Solving for Novel Uncertainty
in produce markets
Preface
At the start of Pipehaul, the core hypothesis was that we could solve the logistical inefficiency in produce markets by deploying few optimisation techniques. Our focus was on logisitical moment between two APMC mandi(s) on either sides, production and consumption regions.
But the process of viewing the agri-space with a logistical lens helped us see the root cause of inefficiency, Trade Practises.
The practices involved in the trade are determining how the logistics happens in produce markets. In order to solve, we had to address a novel uncertainty of grading and trading produce between Farmers and Buyers.
The farmers in the production region and the buyers in the consumption region form the starting point and ending point of post-harvest produce supply chain.
Brief on uncertainty
The writing of Jerry Neumann who is a venture capitalist writing the blog reaction wheel. He writes in his post about addressing different types of uncertainty
The two primary types of uncertainty we care about are novelty novelty—when you can’t predict something because no one has done it before—and complexity—when you can’t predict something because the system you are in is changing in an unpredictable way
Owning the Trade
In order to create a marketplace model, we started directing our own trade with Subjimandi.app while pipehaul would handle the logistical capabilities.
In order to achieve feasibility, we reduced the number of moves produce takes before reaching the end destination.
We did a qualitative analysis of few shipments of our with respect to the Mandi system. We mapped the overall cost of the post harvest supply chain with costs per stakeholder. In the Mandi(s) system, the overall cost had both fixed and variable parts(commission). In our model, the cost was significantly less and had only fixed components. The benefits were significant enough to substantiate the benefit which supply chain should be picked.
Yet, the challenge was getting the stakeholders, especially farmers and buyers to change the way they sell and buy respectively. This was the novel uncertainty we were aiming to solve by building a marketplace around buying and selling produce grade wise.
Setting Benchmarks: Grade
Frist, let us take a look at what grade looks like with a simple example of Ridge gourd.
The physiological traits of produce determine the grade. The quality, inherent traits are similar for each of the three grades. Grade defines the utility of the produce basing on a set of parameters that are publicly published by the regulating authority. In our case, government of India.
Typically, Farmers sell their produce in “lots”, everything harvested with non-standardised segregation. In essence, they are looking to offload(sell) everything they produced. It may be of ”premium grade” or “no-grade”. The incentive for the farmer is to sell everything and get maximum amount of money for entire lot of harvested produce.
On the other side, retailers are looking to buy produce in “lots” at the cheapest possible rate. They verify the standards of produce through sampling in mandi(s) while the “lots” are non-standardised.
Our preliminary approach on both the farmers and buyers was to convey why we were better alternates. It was a narrow space we could occupy. For farmers, we would have to pay more than the competition. Similarly for buyers, we will have to be cheaper than the competition of local mandi(s).
The arbitrage presents itself due to local mandi(s) in production and consumption region having different prices. The difference has margins for us to work with and make transactions. But the probability of it being profitable is less. It is not sustainable over a longer duration of time.
In addition, we were unable to convey the importance of grade. By playing by the competition and prevalent trade practices, we were assessed only on price. Graded produce became a bonus on top of cheaper produce in the first place.
Getting Squeezed out
In the beginning of operations, we wanted to shape the buying behaviour of the Buyer, grocery retailer. So, we procured “lots” from Farmers or local mandi(s) in production region. We could never predict the graded output of a non-standardised “lot”. Be it mandi or farm, the practice didn’t allow us to predict accurately. So, the sale could only take place once the “lot” is graded.
This resulted in a sunk cost model where the demand for each grade had to be generated in a short span of time but we weren’t able to do so due to lack of distribution across different types of buyers. When we grade the produce, retailers fit one grade, hotels fit an another grade and vendors in weekly markets fit an another grade.But none of their buying behaviours matched on the same day. This created a problem as we were restricted by the produce we graded and who all we could sell.
The economics of arbitrage was not working as well. Grade value was appreciated by buyers only when the price is within the local mandi(s) price band.
