4.9x more profit per pound, a $514K ceiling on a $1M target, and 4 errors recovered
A commodity-scale US agricultural producer
The Brief
An agricultural producer asked us to model a direct-to-consumer launch. The format they had been told to build would have destroyed four fifths of the profit on every pound sold. We found it in their own numbers, before a dollar was spent.
The producer moves well over ten million pounds a year into the commodity market, where price is set by the buyer and the grower’s name is worth nothing. Leadership had set a target of one million dollars in branded revenue, and had been advised to build a direct-to-consumer store around bulk formats, on the reasoning that a larger pack carries a larger margin.
Two things made this hard. First, the operator did not think in the units the analysis was written in. Marketplace work is reported in margin per unit sold; a farm thinks in pounds moved and dollars per pound. The existing analysis was correct and almost useless to them, because it never answered the question in their unit.
Second, we inherited a completed analysis and were asked to extend it, not to audit it. Extending a number you have not verified is how an error becomes a strategy.
Profit per pound, by route to market
Every channel expressed in the operator’s own unit, modeled against real listing prices and real fulfillment costs.
| Route to market | Profit per pound | Verdict |
|---|---|---|
| Bulk direct format, as originally specified | $0.83 | Value-destroying. Worse than wholesale. |
| Wholesale | $0.99 | Low per pound, but the only route that absorbs real tonnage. |
| Marketplace, standard pack | $2.41 | The working baseline. |
| Multi-item bundle, four existing packs What we recommended | $4.05 | 4.9x Best route for the volume crop. Zero new packaging. |
The Approach
Verify everything against the raw exports, not against the finished documents. Then translate the entire question into the unit the client actually operates in.
Re-derive, do not review. We rebuilt every inherited figure from the original marketplace exports rather than checking the documents against each other. Internal consistency is not correctness: a wrong number copied faithfully into four files looks like four confirmations. This is the discipline that found the errors below, and none of them would have surfaced any other way.
Change the unit. We rebuilt the financial model around profit per pound of production sold, alongside pounds moved. Every channel, pack and price was then expressed in the operator’s own unit, which made the channels directly comparable for the first time and immediately exposed which of them could not do what was being asked of it.
Test the recommendation before writing it. The bulk-format thesis we had been given was modeled against real listing prices and real fulfillment costs rather than assumed ones. It failed. We rebuilt the thesis rather than presenting it, then propagated the correction through every section that depended on it.
What we Did
Rebuilt the channel economics from primary data. Fulfillment fees were derived from live fee data on real competitor listings at each pack weight, not estimated from a fee table. Carrier rates, third-party pick-and-pack benchmarks and observed bulk market prices replaced the placeholder assumptions in the inherited model.
Found the two inputs that changed everything. The inherited model priced the proposed bulk direct format at a level the market does not pay, and omitted pick-and-pack labor entirely. Correcting both inverted the conclusion.
Found the format the data actually supported. Fulfillment and shipping are charged once per order, not once per item. That single mechanical fact means the lever on a direct channel is basket size, not pack size. A heavy single item cannot carry its own shipping. A multi-item bundle can, and it needs no new packaging at all, because it is the existing marketplace packs sold together.
Modeled the ceiling on the channel they were counting on. We modeled advertising as spend levels rather than growth stages, which surfaced a constraint the previous model had hidden. Efficiency worsens as spend rises, because heavier budgets buy broader and less qualified traffic. Past a certain point the channel converts advertising into volume and brand, but not into margin.
Each panel has its own scale, so the two series are never read against a single axis. Contribution turns negative once advertising passes 37.6% of sales.
Answered a pricing question that was not a pricing question. Leadership had asked why a competitor’s product was priced at roughly twice theirs. It was not a pricing difference. It was a pack-size difference: the two products were the same crop at a twentyfold difference in quantity, aimed at entirely different buyers. Reframing it opened a bulk format that moves ten times the volume of the one on the shelf.
Results in Detail
This was a pre-launch strategy engagement. There is no post-launch revenue to report, and we will not manufacture any. What follows is what the analysis found, and what each finding was worth before a dollar was committed.
4.9x the profit on every pound sold. The format the client had been told to build earns $0.83 a pound. The format the data supports earns $4.05, with no capital spend, because it uses packaging that already exists. The original format would also have required a new packaging line.
The $1M target was not reachable inside the modeled spend range. The hardest advertising push the model supports reaches roughly $514,000 a year on the marketplace channel that target depended on, and returns less contribution than a budget a third its size. Contribution turns negative once advertising passes 37.6% of sales.
Found before the budget was committed rather than after. The finding is what forced two further channels into the strategy.
Increasing advertising spend sixfold produces 4.3x the revenue and 12% less contribution. A clear stop line on a budget that would otherwise have been increased on the assumption that scale improves efficiency.
