Furrow - Why We Failed?

As many of you will know, I recently made the decision to shut down Furrow. I thought I’d start a Substack to document and share my learnings from the journey and keep my network up to date.

While there are a lot of reasons, both internal and external, that are laid out below, as the founder/CEO, the ultimate responsibility for Furrow’s failure rests on my shoulders. These factors didn’t happen to me, they were a result of my decisions and leadership. I learnt so much as a result of this journey and these lessons will travel with me into whatever I decide to do next.

The Furrow Story (skip to below for the reasons we failed)

At Furrow, our mission was to build out the technology, logistics and infrastructure to enable anyone to buy from any farm. I came up with the idea during the early days of the pandemic when the pandemonium in our food supply chains made me realise how broken and damaging the system was. I believed we no longer needed supermarkets to aggregate supply for us when the internet enabled the distribution of produce/harvest availability information direct from the farm. I also believed that advances in packaging and 3rd party logistics meant the distribution of the physical produce would also be possible direct from the farm. If we were able to find and build an operating model that supported physical distribution, we could move the food supply chain away from a pipeline model with multiple middlemen and the damage this was causing, and into a platform model with farmers selling direct to consumers. I felt that existing players were focused on digitising components of the existing supply chain, leading to small incremental improvements, and that if we wanted to unlock a step change in value, we would need to re-design the supply chain around digital technologies. (This article is an interesting read on the impact of electricity in factories as it first replaced steam motors in the same factory layout to unlock small improvements, and then resulted in factory redesign which unlocked a step change in performance and led to modern manufacturing)

We explored several different operating models at Furrow but ultimately couldn’t find a model that worked for consumers, farmers and us. Our first trial was partnering with DPD to collect from farmers nationwide and deliver overnight into our hub in London (initially my house). We then distributed the produce on a hyperlocal basis to maximise freshness and quality and reduce costs, rather than sending it back out overnight which would also entail high packaging costs. By delivering locally, we could deliver in ambient paper bags, rather than having to maintain the cold chain overnight and package produce in boxes that could survive DPD’s network.

Ultimately we realised this model didn’t work for three reasons. Firstly, cost. We had a warehouse in a central London location that we were haemorrhaging money on. Secondly, we were working against DPD’s natural way of working. They are used to collecting from a single location and delivering nationwide, not the other way around. This meant deliveries were frequently lost, delayed or mis-collected, each time adding to our costs as we had to refund customers or repeat our delivery route the next day with the delayed items. Finally, and probably most importantly, we were a demand-constrained marketplace that was voluntarily constraining our demand by only delivering in a few locations in London. This made it hard to grow, both because we were making the chicken and egg problem even harder and because we couldn’t take full advantage of nationwide marketing channels like SEO and SEM (I know you can specify a location for SEM but this leads to customers getting saturated with ads for the same ad spend, rather than spreading it out across a broader population). This further exacerbated the cost issue as we were paying for a half-empty warehouse that we didn’t grow fast enough to fill.

Around May this year, we made the decision to shut down our existing operations and pivot them into something more viable that didn’t burn as much cash. We spent the next 6 weeks exploring models and executing the pivot. Our new approach solved the issues in the old model but presented new ones of their own.

We moved to a model where local farmers would deliver processed orders to a partner farm that would act as a regional hub. They would process the items into customer orders for us before we sent them out nationwide via DPD. This enabled us to open up demand nationwide and work compatibly with DPD’s model. It also reduced costs as we no longer needed to pay rent. Instead, we could pay our partner farm a processing fee for each order processed, turning the fixed cost of rent into a variable cost that used the under-utilised assets and space on the farms.

This all looked good on paper but the practicalities of this were much harder. Farmers pushed back on the approach as they now had to pack orders and deliver them to the hub. Although we paid them for this, they felt it wasn’t worthwhile, particularly when they were used to having a box of unprocessed orders collected by DPD in our old model. We hoped the concerns over delivery would decrease as we brought on more supply since they could share the deliveries, but we didn’t scale fast enough to see if this was true.

Additionally, at our trial partner hub, we were working around the existing farm operations, rather than having a blank slate to build our own custom approach. This led to mistakes with packing and customers would often receive the wrong items. Partnering with the farm and getting them to run operations meant we had limited control over mistakes and efficiency. We had a great relationship with them and they were working as hard as they could, but ultimately the setup meant it was hard for us to implement optimal processes. This then led to a customer experience that didn’t meet their expectations.

Ultimately we ran out of time to get the operating model to work (if it ever can with today’s technology) and didn’t have the metrics needed to raise our next round of financing. We burnt too much cash pre-pivot and this caused us to run out of time to figure things out further with the new model. We had a signal from the market that we were onto something but we didn’t have a product that delivered the experience customers wanted. Customers were interested in what we were doing but we didn’t meet the quality, price and experience they expected which led to poor retention.