For example, If you are grading a non-standardised “lot” of 1 ton. After grading you got 200 Kgs of Premium , 300 Kgs of Standard and 500 Kgs of No-Grade. The current crop of buyers who were predominantly retailers wanted only the Premium and Standard. The remaining 500Kgs we had to find much more non-organised vendors who sell in weekly markets.
The complexity further increases if the distribution in a “lot” is skewed on the Non-grade. These have to be sold at a lower price than the procurement price of the “lot” because you are selling it to a different type of buyer. They are buying this grade because it is feasible for them.
The cost of procurement for all these grades are the same. The cost of grading is distributed equatively to all grades. But, the sale price of each grade was different. We had to create pricing tiers basing on how much we spent on the ”lot” rather how much the customer is willing to pay. The the pricing was far higher than local mandi(s) prices.
No matter how well we tried to guess the “lot”, it was never accurate. Trust us when I say, we did this over 200 times before stopping it.
The practice of buying “lots” and selling graded produce was hurting us because the team selling the grade were deterred due to the pricing, grade availability and lack of buyer segments to sell all grades.
Flip the trade, align your model.
We paused the operations of procurement and sales for a week to caliber our model. We wanted to align it to the outcome we would like to achieve, graded produce trade.
For us to sell all grades, we need to procure all grades. But instead of procuring “lots” we need to move into procuring “grades” from the farmers.
Everyone in the team said it was improbable. We cannot convince the farmer. Yet, it was not a model feasibility problem, we could grade at farm. In essence it was communication problem where the value we were trying to provide to a farmer was being evaluated by the competition of local mandi(s).
If you see it from the lens of farmer. Every entity in the agri-space is selling to them. Starting from Input companies to post-harvest procurement companies that transact with farmer. Each one of them is selling farmers how they are better for them.
For example, a trader who buys at farm sells the price at which he/she procures. So, in a way the farm buyer is also selling price.
This is when the team on the front lines was asked to think as sales force of farmers.
What does a farmer have to sell?
His or her harvested produce.
So, for someone to buy this produce, how do we communicate what type of produce?
Grade defines the type of produce and helps you find buyers for farmer’s produce.
In terms of business model, we stopped making money on produce sales. We made price of graded produce a single price on the marketplace.
For example, the price of premium ridgegaurd is ₹ 15 / kg for both Buyer in consumption region and farmer in production region. I wrote about this in the previous article, alluding to normalisation of prices.
The end game was to build a produce marketplace where buyers and sellers transact to buy produce of their needs and wants. Meaning, the producer gets the same price as the buyer pays for produce the total cost will differ due to logistics involved in getting the produce to their location.
By switching gears of making money in delivery of trade we were differentiating our value proposition. For both buyer and farmer to trade, they have to value grade first.
The first conversion was the toughest. But since then, we did over 900 farm procurements till we closed our operations. The second order benefits of this were numerous. We will address them in a later post.
Back when we paused operations for the first time, no buyer or farmer missed our service. This was when I realised the futility of businesses innovating in essential commodities. But, post our closure after the implementation of graded produce. To this day, we get at least few calls every harvest about our operations.
The uncertainty around the graded marketplace was put to rest with the adoption of www.subjimandi.app in retailers and repeat farmers who were patient to receive our services when we were stretched too thin.
Limitations:
There is a very high chance that the success of this model would not be possible at a larger scale. It may not have been sustainable economics at scale. The business was not viable or venture fundable. Unfortunately, I failed before answering these questions.
The simple nature of our marketplace confounds many outsiders and other industry folks. They don’t believe that grade is not the norm, quality in most cases is subjective.
Yet, many startups re-iterates the same discoveries without solving any new uncertainty. The dominant funding in this space rewards speed over velocity. The moat or edge of a startup is to better manage risk to navigate the complexity of agri-space, be it climate, capital or supply chain risk.
Thus prevailing the existing norms of trade practices and ancients rivers of cashflows.
Signing off till next time,
Vivek V.S, thinking about uncertainty