A contractor can estimate twenty jobs and still have twenty opinions.
The estimating system begins to improve when completed jobs produce usable evidence.
That requires two things:
- preserve what you predicted;
- record what actually happened.
Do not overwrite the estimate after the job.
Without the original baseline, you lose the ability to learn from the difference.
Freeze the sold estimate
Before work starts, preserve:
- estimated quantities;
- estimated purchase quantities;
- estimated labor-hours;
- labor cost;
- equipment assumptions;
- direct job costs;
- overhead calculation;
- quoted price;
- explicit assumptions.
Keep the calculated price as well as the final rounded quoted price where the company uses both.
The sold estimate is the reference point.
Record enough actual data to explain the job
You do not need a large enterprise platform to begin.
At minimum, capture:
- labor-hours by job;
- major material purchases;
- equipment or rental expense;
- subcontractor invoices;
- disposal;
- significant scope changes.
For important or repeatable work, go deeper and record labor by major operation.
A total labor figure can tell you that the job missed.
Operation-level data can tell you where it missed.
Calculate variance, then diagnose it
Suppose the estimate allowed 84 labor-hours and the unchanged work actually consumed 101.
Variance:
101 - 84 = 17 labor-hours over
Percentage variance:
17 / 84 = 20.2% over estimate
Do not stop at the percentage.
Ask why.
Possible causes include:
- quantity understated;
- production rate too optimistic;
- access worse than documented;
- crew skill mix changed;
- equipment failure;
- material delay;
- rework;
- customer change;
- hidden condition;
- unsupported estimating guess.
Different causes require different responses.
A production-rate problem should not automatically be “fixed” with a blanket markup.
Keep changed scope out of the production-rate database
Suppose a 400-square-foot patio was estimated correctly.
During the job, the customer adds another 100 square feet through the company’s approved change process.
If the original 400-square-foot estimate is compared with total hours for the final 500-square-foot scope, the data will falsely suggest the original production assumption failed.
Separate changed work.
Historical data is only useful when the boundaries are clean enough to compare.
Hold a short post-job review
For larger or important jobs, ask the people who performed the work:
- Where did the estimate feel realistic?
- Where did it underestimate effort?
- What work was not obvious from the scope sheet?
- What equipment or staging change would have altered productivity?
- What should the next estimator know?
Treat the crew as a source of production information.
The objective is not blame.
The objective is a better next estimate.
Build a production-rate library slowly
Choose recurring operations that materially affect price.
Examples can include:
- mulch spreading;
- shrub planting by size;
- sod installation;
- sod removal;
- edging;
- paver installation by a defined system;
- fence installation;
- cleanup;
- material handling;
- disposal.
Define the operation before averaging jobs together.
“Planting” is too broad.
A more useful internal definition describes:
- plant size;
- whether the bed is already prepared;
- staging distance;
- what the operation includes;
- what it excludes.
Now comparable jobs can actually be compared.
Use ranges and context
Suppose five comparable shrub-installation records produce labor-hour-per-shrub figures of:
- 0.25;
- 0.29;
- 0.27;
- 0.40;
- 0.30.
The 0.40 result deserves investigation before it is blended into a standard.
Perhaps:
- access was difficult;
- soil contained rock;
- plant size differed;
- the crew was new;
- the scope definition was inconsistent.
Add context tags such as:
- easy, moderate, or difficult access;
- hand work or machine-assisted;
- plant size;
- soil condition;
- crew size;
- haul distance;
- new construction or existing landscape.
The goal is useful context, not false precision.
Turn repeated work into assemblies
An assembly is a repeatable group of costs and production assumptions.
A defined shrub-installation assembly might include:
- one shrub;
- expected planting labor;
- amendment;
- a small material-handling assumption;
- defined equipment treatment.
The assembly needs boundaries.
Bad:
Shrub — $125 installed
Better:
A defined shrub size in a prepared bed with normal access, material staged within a stated distance, and explicit included and excluded work.
Now the estimator knows when the assembly applies — and when to return to a full estimate.
Keep cost and price separate in the price book
A useful assembly can show:
- material cost;
- labor-hours;
- labor cost;
- equipment;
- other direct cost;
- overhead recovery under the chosen model;
- selling price under current pricing rules.
If the only stored value is a final price, changes in wages, supplier costs, equipment, overhead, or productivity are harder to update intelligently.
Every price-book entry also needs a date and a definition.
Diagnose the type of estimating failure
When a job underperforms, classify the problem before changing the model.
Useful categories include:
- Quantity error
- Production error
- Cost error
- Scope error
- Overhead error
- Pricing error
- Execution error
An execution problem caused by rework or poor staging should not automatically change an otherwise accurate estimating production rate.
A stale supplier price should not be treated as a labor problem.
The category points to the corrective action.
Build evidence no generic calculator can provide
Over time, estimate-versus-actual records can tell you how your crews, equipment, sites, suppliers, and methods behave.
That is more useful than copying a generic unit price because it is tied to the conditions your company actually encounters.
The loop is simple:
Estimate → Freeze → Perform → Record → Compare → Diagnose → Update
Then feed the lesson back into the next Six-Pass Landscape Estimate.