What it means
Many business costs are mixed, meaning part of them stays the same regardless of volume and part moves with it. A delivery operation pays for vehicles and depot rent whatever happens, then pays fuel and driver hours in proportion to the number of drops.
The method assumes the relationship between activity and cost is a straight line, so two points are enough to define it. Subtract the low period's cost from the high period's cost, divide by the difference in units, and you have the variable cost per unit.
Once you have the variable rate, the fixed element falls out of either data point. Take the total cost at the high level, subtract the variable cost of those units, and what remains is fixed, and the same calculation at the low level should give the same answer as a check.
The reason people use it is speed and transparency, since anyone in a meeting can follow the arithmetic. The reason accountants are cautious is that it uses only two observations out of possibly dozens, and those two are the extremes, which are exactly the periods most likely to be unusual.
Regression analysis on the full data set is the more accurate alternative and is a standard spreadsheet function. The sensible practice is to use the high-low method for a first estimate or a sanity check, then move to regression when the number is going into a budget, a pricing decision or a break-even analysis.
In practice
Real-world examples
Example 1
A print shop uses the high-low method on twelve months of electricity bills and finds a fixed element of $1,800 a month plus $0.42 per thousand impressions. The owner uses the split to quote confidently on a large job without worrying that energy cost will erode the margin.
Example 2
A logistics manager applies the method to warehouse labour and finds the fixed element is far larger than expected. That prompts a review of shift patterns, because the business was paying for base cover it only needed during peak weeks.
Example 3
A software company splits its cloud hosting bill into a fixed platform charge and a variable per-user cost of $1.85. Sales then uses the variable figure in contribution margin calculations when pricing volume discounts for enterprise customers.
Think of it
Imagine estimating your monthly mobile phone bill by looking only at your busiest month and your quietest month. By comparing the two, you can guess the base plan cost versus your extra data charges.
Formula
Calculation
Variable cost per unit = (cost at highest activity - cost at lowest activity) / (highest activity units - lowest activity units).
Fixed cost = total cost at any activity level - (variable cost per unit x units at that level).
A packing operation records its monthly overhead against boxes packed. The busiest month was 12,000 boxes at a total cost of $86,000. The quietest was 4,000 boxes at a total cost of $46,000.
Variable cost per unit = ($86,000 - $46,000) / (12,000 - 4,000) = $40,000 / 8,000 = $5.00 per box.
Fixed cost using the high month = $86,000 - (12,000 x $5.00) = $86,000 - $60,000 = $26,000.
Checking with the low month = $46,000 - (4,000 x $5.00) = $46,000 - $20,000 = $26,000. The two agree, as they must.
The resulting cost equation is total cost = $26,000 + $5.00 per box. Forecasting a month of 9,000 boxes gives $26,000 + (9,000 x $5.00) = $26,000 + $45,000 = $71,000.Case study
Seen in the real world
Ashgrove Beverages is a fictional drinks bottler used here as an illustrative example. Its production overhead looks unpredictable, ranging from $46,000 to $86,000 a month, and the management team has been treating the whole amount as uncontrollable.
The finance analyst applies the high-low method to twelve months of data, using the busiest month at 12,000 cases and $86,000 and the quietest at 4,000 cases and $46,000. The result is a variable cost of $5.00 per case and a fixed cost of $26,000 a month. Suddenly the picture is clear: at the average volume of 9,000 cases, total overhead should be $71,000, and any month materially above that needs explaining.
Two months later, actual overhead at 9,000 cases comes in at $79,000, some $8,000 above the model. Investigation finds an unbudgeted maintenance contract rather than a fault in the method. In this illustrative outcome, the company keeps the high-low estimate as a monthly control tool but rebuilds the cost line with a full regression once a year, because the two extreme months alone had ignored a gradual rise in the fixed base.
Watch out
Common mistakes
- Picking the months with the highest and lowest cost rather than the highest and lowest activity. The method is driven by the activity measure, and choosing on cost distorts the variable rate.
- Using an abnormal period without adjusting for it. A month disrupted by a strike, a shutdown or a one-off repair will drag the whole cost equation off course.
- Projecting far outside the activity range in the data. A model built between 4,000 and 12,000 units says nothing reliable about 30,000 units, where new equipment or extra shifts would change the fixed base.
Questions
People also ask
Why use only two data points when I have twelve?
Because it is fast and easy to explain, though regression on all twelve is more accurate and should be used for anything important.
Can the method produce a negative fixed cost?
Yes, and when it does the answer is a warning sign that the data is not linear or that one of the extreme periods is distorted.
How often should the split be recalculated?
At least annually, and sooner after any change in pricing, capacity, headcount or supplier contracts that would shift the underlying cost structure.