Work record / Stories

Building the pricing model instead of trusting the gut

LIEF was pricing bid by bid on judgment, so I had the team build a model from every bid we had actually run.

LIEF kept bidding with a manufacturer of a patented prefab wall system, and that relationship had become a steady pipeline. Yet each bid was still priced separately on judgment, with no model linking labor cost to material cost across the whole book.

I told my team to stop pricing new bids by feel and to build a model from the bids the group had actually run.

Here is why. With enough real transactions, the link between two cost variables, labor and material in this case, can be measured rather than guessed. A model drawn from real bids removes guesswork, and it flags a bid that looks off before it leaves the building.

Fourteen live bids showed labor fitting material at an R-squared of 0.999. A bid that lands well off that line is now a cue to recheck it before sending. As the model matured, the pricing margin of error tightened, plus or minus 14 percent at first and plus or minus 6 percent later.

A person still decides what to do when a bid leaves the expected line. The model shrinks the area where judgment is needed and leaves judgment in place. After enough real transactions, price the next job from the pattern in your own data, not from memory.

Story details

All stories · Work record