Benchmark Methodology — Supply Index

1. Data collection

All price data on Supply Index originates from real supplier invoices submitted by verified UK aesthetic or dental clinics. We do not scrape supplier websites, accept price list submissions from suppliers, or use list prices as data points.

Clinics submit data via three methods:

📄

Invoice upload

Clinics upload PDF or image invoices. We extract product names, quantities, and prices using OCR with manual review.

⌨️

Manual entry

Clinics manually enter price data via a structured form with supplier, product name, quantity, and net price fields.

🔗

Accounting integration

Coming soon: direct integrations with Xero and QuickBooks for automated invoice extraction.

2. Validation pipeline

Every submitted data point passes through a multi-stage validation pipeline before being included in any benchmark calculation.

Stage 1

Clinic identity verification

Before any data is accepted, the submitting clinic must have a verified account. For aesthetic clinics, we verify against the CQC register. For dental practices, we verify against the GDC register of registered practices. Clinics must also provide a Companies House number or VAT number.

Stage 2

Invoice authenticity checks

Uploaded invoices are checked for: (a) a valid UK supplier VAT number matching known distributors; (b) an invoice date within 60 days of upload; (c) product descriptions that can be matched to our product catalogue. Invoices failing these checks are flagged for review.

Stage 3

Product matching

Each line item is matched to a canonical product in our catalogue using a combination of exact-match, fuzzy-match, and manual review. Products that cannot be confidently matched are placed in a pending queue for manual review by our team. We do not include unmatched data in benchmarks.

Stage 4

Price plausibility checks

Prices are checked against expected ranges derived from our running dataset. Any price more than 3 standard deviations from the current median for that product is flagged for manual review. We check whether the deviation is explained by quantity discounts, unusual packaging (e.g. multi-pack pricing submitted as unit price), or data entry errors.

Stage 5

Duplicate detection

Each invoice is assigned a fingerprint based on supplier VAT number, invoice number, date, and total amount. Duplicate submissions from the same or different clinics are detected and deduplicated. This prevents any single clinic from inflating or skewing benchmarks.

3. Benchmark calculation

Benchmarks are calculated on a rolling basis and updated weekly. We report the following statistics for each product with sufficient data:

Statistic	Description
P10	10th percentile — only 10% of clinics pay less than this
P25	25th percentile — the lower quartile, a realistic "good deal" benchmark
P50 (Median)	The middle value — half of clinics pay more, half pay less
P75	75th percentile — the upper quartile, above this suggests overpaying
P90	90th percentile — only 10% of clinics pay more than this
Mean	Arithmetic mean, shown alongside median for transparency
Sample size	Number of distinct clinic submissions included in the calculation

Minimum data threshold: We require a minimum of 20 verified data points from at least 5 distinct clinics before publishing a benchmark for a product. Below this threshold, products show a "Benchmark pending" status. This prevents small datasets from producing misleading benchmarks.

4. Data weighting and freshness

Not all data points are weighted equally. More recent data is weighted more heavily to ensure benchmarks reflect current market conditions. We apply exponential decay weighting:

Data submitted within 90 days: full weight (1.0)
Data 91–180 days old: 0.8 weight
Data 181–365 days old: 0.6 weight
Data 366–540 days old: 0.3 weight
Data older than 540 days: excluded from calculations

We display the effective sample size (weighted N) alongside each benchmark so users can assess the reliability of the data.

5. Anonymisation and privacy

Individual clinic prices are never disclosed. All benchmarks are aggregated statistics only. Additionally:

We do not publish benchmarks for products with fewer than 5 data points
We do not publish regional breakdowns unless the region has ≥8 data points
Invoice documents are deleted from our servers within 30 days of validation
Extracted price data is stored without any clinic identifiers at the benchmark calculation level
Clinic accounts can request deletion of their contributed data at any time

6. Limitations

We are committed to honest communication about what our benchmarks can and cannot tell you:

⚠Benchmarks reflect prices paid by clinics in our network, which may not be perfectly representative of all UK clinics.

⚠Volume discounts, loyalty agreements, and buying group memberships mean that the correct benchmark for your clinic may differ from the overall population median.

⚠Product specifications can vary (e.g., 1ml vs 2ml packaging, including vs excluding needles). We standardise where possible but users should verify unit comparability.

⚠Benchmarks are informational only and do not constitute commercial advice. Always negotiate directly with your supplier.

Questions about our methodology?

We welcome scrutiny of our approach. If you spot an error or have a methodological concern, please get in touch.