Two quotes can name the same item and still describe different purchases.
One includes delivery. One does not say. One includes setup. Another lists setup as optional. A single “total” column can hide those differences.
Before asking AI which quote is better, ask it to normalize only what the documents actually state.
Use these columns: supplier, scope, base price, included charges, excluded charges, unknown charges, timing, assumptions, and source location.
In a fictional comparison, Quote A is $800 with delivery included. Quote B is $750 with delivery unspecified. The supported conclusion is that B has the lower stated base price. The delivered cost comparison is unresolved.
Try:
Compare the supplied quotes using only their stated terms. Cite the relevant section or phrase for each entry. Keep “not stated” separate from “included” and from zero. Do not rank total cost unless the scopes and required charges are comparable. List the smallest set of questions needed to make the quotes comparable.
Check the table against the documents, especially units, quantities, taxes, dates, and optional items. If AI performs arithmetic, recalculate it independently. A correct calculation on mismatched scope is still misleading.
Then send the missing questions through your normal purchasing process. The useful output may be three questions, not a winner.
For a purchase with ongoing costs, include the comparison period explicitly. “Cheaper this month” and “cheaper over a year” answer different questions.
This week, take an old quote comparison and circle every blank cell. Which blanks were treated as zero? That is where to start improving the next comparison.
Try a free sample
Want a reusable starting point? Open the SOP Starter Pack and copy the free worksheet at the top. The paid section adds guidance and worked examples to help you use it.
Become a paid reader for the complete pack archive and two new implementation packs each month.
—Future Proof