could program the robot to only hug the crops, following rows all day long.  The increased efficiency would mean a single robot, rather than handling only a 150 square-foot garden, could instead tend 750 square feet or more.  

Some city dwellers have awkward-sized lawns—too big to ignore but too small justify buying an expensive power mower.  We thought a version of Tertill might fill the bill.  So, we mocked up a prototype where X-Acto blades mounted on pivots replaced the string trimmer.   (Had we decided to develop the concept, we would have engineered a safer cutting mechanism!)  I tested the idea in my lawn, and it did indeed cut the grass. 

A common issue Tertill encountered was that many gardeners plant their crops much more densely than instructions on the seed packet recommend.  When plants are too close together Tertill can’t maneuver properly.  Although Tertill was already challengingly small, we spent considerable time and effort building an even smaller version, unofficially called TertillTeeny.  It never managed to match the maneuverability of its larger sibling.

Other Product Concepts

Tertill included a smartphone app.  The app let users know the status of, and statistics about their robot—e.g. how much time it spent patrolling and charging.  We researched greatly expanding the app to include information about climate and soil conditions in the user’s region.  This would help gardeners monitor and improve the health of their gardens.  For this we would charge a subscription.  But it was a big project and we never fully developed that idea.

We investigated developing non-robot products related to gardening.  One was the Tomatometer (Tōm.e.tom’.e.ter) or tomato meter.  The idea was to build a device able to perform a multispectral analysis on tomato plants.  (We chose tomatoes because tomatoes are grown in 85% of home gardens.)  By measuring the reflection and absorption of different wavelengths of light our instrument could determine the health of the plant.  It would then advise the gardener if the plant needed more or less sunlight, water, or nutrients.  Over the summer I planted eight tomato plants and systematically gave pairs of them too little, too much, or the right amount of sunlight and water.  I carefully monitored the progress of the plants, recording the measurements made by the Tomatometer.

Sadly, my meter did not show dramatic differences between healthy and less-than-healthy plants.  Scientific literature suggests that a product like the Tomatometer should be workable and I would have liked to refine my construction and measurement techniques.  But there was no time as that would have required another growing season.

Our focus was on gardens, and we sought any angle that would make gardening easier and more productive. One such idea was fertilizer.  

Around this time a company called Sunday (www.getsunday.com) was making a big splash.  They provided custom lawn fertilizer designed for their user’s climate and soil conditions.  Their product was supplied as a liquid in a plastic bag.   By connecting a hose, fertilizing became as easy as watering.  Why couldn’t we offer a similar convenience for gardens we wondered? 

We began a pivot to fertilizer.  It was a lot of work.  One hard part is that each state had its own standards and required certificates specifying the sort of fertilizer can be sold in that state.  We jumped through 50 hoops.  But finally we were ready to offer our easy-to-apply, tuned-for-your-garden product.  We launched the product, we promoted it vigorously, and the market yawned.  Our fertilizer product was no more successful than Tertill.  

We had pretty much tried everything we could think of to make Tertill appealing, to reach gardeners who might be interested, and to find other products that would appeal to the gardening market.  All to no avail.  After eight years of diligent effort we reluctantly accepted the verdict of the marketplace.  

Tertill Necropsy

We began with the idea of building a product that would solve the age-old problem of weeds in gardens.  We sought an effective solution that would be low in cost, effortless to use, and would require no herbicides.  The product we built was all of those things.  Tertill lived in the garden, it prevented weeds from growing and could often be found for $200—the same price as the original model of Roomba.  

The maddening aspect of the exercise was that Tertill worked and it worked well.  Our little robot reliably kept my garden free of weeds, many customers reported the same and sang the praises of our machine, and Tertill earned a place in somewhere between 5000 and 7000 gardens.  Researchers at the agricultural departments of the University of Manie at Orono and Cornell University conducted tests that confirmed Tertill’s efficacy.  But it wasn’t enough.  Nearly out of money but hoping to endure until the market caught up with our robot, we merged with my previous company, Harvest Automation, with whom we shared offices.  But the market did not catch up; we were forced to discontinued Tertill.

Why did Tertill meet such a cruel fate?  Sadly, one rarely gets a satisfying answer to such questions.  Afterall, had we known what was going wrong we’d have fixed it.  What we are left with are a few possibilities.  

When a technology works but customers don’t adopt it, the problem is usually held to be one of “product market fit.”   We knew of some potential misalignments between the things that might interest customers and what Tertill provided.

As we began development, I computed that Tertill should have a market maybe one tenth as large as Roomba’s.  There are about 120 million households in the US.  All of them have floors but only about 42 million have gardens.  Gardens seem much more varied than living room floors.  There are gardens with multiple, small, raised beds, square-foot gardens composed of many small, separated spaces, and deck gardens where all plants are grown in containers.  Tertill was not ideal for any of those venues. Allowing for such gardens, I figured maybe 12 million or so gardens would be compatible with Tertill.  

iRobot sold about 50 million Roombas over 20 years; selling 5 million Tertills over the same period would be fine, I thought.  But we didn’t come close.   Tertill’s incompatibility with several popular garden configurations worked against it.  So, maybe my estimate of the number of gardens where Tertill could flourish was too optimistic, and there just too few places where Tertill worked well.

Another fundamental truth of our market is that gardening is primarily a hobby.  Commercial farmers grow crops with great efficiency, home gardeners much less so.  When the cost of the gardener’s time is included in the equation, produce bought at the supermarket is generally less expensive than crops grown at home.  

We promoted Tertill, in part, as a tool to improve yield by eliminating the competitive pressure of weeds.  But even the weediest of gardens still produce some crops (I can attest).  It may be that most gardeners see weeds as a nice-to-not-have rather than a scourge that must be defeated at all costs.  Weeds didn’t prevent gardeners from enjoying their pastime—that lessened the urgency for buying a tool that would eliminate them.

Finally, cost may have been a problem.  Even when they could buy the robot at the $200 price consumers said they expected, it wasn’t fully in line with other garden expenses.  Gardeners spend an average of less than $100 annually on their gardens.  So spending twice that amount for a single implement may have felt excessive even if consumers expected our robot to cost that much.  

It’s conventional wisdom that only about one in ten startups comes close to the success its founders hope for.  That puts Tertill in good, if melancholy, company.  Developing that little robot and bringing it to market was great fun.  Seeing it actually work and deliver the performance we imagined, even better.  I have no regrets.  The greatest fun would have been getting millions of customers to agree with us.  Maybe next time.