The pricing meeting before a submission deadline is usually two arguments wearing one agenda. One side wants margin: we did the costing, we deserve 25 percent. The other side wants the win: cut it, they will go with the cheapest anyway. Both are guessing, because neither has looked at what happened the last twenty times this buyer bought this thing.
That history compresses into three numbers, and together they answer the three questions that decide a price: how crowded is the field, how aggressive do winners run, and what does this buyer historically pay. The three terms, and percentile, are defined in the glossary.
01Number one: bidder density
Bidder density is the count of bidders per tender in your category, read across the last few years of comparable awards. It is the single best proxy for how hard a win will be. In a field of six broadly similar bidders, a random one wins about one time in six. You are not random, but the baseline sets expectations, and your win rate, pricing latitude and cost of sale all move with it.
Density also tells you what game you are in. Thin fields, one to three bidders, usually mean high barriers: pre-qualification, niche capability, or an incumbent everyone else has stopped challenging. Price matters less there; eligibility and relationships matter more. Crowded fields, eight bidders and up, are commodity territory where the spread compresses and the cheapest compliant offer usually takes it.
02Number two: the winner versus median spread
Open a single bid table and sort by amount. The median bid is what a typical competitor thought the work was worth. The winning bid sits some distance below it. That distance, expressed as a percentage and averaged across the category, is the winner versus median spread: how aggressive winning actually is around here.
If winners in your category typically land 8 to 12 percent below the median bid, that is a pricing instruction: bidding at the pack median is a plan to lose politely. A category where winners sit only 2 percent under median is telling you bids cluster tightly and technical scores do more of the deciding.
The spread has a second use: it converts award values into bid expectations. If winners run about 10 percent under the median bid, a category with a median award of 2.5M implies a typical pack median near 2.8M. You will need that inversion below.
03Number three: award-value percentiles
Line up the awarded values of comparable tenders and take the 25th percentile, the median, and the 75th percentile. That band is what the buyer historically pays for this kind of work, and it is the sanity check on your absolute number.
A price above the 75th percentile needs a story: bigger scope, longer term, or a genuinely premium offer. A price below the 25th percentile deserves suspicion of your own costing before congratulations. The band only works on comparable work, so normalise for contract duration and scope before you trust any percentile.
The chain, end to end
Pricing a three-year cleaning contract, amounts in USD
A ministry tenders a three-year office cleaning contract. The category history, 24 comparable awards with published bid tables over three years, gives you the three numbers.
Density: the typical tender drew 6 to 8 bidders. Crowded and price-driven, but not a stampede.
Spread: across those bid tables, winners ran about 10 percent below the median bid.
Percentiles: awards came in at 2.0M (25th), 2.5M (median), and 3.2M (75th).
Your cost model says delivery costs 2.1M over the term. Now chain the numbers. If this tender draws a typical field, the median award of 2.5M implies a pack median near 2.5M divided by 0.90, about 2.8M, and a winning bid near 2.5M.
So a bid of 2.45M puts you where winners historically land. Margin check: 2.45M minus 2.1M is 0.35M, about 14 percent of the bid. Percentile check: 2.45M sits between the 25th percentile and the median of what this buyer pays, so the price will not startle anyone.
The decision is now a one-liner. If your margin floor is 12 percent, bid 2.45M with a straight face. If your floor is 18 percent, the winning zone is below your floor, and the right answer is no bid. Walking away from a tender you were always going to lose at your price is the cheapest outcome available.
Density set the competitive frame, the spread converted award history into a target price, and the percentiles sanity-checked the absolute level. None of them replaced your cost model; they told you whether your cost model and the market can meet.
Reading benchmarks well
- Sample size. Twenty-four awards over three years is eight a year, and a percentile on that moves when three unusual awards enter the window. Ask how many observations sit behind a figure before you lean on it.
- Category drift. The 2023 cleaning contract is not the 2026 cleaning contract: scopes grow, terms stretch, indexation clauses appear. Normalise to annual value and comparable scope.
- Survivorship. Award statistics only contain tenders that were awarded. Cancelled and re-tendered procurements are often exactly the ones where bids came in far from the buyer's budget.
Doing this by hand means downloading bid tables one tender at a time and maintaining the spreadsheet forever. Ishara's bid benchmarks compute the bidder-count histogram, the winner versus median spread and the award-value percentiles from published bid tables, scoped to any buyer or to the whole market. The price a government bid use case walks the same chain on the product's screens.
You can now price a bid from history. The last lesson in the course zooms out from one tender to all of them: turning the flow of opportunities into a managed funnel with gates, stages, and coverage math.