Guides
How to observe a food truck operation with a skeptical eye: a 25 point field protocol
A 25 point field observation protocol for watching a food truck through setup, peak, board and wind-down, plus what observation cannot tell you.
What to take away
- No trucks are named on this page. Publishing commentary about identifiable businesses that nobody verified is not research, it is invention.
- What replaces the list is a field observation protocoltwenty five things you can see for yourself, standing in front of any truck, in about twenty minutes.
- Watching an operation trade tells you more than any interview, because throughput, layout and queue behavior are visible and nobody can overstate them.
- Observe operations unlike yours as well as similar ones. A dessert truck and a taco truck solve the same throughput problem differently.
- Write the observations down at the time. A remembered impression of four trucks is not a comparison.
Why nobody is named
Three reasons, and they compound.
Named operations change: menus, owners, cities, and formats move. A list written today misleads within a season. Commentary about a specific business is a claim about that business, and claims should be checkable.
The interesting part was never the name. It was the decision the operator made, which you can observe directly.
So go and look. The protocol below works on any truck you can walk up to.
How to run the observation
Pick a service you are not working. Arrive before the peak, stay through it, and stay a little after. Stand where a customer stands rather than where an operator would.
Take notes at the time on a phone or a card. Do not photograph people, do not obstruct, and do not interview the crew during service. If you want to talk, come back at the end and ask one question rather than five.
Before the queue forms: the setup
Before the queue forms
- How long did setup take, from arrival to first order taken?
- Where is the generator, relative to the window and to where the queue will stand?
- What is the truck parked on, and how level is it?
- Which side does the window face, and how much depth does the queue have?
- Is there any shelter, shade or wind protection for people waiting?
- How is the board displayed, and is it readable from thirty feet and from the back of the queue?
During the peak: the line
During the peak
- How many people are working, and what is each one's fixed job?
- Does the person taking orders ever leave the window to cook?
- How many distinct cooking or assembly steps happen per order?
- What is the ticket time from order taken to handover, timed on three separate orders?
- Where does the queue physically stall, if it does?
- Is the handover point separate from the ordering point?
- How many questions are asked at the window per ten orders, and what are they about?
- How is a run-out communicated, and how far back down the queue does the news reach?
- What happens when a payment fails?
- Does anyone leave the queue, and at what point?
Point ten is the one to be rigorous about. Time it rather than estimating, and time three orders rather than one. Ticket time is the closest thing to a universal measure of how well a truck is designed.
The board and the offer
- How many items are on the board, and how many decisions does an order require?
- How many of those items appear to share components?
- Is there an obvious default, a package or a recommendation?
- Is portion size communicable without asking?
- Is allergen or ingredient information available without a conversation?
- What is packaged in advance, and what is assembled to order?
Point eighteen is the most instructive and the least obvious. A board of eight items built on four components is a completely different operation from eight items built on twelve, and you can usually work it out by watching what gets pulled from where.
Board width vs components
Eight items, four components
- Items on board
- 8
- Components pulled
- 4
- Decisions per order
- Fewer
- Assembly load
- Lower
Eight items, twelve components
- Items on board
- 8
- Components pulled
- 12
- Decisions per order
- More
- Assembly load
- Higher
After the peak: the wind-down
After the peak
- How long does breakdown take, and how many people does it need?
- What happens to waste, and is there any sign of where the water and greywater go?
- Does the crew look like they could do it again tomorrow?
The last one is not sentimental. Sustainability of the block is the constraint that ends more trucks than demand does, and it is visible at the end of a service in a way it never is in a busy photograph.
What to do with the notes
What to compare it against
- Their ticket time
- Yours, measured the same way
- Decisions per order
- Your own board
- Where their queue stalled
- Where yours stalls
- Their crew count at peak
- Yours
- Their setup time
- Yours
What it might change
- Their ticket time
- Board width or station layout
- Decisions per order
- Packages and defaults
- Where their queue stalled
- Which step to redesign
- Their crew count at peak
- Whether you are understaffed or overstaffed
- Their setup time
- Whether your block is longer than it needs to be
Do not copy anything wholesale. Their permit conditions, pitch, support kitchen distance and crew are different from yours, and a decision that works for them may be impermissible or impractical for you.
That is particularly true of anything involving on-board preparation, since what a mobile unit may cook, cool or assemble is set locally. The FDA's state-by-state directory of the agencies that administer retail food regulation identifies who decides that for you. Anything you want to adopt has to be checked against your own authorities and recorded in the compliance register.
What observation cannot tell you
It cannot tell you whether the operation is profitable. A long queue at a thin margin over a six hour block can be a loss, and none of it is visible from outside. That calculation is your own, and it is priced from the whole block rather than the plate.
What observation cannot tell you
- Whether the operation is profitable
- What their permits allow
- Commissary cost and drive distance
- What the job costs the crew
- How customers found them
It cannot tell you what their permits allow, what their commissary arrangement costs, or how far they drive. The Small Business Administration's guide to planning and running a business is a reminder of how much of an operation sits in registration, insurance and finance decisions that never appear at the window.
It also cannot tell you what the job costs the people doing it. The Bureau of Labor Statistics description of what running a food service operation involves covers staffing, supplies, preparation and sanitation. It adds customer issues, budgets and payroll records.
Schedules often include nights, weekends and holidays, and a crew that looks effortless has usually been doing it for years, a training question covered in how a truck is staffed and trained.
And it cannot tell you how people found them, which is the invisible half of any truck's performance and the problem set in how a moving business publishes itself. Whether what you learn justifies a new pitch, a new unit or nothing at all is a growth decision with its own criteria.
Common questions
How many operations should I observe?
Enough that patterns appear rather than impressions, which is usually four or five run properly. One is an anecdote. Ten observed casually is worse than four observed with the protocol.
Should I tell them what I am doing?
If you speak to them at all, yes, and afterward rather than during. Most operators are generous with a single specific question at the end of a service. Nobody is generous during a peak, and standing in a queue taking notes without buying anything is worth avoiding.
Is it worth watching trucks in other cities?
Yes, for design and layout, which travel. Not for regulation, permits or pitch strategy, which do not. A layout idea from another market can be adopted; a parking practice cannot.
What if the truck I watch is doing everything badly?
That is often the most useful observation of the lot, because failure is legible in a way that success is not. Note exactly where the queue stalled and what caused it, then check whether your own operation has the same weakness in a less obvious form.
