Methodology

How price observations are collected, what the signals mean, and what they don't.

What “observed” means

Every price on MarketSignalIndex is an observation — a price we have seen a real seller ask for a specific part, on a specific date, from a public or contributor-submitted source. We do not model, estimate, or interpolate prices. If we have not observed it, we do not show it.

Observations are collected from two places: verified contributor feeds (resellers who submit their active inventory directly) and public market listings (prices observed from major reseller and secondary market channels). Contributor data is labeled and carries higher confidence. All other observations reflect asking prices — not necessarily what the market actually cleared at.

Data is refreshed monthly for most market classes. Each page shows a Data as of date so you always know how fresh the signal is. In fast-moving markets, check that date before using a signal to benchmark a live quote.

Data sources

MarketSignalIndex aggregates observed prices from three source types, each representing a different layer of the enterprise hardware market:

Verified contributor feeds

Sellers who opt into the Contributor program submit their active inventory directly. Contributor prices reflect live asking prices from known, vetted resellers. These carry the highest confidence level and are updated on a regular cadence agreed with the contributor.

Observed seller prices

For non-contributor sellers (major resellers, distributors, and marketplaces), prices are observed from public listings and recorded against specific part numbers. Each observation is timestamped. These represent asking prices — not necessarily transacted prices.

Secondary market sold transactions

Completed sale data from secondary market channels captures actual cleared prices, not asking prices. This is the most accurate signal for where the market is actually transacting. Secondary market data is sourced via official marketplace APIs and covers the trailing 90 days of sold transactions.

Update cadence

Each canonical market page shows a Data as of timestamp. This reflects the most recent observation date across all listings for that market class.

  • Contributor feeds — refreshed on contributor's agreed schedule (typically weekly to monthly).
  • Observed seller prices — refreshed monthly. Listings older than 90 days are considered stale and excluded from signal computation.
  • Secondary market transactions — trailing 90-day window, refreshed monthly. Each data point reflects an actual completed sale.

Signals degrade with time. A canonical page with a Data as of date older than 60 days should be treated as directional only — not a current quote benchmark. Fast-moving markets (DDR5, AI-relevant SKUs) can shift 20–30% within a quarter.

Signal definitions

All signals are computed from observed price samples for a given canonical market class. Signals are recalculated on each data refresh.

Clearing range
The observed low–high price band across all active listings for a given certification tier (OEM, Original, or Compatible). Represents where sellers are currently asking, weighted toward the middle of the distribution. Use the range as a benchmark when evaluating a quote — a price near the top of the range warrants scrutiny; a price below the bottom may indicate condition issues or a distressed seller.
OEM premium
The percentage difference between the median observed price for OEM-certified parts and the median for Original manufacturer parts in the same market class. A positive value means OEM parts are trading above Original. This premium reflects the value of vendor certification, firmware qualification, and extended warranty support — not a difference in the underlying silicon. Formula: (OEM median − Original median) / Original median × 100
Compatible discount
The percentage by which Compatible (third-party) parts trade below Original manufacturer parts. A higher discount suggests more price-competitive alternatives exist, but also warrants closer attention to compatibility verification for your target platform. Formula: (Original median − Compatible median) / Original median × 100
Spread
The price spread across all observed listings as a percentage of the midpoint price. Measures market volatility for this class. A spread above 40% typically indicates mixed conditions (new vs refurbished), multiple quality tiers trading together, or thin data with outlier prices. Formula: (max − min) / midpoint × 100
Liquidity score
A qualitative rating based on the total number of price samples observed for a market class. More samples mean more confidence that the signal reflects real market depth rather than one or two sellers setting the range.
  • High — 25 or more samples
  • Good — 12–24 samples
  • Medium — 6–11 samples
  • Low — 3–5 samples
  • Insufficient — fewer than 3 samples (signals not computed)

Certification tiers

Every SKU in the database is classified into one of three certification tiers. Signals are computed separately per tier so you can compare apples to apples.

OEM

Parts certified by a server vendor (HPE, Dell, Lenovo, etc.) for use in their specific platforms. OEM parts may carry firmware locks that trigger warnings or refuse to seat in non-certified configurations. They typically command a price premium over equivalent Original parts due to vendor qualification, extended warranty coverage, and compatibility guarantees.

Original

Parts manufactured by the component OEM (Samsung, Micron, Hynix, Intel, Seagate, WD) and sold without server-vendor certification. These are the same underlying components that may be used across multiple platforms and OEM configurations. Original parts generally offer strong compatibility without the OEM premium.

Compatible

Third-party or generic parts that meet the published specification but are not manufactured by the original component maker. Compatible parts are typically the most price-competitive option but require careful compatibility verification — particularly on platforms with OEM lock enforcement. Always verify against your target platform's compatibility list.

Coverage and limitations

What is well-covered

  • DDR4 RDIMM and LRDIMM server memory — active secondary market with multiple sellers
  • Enterprise nearline HDDs (Seagate Exos, WD Ultrastar) — mature, liquid secondary market
  • Intel Xeon Scalable and AMD EPYC processors — coverage growing as OEM variants are added

Known limitations

  • Asking prices vs cleared prices — most observations are asking prices from active listings. Secondary market sold data reflects actual cleared transactions but covers only the trailing 90 days.
  • Condition mix — listings span new, refurbished, and pulled conditions. Signals reflect the observed mix. A wide spread often indicates condition variation rather than market volatility.
  • CPU data depth — CPU canonical pages currently have limited SKU coverage. OEM variants (HPE P-series pulls, Dell AA-series) are being added. Signals on CPU pages should be treated as directional until coverage improves.
  • Quote-only sellers — major distributors (CDW, SHI) do not publish public prices. Their volume is not reflected in these signals.
  • Refresh lag — data is refreshed monthly for most sellers. In fast-moving markets, prices may have moved since the last refresh. Always check the Data as of date before using a signal to benchmark a live quote.

How to use these signals

These signals are designed to support decisions, not replace them. Here is how practitioners typically apply them:

Benchmarking a quote

When you receive a quote from a broker or reseller, compare the offered price against the clearing range for the relevant certification tier. A quote significantly above the range top warrants negotiation or a competing offer. A quote below the range bottom is worth scrutinising — verify condition, warranty terms, and seller reputation.

Reading the OEM premium

If the OEM premium is low (under 10%), Original manufacturer parts offer comparable value without the vendor-lock risk. If the premium is high (30%+), it may reflect genuine scarcity of certified inventory or a specific platform requirement — or simply a thin market with one expensive OEM seller skewing the median.

Interpreting spread

A wide spread does not always mean a volatile market. It often means the observed listings span multiple conditions or quality tiers. Before concluding a market is volatile, check whether the spread is explained by condition mix in the underlying listings.

What signals don't tell you

  • Whether a specific seller is trustworthy
  • The physical condition of a specific unit
  • Warranty terms or remaining life on pulled drives or processors
  • Platform-specific compatibility for your exact server model
  • Real-time prices — always check the Data as of date

Questions or corrections

If you spot a price that looks wrong, a SKU that's missing, or a signal that doesn't match your market experience, we want to know. Email us directly — corrections improve the data for everyone.