Four material errors recovered from the inherited analysis, including a category understated by 58% and a fulfillment fee where the correct figure was 65% higher than the one carried in the model, which had overstated one product’s margin by 16 points.
Every downstream decision that rested on those figures was rebuilt on correct ones. The largest correction made the lead product’s case stronger, not weaker.
Error 2. Fees charge on pack weight, not contents. The correct fee is 65% above the one the inherited model carried.
The marketplace channel moves 0.39% of annual production at the planning share, and 0.25% to 0.55% across every advertising level modeled
It reframed the entire strategy. The marketplace is not a volume solution and never could be; it is where the brand earns proof cheaply. Pounds move through wholesale. The client stopped asking the marketplace to do a job it cannot do.
The marketplace channel against total annual production, drawn to scale. 0.39% at the planning share; 0.25% to 0.55% across every advertising level modeled.
The finding that proves the promise
- Our own first-pass thesis was that bigger packs earn more per pound. The client’s raw data said the opposite: bulk buyers pay far less per pound, so a bigger pack earns less, not more. We rebuilt the recommendation rather than defend it.
- The corrected answer was also the better one. It is more profitable, it needs no capital, and it is easier to execute. Being willing to be wrong in front of the data is not a cost of the method. It is the method.
Tools & technologies
What the build runs on
verified
Evidence appendix - 12 figures
Figure
Source
How it is reproduced
$0.83 / $0.99 / $2.41 / $4.05 per pound
Financial model, channel economics tab
Order value minus (raw cost + packaging + pick-and-pack + shipping + payment processing), divided by order weight. Fulfillment and shipping are charged once per order.
4.9x
Derived
$4.05 divided by $0.83.
$514,000 top case; 37.6% threshold
Financial model, ad scenarios tab
Revenue equals spend divided by target advertising cost ratio. Contribution equals revenue times blended margin, minus spend. Contribution reaches zero where the ratio equals the blended margin. The $514,000 is the highest of three modeled spend levels, not a derived cap, and the 37.6% blended margin is a planning input rather than an output of the SKU model.
6x spend, 4.3x revenue, 12% less contribution
Financial model, ad scenarios tab
Compare the conservative and aggressive rows.
Error 1: a category understated by 58%
Raw marketplace export
Child-listing revenue column, summed after de-duplication. The inherited figure matched no consistent clean of the source: its unit count implied bulk listings were excluded, its revenue implied they were included.
Error 2: a fulfillment fee 65% higher, overstating one margin by 16 points
Live fee data on comparable listings
Fees charge on pack weight, not contents, so two packs of the same weight carry the same fee. The inherited model charged one product far less than an identical-weight sibling. The corrected fee of $7.23 is 65% above the $4.38 the inherited figure implies. Margin moved from 51.4% to 35.5%.
Error 3: a category pool labelled verified that was not
Raw marketplace export
The published figure appeared nowhere in the source. Re-summing the three category revenues gave a figure 13% lower. Every share in the study is measured against the corrected pool.
Error 4: a flagship competitor row carrying another product’s data
Raw marketplace export
The lead document’s price, unit and review counts were traced to a different listing in the same export. Restored to source.
0.39% of annual production
Financial model
Category units times share, times pack weight, times twelve, divided by annual production. The share is a 5% planning assumption; the modeled advertising range gives 0.25% to 0.55%.
Well over ten million pounds a year
Grower-reported production, stated as a band
Exact figure withheld as identifying. The band is what the argument needs.
A twentyfold pack difference; ten times the volume
Raw marketplace export
The two listings compared were the same crop at 0.5 lb and 10 lb. At equal unit velocity the bigger pack moves twenty times the pounds. The ten times figure holds the shelf pack at twice the unit velocity of the bag, which is the conservative case.
$1M revenue target
Client-stated objective
Taken from the engagement brief, not derived.
Source captures from the research tools - 6 Screens
Helium 10 Xray · category sizing
Category sizing for one SKU. The $611,719 total spans all 46 listings the search term returns, including non-premium items; the category figure used in the model is the filtered premium subset.
Helium 10 Magnet · search summary
Keyword discovery from a single seed term: 709 keywords returned, 190 organic and 111 paid.
Helium 10 Magnet · keyword table
The unfiltered return carries broad and adjacent terms alongside the relevant ones. Several rows here return no search volume and were removed before analysis.
Jungle Scout · product database
The account was signed out when this view was captured, so the sales and revenue columns are masked. The fee and dimension figures used in the model were pulled from a signed-in session.
Google Trends · seasonality
Twelve-month view, United States. The index is normalized to its own peak rather than to absolute volume, so it is read here for timing and direction only.
Google Trends · related queries
Related queries for the same term and window. Interest is relative within the category, and the top ten are dominated by adjacent species.