Profitability Projections For A "Minimum Viable Bakery"
The Economics Of A Kolache Bake
Tom and I are starting a baked goods business, in a pandemic.
Here’s how we expect to make (or lose) money.
Costs for a Bake [1]
Fixed
$200 - day kitchen rental
$240 - labor
Variable (ish)
$0.75 - Ingredients per kolache
Prices / Profitability at Various Scales
Our current intent is to sell kolaches for $2.25 wholesale, and $4 retail. [2]
Cost per unit at various scales [3]:
100 units: $5.15 cost per kolache
At 100 per unit, this really sucks. We lose money for every unit we move.
300 units: $2.22 cost per kolache
At 300 units, surprisingly, we can be unit breakeven for wholesale pricing and eek out a small profit at retail pricing [4].
It’s not a scale we’re interested in operating at, so if for some reason we got stuck here, this wouldn’t last long.
500 units $1.63 cost per kolache
At 500, we are a bit more than “ramen profitable”.
An example of what this looks like:
We do a single weekly bake in the rental space.
5 coffee shops take delivery of 50 units to sell over the weekend (16/day)
We sell 250 kolaches directly to customers
VERY feasible, and this is the lower bound of where we start getting interested.
In this scenario, we make ~$800/week (combined) for running a single bake.
Scale — 1000+ units
As a reference point, selling 1000 units in a day is not unreasonable for a kolache shop. 2000 is a big day. Pre-orders and increased demand on a holiday can take a single location kolache store to 5000 units.
If we can increase this “only bake on one day, in a rental kitchen” model to 1000 units per bake (operationally very reasonable), then Tom and I can get to a point where for baking 1 day a week, this has a net revenue of ~$100k/year [5].
Minimum Viable Bakery
Profitable “once a week” bakes in a rental kitchen is not our end goal — it’s is effectively the end range of the success path for the “minimum viable product” version of this concept.
The important part about our approach here is that the day rental model allows us to deliberately accept higher short term in the validation/proving phase in order to remove the large downside risk in the event of failure.
In practice, if we start moving down this path it’ll be enough validation for us to move to more permanent facilities and labor, which will increase our breakeven unit count but reduce our total costs at higher scales.
—
Footnotes:
[1] These are pretty stable over the range we’re currently operating, but with large volume changes, labor and ingredients will change. Also omitting a few things for clarity/simplicity in a blog post. Packaging, transportation, etc. And we’ve excluded founder labor, which will eventually be replaced by paid labor.
~
[2] To anchor that to how sane that is, here are some data points:
In Chicago, we haven’t gotten pushback on price from our initial conversations with retailers for the wholesale price.
A breakfast sandwich in Chicago could cost anywhere from $4-8
In Houston, a “premium” kolache typically caps out around ~$3.75 though there are some above the $4 price point.
I’ve heard wholesale kolache pricing in Houston for a savory kolache ~$1.65
A croissant sold to a coffee shop can be anywhere from $.80 to $1.50 from the wholesaler.
Basically, we’re not the cheapest, but it’s not unreasonable. Notes here for the benefit of non-Chicagoans who may not be familiar with pricing or non-food industry folks who may want context for wholesale pricing.
~
[3] These costs per unit include facility as well, because our facility costs right now are NOT a fixed cost. Since we’re doing day rentals, we only have to pay this *IF* we bake.
~
[4] I’ve excluded some costs for simplicity, but we could theoretically optimize our way to profitability from this. We wouldn’t continue a standalone operation at this scale, but as a building block to a multi-product food empire, this is still a valuable milestone for us.
~
[5] Sales - ingredients - labor - facility. For simplicity a lot of stuff has been excluded (packaging, transport, professional fees, etc.), which will lower the number, but this is IMO a valid first order approximation.