Logging Without a System
Everything on this site so far assumes a record. Many operations — small fleets, single-facility shippers, brokers — do not have one, and are told the answer is software. For a software-oriented reference on recording work time, additional details.
It is not, at least not first. Six columns kept consistently for a month produces more usable information than most telematics deployments produce in a year, because the columns were chosen for the question rather than for what the system happens to emit. For broader transportation and measurement context, see Bureau of Labor Statistics.
The six columns
Date and load reference. One row per visit.
Facility. Name and city. You are going to group by this.
Gate in. The event, not "arrival" — the moment the vehicle reached the gate, from whatever source you have.
Gate out. Same, at the end.
Mode. Live load, drop-and-hook, or drop-and-pick. Mixing these makes your own data uninterpretable, and it is one keystroke.
And a cause note, only when the visit was long. Three words. Waiting for door. No paperwork. Lumper late. Blank on normal visits, which is most of them.
That is it. Fifteen seconds per visit, and the sixth column is what turns a dataset into a diagnosis.
Two optional columns worth adding later
Once the six are habitual.
Check-in time, which lets you separate the yard segments and see whether your queue is outside the gate.
Remaining hours at departure, which is the safety and scheduling exposure and the number that matters most to a driver.
Do not start with eight columns. Start with six, add these when the habit holds, and accept that a record with gaps is usable while an abandoned one is not.
Who fills it in
The driver, by message, at two moments. Gate in, gate out. Whoever receives the messages transcribes them, or the messaging system timestamps them and you read the times off later.
Not a form. Not an app to install. Not a login. Anything requiring more than sending two short messages will be skipped exactly on the days that matter — the busy, difficult ones, which are the visits you most need in the data.
And record the fast visits. The fast ones are what make the slow ones credible, and a log containing only bad days is a complaint rather than a dataset.
What a month is worth
Per facility: a median, a 75th percentile, a worst case, and the share of visits over your free-time allowance. Four numbers that no industry survey can give you about your own operation.
A cause list. The three-word notes cluster fast. Usually two or three causes account for most of the long visits, and at least one of them is fixable by scheduling rather than by anyone spending money.
And an evidentiary basis. A consistent record with a stated method supports a claim that would otherwise be refused outright.
Thirty visits is enough to see a shape at one facility. A hundred across several is enough to compare them.
When to buy software
When the manual log has proved which questions you have.
That is the whole sequence, and doing it in reverse is the common error: buying a system, receiving whatever it measures, and discovering it does not answer the question — usually because it starts the clock at check-in or uses an unspecified geofence.
A month of paper tells you what to require of a system, which turns a purchase from a hopeful one into a specification.
The one rule
Consistency beats precision.
A slightly imprecise time recorded on every visit is far more useful than an exact one recorded when someone remembered. The analysis you will do — medians, distributions, comparisons between facilities — is robust to a few minutes of noise and destroyed by selective recording.
The short version
- Six columns: date and load, facility, gate in, gate out, mode, and a three-word cause note when the visit ran long
- Fifteen seconds a visit, and the cause note is what turns data into a diagnosis
- Two short messages from the driver, not a form or an app — anything heavier gets skipped on the days that matter
- Record the fast visits; a log of only bad days is a complaint rather than a dataset
- A month gives you a median, a 75th percentile, a worst case and the share over your allowance, per facility
- Buy software after the manual log has established which questions you have, not before