I saw the writing on the wall and decided to shut down while we had the cash for everyone to have a soft landing and return funding to investors, rather than trying to force it to the end. Maybe we could have raised our next round but I believe we would always be on the back foot from then on as we’d likely get poor terms which would challenge future investors and make it difficult to attract the talent we needed to make Furrow work.

I ultimately believe there is something in this space. The world is moving towards platforms and coordinated decentralisation with no gatekeepers. The challenge in food is that the physical distribution of produce is not yet up there with the digital distribution of harvest availability information.

Why we failed

Ultimately we couldn’t find a distribution model for a farm-to-table supply chain that met customer and farmer needs, while also being, or having a route to become, profitable. There are a few underlying issues behind this.

Food is a tough, unforgiving industry

Food is probably the hardest industry for logistics. Maintaining freshness and quality across hundreds of product types, each requiring slightly different conditions is no easy task. We thought we could overcome this with the freshness of our produce and the shortness of our supply chains by harvesting products to order and shipping overnight. However, we still faced challenges around packaging, maintaining the cold chain for chilled products and ensuring the products arrived in good condition. Getting the right products to the right place at the right time in the right condition is difficult.

Combine this with the low margins that are inherent within the industry due to customer expectations and years of driving down costs by supermarkets (the UK has some of the lowest spend on food as a percentage of income in the world, most of which is due to farmers being squeezed and the environment being degraded as a result) and it is a recipe for a difficult time.

Additionally, customers have high expectations when it comes to food. Global supply chains mean there is an expectation for almost all products, all year round, with little regard for seasonality. This is changing but I do not think customers will sacrifice their diet, particularly in the winter months, in order to eat more sustainably. The sustainable abundance mindset of energy applies to food as well. We need to find ways to meet our demands for food by sustainable means, rather than cutting back.

Didn’t remove steps from the value chain

In Working Harder and Smarter, Packy McCormick discusses how advances in atoms enable us to work harder by cutting out or rapidly speeding up entire steps of a value chain, whereas advances in bits enable us to work smarter by coordinating our activities in a more efficient way.

Furrow fell into the working smarter camp (sort of, and even this may be a bit of a stretch). We were rearranging the activities of the value chain to be done by different participants. Before wholesalers and other participants would do the steps of packing and distributing the produce, now we were asking farmers to do this instead. Sure we were giving them more money but their work was increasing too. We still had to do the same number of steps in the value chain to get products to customers and there were still costs associated with them. The overall amount of work that needed to be done was the same so we weren’t stripping costs out of the system.

Didn’t have a 10X product

Not removing steps from the value chain meant we didn’t actually have a 10x better product. We were 10x better than going to a farmers market, and possibly going to a supermarket, but we weren’t 10x better than Ocado, Tesco.com etc. Consumers ultimately care about convenience, price, quality and selection. Provenance is nice to have that some consumers want but they ultimately care about the other points. To reach real scale, you need to improve the criteria above and we weren’t doing that.

Paraphrasing Dalton Caldwell, “a weak signal from the market is the most dangerous. If it was a strong signal, you’d know and have the makings of product-market fit. If it was no signal, you know your idea is a dud. But a weak signal gives you false hope that maybe there is something there with a few twists and tweaks. Founders can risk wasting years of their lives on this false hope”. Furrow probably fell into the weak signal risk. I could say we proved there was a problem and people were willing to pay for it, but we couldn’t find a way to give them what they wanted, but realistically if we were onto something more people would have put up with our bad product and we wouldn’t have poor retention and limited word of mouth growth.

Dalton Caldwell @daltonc

A dangerous (and common) scenario is a startup with a little bit of traction... just enough traction that they aren't willing to change ideas/markets and not enough traction for them to actually know they have product market fit. It's the uncanny valley of product-market fit.

Having said that, I think there is enough customer signal that if you can fix the operating model to remove steps from the value chain and achieve a cheaper price/better quality/more selection/more convenience AND provide the provenance/satisfaction of buying direct from the farm, then you’re onto a winner.

Didn’t understand our suppliers well enough

I didn’t chat with our farmers enough to really understand the issues they were facing. I was more motivated by the consumer side since this was a problem I wanted to fix for myself and I thought the farmers would fall in line. I assumed they would be like me and motivated by money, whereas most, or at least the initial farmers and producers we had on board, just wanted to farm. This meant our new operating model, despite paying them, didn’t go down well and we’d either have to reconsider it to make it better for the farmers or pay them far more, which would impact our already tight margins. We may have been able to solve this by finding an initial base of more financially motivated farmers but I believe we needed to prove our model with smaller producers before earning the right to work with the larger farms. These small producers are typically lifestyle farmers.

There is a quote from this fantastic retrospective on Advisable that I am going to paste below as it sums up our situation pretty accurately.

Related to the above, we didn't understand our customers well enough. This is a tricky one however as given the fact that we were trying to build something that we hoped they would want but not necessarily something that they were asking for - we just weren't building to solve a problem that they had.

