The Software Price Index

The VendorBenchmark Software Price Index, versions 1.0 and 1.1

The index publishes for twenty flagship vendors and only those. The universe is our calibration gate: a vendor enters when its reference dataset is dense and defensible enough that we would stand behind the numbers in print, and not before. An uncalibrated vendor has no index page and the API returns not found rather than a guess.

1. Verified deals. Anonymized observations contributed by customer organizations through the VendorBenchmark outcome network, with the contributing organization's consent. Each contribution carries the vendor, the canonical metric, a coarse company size band, and nothing that identifies the organization.
2. Curated anchors. Reference agreements assembled and maintained by our analysts: per-vendor datasets of anonymized closed deals with signing date, deal size, negotiated discount, and net fee, the same reference book behind our in-app benchmarking tools.
3. Public list anchors. Published list prices for flagship SKUs, refreshed from vendor price lists. Where a curated dataset records discounts rather than per-unit fees, the paid figure translates the observed discount distribution through the flagship SKU's list price, and the page says exactly which anchor was used.

Each cell of the index covers one vendor, one seat band, and one quarter. Verified and curated observations pool into one distribution per cell; a verified observation counts once per contributing organization per window (one organization, one voice), so no single contributor can move a published figure by volume. Every published figure is stamped with both counts, in the form N verified deals, M curated anchors.

Verified observations enter a cell only when at least five distinct organizations contribute to it in the window, the same k-anonymity floor the outcome network enforces everywhere else. Below five, the verified pool is excluded entirely rather than thinned. The index publishes aggregates only: medians and quartiles, never a raw observation, and nothing traceable to a contributor. Curated reference deals are anonymized at source before they enter the book.

A cell publishes only when its pool clears a density floor: at least ten curated observations or at least five distinct verified contributors in the window. A cell that clears neither is not published, the page shows insufficient data in that cell, and the API omits it. We do not interpolate, extrapolate, or borrow from neighboring cells. Admitting the gaps is what makes the filled cells credible.

The index stamps once per calendar quarter, covering complete quarters only. Each stamp aggregates the trailing four quarters of observations ending at that quarter, a rolling annual window that smooths thin quarters without hiding movement, the standard treatment for transaction-based indices. Curated deals date by signing date; verified observations date by the quarter they were contributed. The direction of travel compares a cell's median against the same cell one quarter earlier: rising and falling mean more than one percent of movement, flat means within it.

1. Median paid price. The median net fee per unit per year observed in the cell (for perpetual license datasets, per unit one time). For discount-based datasets this is the discount distribution translated through the named flagship list anchor.
2. The paid corridor. The 25th to 75th percentile of paid prices, the gold band on the charts.
3. The discount corridor. The 25th to 75th percentile of negotiated discounts off list, with the median.
4. Renewal uplift cap. The median contracted renewal uplift cap, what the paper allows, from verified network observations only. This measure fills more slowly than price; its cells publish when the network clears the k-floor and say insufficient data until then.
5. Realized renewal uplift (version 1.1). What renewals actually landed at versus the prior term, as a 25th to 75th percentile band with the median, from the outcome network's verified observations blended with our curated anchor calibration. Defined below.
6. Direction of travel. Rising, falling, or flat versus the prior quarter, from the cell's own history.

Version 1.1 adds one measure and changes nothing else. Where the contracted uplift cap records what the paper allows, the realized renewal uplift records what the renewal actually landed at: the percentage change of the new annual fee against the prior term, per year, negative when the buyer paid less. The source is the VendorBenchmark outcome network: customer organizations that consent to contribute have their confirmed negotiation outcomes anonymized at the source to a vendor family, a value rounded to one decimal of a percent, a coarse deal-size band, and a calendar quarter, keyed by salted one-way hashes with no organization identifier stored.

The same honesty rules apply, and one is stricter. One organization, one voice: a contributor's latest observation in the window is the only one that counts. The k-anonymity floor holds at five distinct contributors per cell, and below it the cell publishes nothing at all; unlike the paid measures, the realized uplift measure has no curated fallback pool, because a curated estimate of realized behavior is an opinion, not an observation. Above the floor, the verified distribution is blended with our hand-calibrated renewal-uplift anchor for the vendor family: observations are freshness weighted with a twelve month half life, the anchor enters as a capped pseudo-count so fresh verified deals can always move the band, and every published figure carries both counts.

Version 1.1 publishes alongside version 1.0, which continues to stamp quarterly and is never restated. The new version's first rows appear in the first quarter any flagship vendor cell clears the contributor floor; until that day the index serves version 1.0, and this page is the announcement of record for what 1.1 will publish. There is no partial state: a version either has published rows or it does not exist.

The snapshot schema is a contract. Published cells are append-only: once a figure is published for a version, it is never updated, deleted, or silently restated. A methodology change ships as a new index version published alongside the old one, so a contract clause or article that cites version 1.0 keeps meaning what it meant on the day it was written. Contract language that references the index, for example a repricing rider, should cite the index by name and version.

Cite the index by name, version, vendor, and quarter: VendorBenchmark Software Price Index v1.0, Workday, 2026Q2, vendorbenchmark.com/price-index/workday. The version to cite is the one printed under the figure you read; every page and API response states it. For press inquiries about the underlying pools, write to the monitored inbox on our contact page; we can confirm counts and methodology but never a contributor.

The public endpoint serves each vendor's published series and the latest quarterly summary, rate limited and CORS enabled, at https://app.vendorbenchmark.com/api/public/price-index?vendor=workday. The keyed tier at /api/v1/price-index returns every vendor, seat band, and quarter in one machine-readable call for portfolio models and diligence work; it authenticates with a VendorBenchmark API key (created under Settings, API) and is part of the Professional and Enterprise plans.