How the deal score is calculated
Every deal on LEFO Search carries a number from 0 to 100. It is not a gut feeling: it is the sum of six parts, each with a fixed maximum, and the number on the badge is exactly the sum of the six lines you see when you hover over it. This page reads the weights from the very function that applies them to the results, so it cannot describe a formula other than the one in use.
The six parts
| Part | Max | What it measures | How points are earned |
|---|---|---|---|
| Price | 35 | How much the deal costs compared with the median of all deals found for the same search. | All points for costing 30% less than the median or lower; zero for costing 20% more than the median or beyond; in between, proportionally. Exactly at the median: 14 points. |
| Discount | 15 | The discount percentage on the starting price declared by the store. | All points from −50% up; below that, proportionally. A declared discount does not make the price lower than it is: it weighs less than the real price. |
| Rating | 15 | The average product rating, in stars. | Proportional: 5 stars = all points, 4 stars = 80%, no rating = zero. |
| Store | 15 | How reliable the selling store is. | Large chains and marketplaces with buyer protection sit near the maximum; a store we do not know gets 70%; comparison storefronts that bounce you to other sites get much less. |
| Reviews | 10 | How many reviews the product has. | Logarithmic scale relative to the most-reviewed product in the search: 10,000 reviews are not worth 100 times 100 reviews. |
| Condition | 10 | New, refurbished or used. | New: all points. Refurbished: 80%. Used: 60%. Not recognised: 70%. |
| Total | 100 | A deal that is excellent on all six really does reach 100. | |
The price curve, in numbers
It is the part that weighs most, so it deserves a table. Price expressed as a percentage of the median:
| 60% | 70% | 80% | 90% | 100% | 110% | 120% | 130% |
|---|---|---|---|---|---|---|---|
| 35 pts | 35 pts | 28 pts | 21 pts | 14 pts | 7 pts | 0 pts | 0 pts |
The median is the one of the search, not an external price list: if you search for «32 inch smart tv», the median is that of the 32-inch smart TVs found at that moment. That is why the same product can score differently in different searches.
An example computed by the real code
Four deals for the same product, median €199.00. The table is generated on the fly by the function that assigns scores in production — it is not a hand-written example.
| Deal | Price | Price /35 | Discount /15 | Rating /15 | Store /15 | Reviews /10 | Condition /10 | Total |
|---|---|---|---|---|---|---|---|---|
| Offer A costs 35% less than the median, −55% declared, 5 stars, the most reviewed, top-tier store, new |
€129.00 | 35 | 15 | 15 | 15 | 10 | 10 | 100 🔥 |
| Offer B priced exactly at the median, no discount, 4 stars, 40 reviews, unknown store, new |
€199.00 | 14 | 0 | 12 | 11 | 5 | 10 | 52 ◦ |
| Offer C identical to B, but with no rating or reviews |
€199.00 | 14 | 0 | 0 | 11 | 0 | 10 | 35 ◦ |
| Offer D costs 30% more than the median and is used |
€260.00 | 0 | 0 | 12 | 11 | 5 | 6 | 34 ◦ |
Two things to notice. Deal C is identical to B but has no stars or reviews: it loses exactly those two parts and nothing else — no hidden penalty. Deal D costs more than the upper limit of the curve, so it gets zero on price: for a comparison engine, an expensive product is expensive even if the store is excellent.
Badges and result order
- 🔥 Top deal from 80 points up.
- ✅ Good deal from 60 to 79.
- Below 60 the number stays visible, but with no merit label: an average deal is not «a deal», yet the score helps you compare it with the others.
The badge colour follows nine bands, from green to red:
100-91 90-81 80-71 70-61 60-51 50-41 40-21 20-1 0
Results are sorted by descending score by default; ties go to the cheaper deal first. The «Sort» menu switches to price, discount, rating or number of reviews. The «Minimum deal score» filter in the sidebar hides everything below the chosen threshold.
What the score does not do
No store can pay to move up. The affiliate commissions that keep the site running do not enter the formula: the «store» part measures reliability, not who pays us, and a store that pays us nothing can happily sit above Amazon. We state this on every page with affiliate links, and you can check it: the six parts above are the only ones.
The limits, stated plainly
- Ratings and reviews come from the providers' catalogues: if a store does not expose them, that deal starts without those two parts even if the product is excellent.
- The discount is the declared one: that is why it weighs less than the real price. Inflated starting prices exist, and the «price» part is there for that.
- Condition detection is automatic and can occasionally be wrong: always check the store's product page.
- Anomalous prices — below 35% or above 320% of the median — are excluded before the calculation: in most cases they are accessories mistaken for the product, or off-market listings.
- Results stay cached for up to 15 minutes: the price you pay is the one on the seller's page.
Who is behind the site and where the data comes from: about us (Italian). Questions about the method: write to us.