Rank stores for lending risk

For automated lending decisions, drop in one file of applications and see which stores to review first. The tool runs three checks on every store and ranks them by how many flag it.

Applications file

Drop a CSV file here or pick one

33,159 applications from 30 stores were read.

The file is read in this browser and never uploaded. If you email the result, only counts are sent.

Store or unit

Application date

Good or bad

Loan amount

Goods price

These checks run with the columns picked: Bad rate, Volume, Above price, and each check needs its columns.

Warning multiple, ×

Above-price tolerance, %

Minimum applications

Confidence, %

Each check tests every store. When more stores are tested, a store must stand out more to be flagged, so higher confidence flags fewer.

These are generated demo applications from 30 stores over two years, with a store planted for each check and one store that trips all three.

Stores flagged

All three1

Two checks0

One check3

4 stores are flagged by at least one check at 95% confidence, with a flag threshold of 1.5× against the book’s rate or their peers’ growth. 1 are flagged by two or more, so look at those first, and treat each flag as a reason to look, not proof.

Flagged store points

Bad rate Volume Above price

Store ranking

Rank	Store	Applications	Bad rate	Bad lift	Volume	Above price	Price lift	Points
1	S12	888	14.1%	2.5×	3.1×	26.4%	6.0×	3
2	S09	846	7.1%	1.3×	0.6×	33.8%	7.7×	1
3	S17	1,200	6.1%	1.1×	3.6×	2.2%	0.5×	1
4	S07	1,066	13.7%	2.4×	1.1×	3.0%	0.7×	1
5	S24	1,576	4.6%	0.8×	1.0×	3.3%	0.8×	0
6	S30	1,544	5.4%	1.0×	0.9×	3.2%	0.7×	0
7	S22	1,476	4.1%	0.7×	0.7×	3.0%	0.7×	0
8	S15	1,471	5.0%	0.9×	1.0×	3.5%	0.8×	0
9	S16	1,455	5.8%	1.0×	0.8×	3.5%	0.8×	0
10	S14	1,399	5.3%	0.9×	1.4×	2.9%	0.7×	0
11	S02	1,397	4.4%	0.8×	0.9×	2.7%	0.6×	0
12	S28	1,371	5.4%	1.0×	0.8×	2.8%	0.6×	0
13	S29	1,345	4.6%	0.8×	1.0×	3.0%	0.7×	0
14	S13	1,331	4.4%	0.8×	0.8×	3.5%	0.8×	0
15	S27	1,233	6.0%	1.1×	0.8×	2.9%	0.7×	0
16	S06	1,228	5.0%	0.9×	1.1×	3.1%	0.7×	0
17	S01	1,136	4.8%	0.9×	1.5×	2.6%	0.6×	0
18	S19	1,049	4.4%	0.8×	0.9×	3.0%	0.7×	0
19	S26	1,026	5.1%	0.9×	1.2×	2.8%	0.6×	0
20	S20	1,010	5.9%	1.1×	1.0×	3.2%	0.7×	0
21	S08	985	4.6%	0.8×	1.4×	2.3%	0.5×	0
22	S11	927	5.4%	1.0×	0.8×	2.7%	0.6×	0
23	S23	901	5.1%	0.9×	1.4×	2.4%	0.6×	0
24	S10	840	5.0%	0.9×	0.8×	2.0%	0.5×	0
25	S05	816	5.3%	0.9×	1.3×	3.2%	0.7×	0

5 more stores are in the download.

Get the result by email

Run the same checks on every application when the decision is made, then add their points into one store score. Start free

Checks and points

Bad lift = Bad rate ÷ Book bad rate
Volume = (Latest month ÷ Earlier month) ÷ Peer growth
Above price = Loan amount ≥ Goods price × (1 + Above-price tolerance)
Price lift = Above price ÷ Book above price
Flag rule = p ≤ (1 − Confidence) ÷ Stores tested  &  Lift or index ≥ Warning multiple
Points = Bad rate + Volume + Above price

Each check tests every store against the rest of the book or its peers, with confidence shared across the stores it tests. A store new in the latest month counts as a volume flag, and each flag adds one point, so points show how many checks point to a store, not how certain any one check is.

How store risk ranking works

A store risk ranking brings checks together to show which stores stand out for review. The tool compares each store's bad rate with the rest of the book, then checks volume against peers and loans above price.

A store flagged by more than one check gives you a reason to look at it first. Points show how many checks flag each store, and they set a reading order rather than a verdict.

Common questions

How do I rank stores by fraud risk?

The tool compares each store's bad rate with the rest of the book. It also checks volume against peers and loans above price, then ranks stores by how many checks flag them.

Which checks can rank a dealer by risk?

The tool checks a store's bad rate against the rest of the book. It also compares volume with peers and measures loans above price.

How is a merchant risk score worked out?

A store gets one point for each check that flags it. The ranking puts stores with more points first, then uses the lifts or indexes of flagged checks and store size.

Why review a store flagged by several checks?

Several checks flagging the same store give you more reason to look at it first. A flag is a reason to look, not proof.

How can I choose which stores to review first?

Start with the stores that have the most points, since each point represents a check that flagged the store. The points set a reading order, not a verdict.