This meant that we were able to talk to them about how it would ideally work for them and get feedback from them on our attempts, but they probably wouldn't be able to direct what we do in the same way that they could if we had been building a more incremental solution - doing something they already do faster/cheaper/better.

In any case, we didn't understand them well enough to build something transformative for them.

Additionally, unless a farmer was just starting out, they already had routes to market for their produce and were multi-homing by selling their produce on Furrow. They didn’t want to move their customers from their veg box or local farmers market to us and farming is not like restaurant delivery where every incremental order can be fulfilled by the kitchen - there is a long lead time on growing incremental produce to expand your customer base. As a result, we were often getting surplus produce from our suppliers, rather than them viewing us as their main source of demand. However, we were a demand-constrained marketplace, and the supply typically went where the demand is, so as we grew this may have been overcome.

Execution

In addition to the structural issues with Furrow, we also didn’t help ourselves with our execution. While there was a lot we got right, there are several things I would have changed if we were to start again.

1. I didn’t prioritise effectively and ruthlessly enough

   This resulted in wasted time and money as I got bogged down too much on unnecessary things that I thought were important. This was particularly the case around culture where I was so determined to make sure we had it set up from the start that I got paralysed choosing the ideal company wiki software and writing culture and communication documents, rather than doing the things that actually set the culture like effectively executing. Luckily many of our prioritisation mistakes were quickly identified, learned from and reversed but they cost us time and money.

2. I was a solo founder that didn’t have the right setup and support network around me to make sure I was nourished by the work

   Starting a company on my own remotely, and then running it remote-first once we had employees, was incredibly lonely and isolating and I think ultimately led to me feeling burnt out - we all need face-to-face human connection and this is something the remote vs in-office debate often misses. I have learned so much about myself in this process and know so much more about what makes me tick and the support systems that I need in place to ensure I am nourished and enriched by my work.

   To be clear, I am not saying I didn’t have the right team around me. I am more referring to making sure I built a broader support network of family, friends, and other founders to make sure I was doing ok.

3. Build a larger team around me so we can move faster

   We could have moved faster if we had a bigger team that had the skills to quickly test a lot of what we wanted to do (for example, I have reasonable tech skills but I am no software developer), but that also had the risk of higher burn while we figured things out. Ultimately I think this would have been worth the tradeoff as startups are all about momentum and investors don’t look at ‘momentum per employee’, but rather overall momentum (within reason, usual internet blog caveats apply). Having a bigger team would also mean we could spend more time growing the business, rather than running it.

4. Reliance on 3rd parties

   Whether this should be a separate point outside of execution I’m not sure, but we had a lot of challenges relying on 3rd parties. From spending time managing the relationship and chasing lost deliveries to being unable to implement our own ideal processes at a partner hub, 3rd parties were a blessing and a curse. They were great since we didn’t have to build out those capabilities ourselves, but being a small fish meant we had very little bargaining power to get the level of service we needed. I don’t know how I’d do Furrow again with less reliance on 3rd parties (short of building out our own nationwide delivery infrastructure) but carefully considering what is a core competency that ideally needs to be in-house would be a start. As Jeff Bezos says: “Does it make the beer taste better?”

Do I still think that what we were pursuing was a fundamentally good idea?

Yes. I do think that a platform-based food system is inevitable. The challenge is finding an operating model that delivers a 10x better solution for customers and sellers while being cost-effective. My view is that existing players have focused on digitising the existing supply chain but that real value is unlocked when you re-design the entire supply chain from the ground up to make the most of new technology.

I believe the world is moving from pipelines to platforms and protocols. We’ve seen this in many industries as gatekeepers have been removed and we’ve moved towards coordinated decentralisation. This has predominantly occurred in the world of bits with things like Substack in journalism and Maven in Education, but also in the physical world with things like residential solar and batteries.

I also believe that people buy from people and the current first wave of eCommerce has lost that. We wanted to fix this once we’d sorted the base operating model by bringing in social features such as farmers hosting live streams and posting photos and videos of life on the farm. If you can bring the connection, intimacy and community of a farmers market to an online browsing experience, customers will have a great time.

A farm-to-table/platform-based food supply chain will happen. Someone just needs to identify the operating model that works. This may not be possible with our current logistics technology but as this continues to improve, it will become feasible.

What’s next for me?

I’m not sure. I am taking some time off to relax, review the past two and a bit years and do some soul-searching. I’m not sure whether I’ll start up something else, join another startup, go into investing or do something else entirely. I’m toying with a few ideas in my head, and am interested in other industries where the pipeline to platforms and coordinated decentralisation shift is occurring, as well as continuing to noodle on decentralised commerce/logistics that enable smaller, lower value items to be shipped cost-effectively.

In the meantime, while I figure things out and think about new ideas, I’m looking for a part-time role as an Entrepreneur in Residence, Chief of Staff/Founder’s Associate, or another special projects type role where I can use my experience to help founders.

If you have any ideas or know of any roles you think I’d enjoy or be a good fit for, please let me know!