PriceProof — the newsroom

How we caught our own matcher inventing two products

2026-10-01T00:00:00Z

Every product page on this site exists because award descriptions mentioning that product were found in federal contracting data. The search is a text match against the description field. That works well for a name like "Databricks". It fails badly for a name that is an ordinary English word — and this week we found two pages that it had failed for, and took them down.

What went wrong

Snowplow, the behavioural data platform, shares its name with a snow-clearing blade. The matcher collected awards for snowplows — municipal equipment, highway maintenance, winter operations — and there were plenty of them. The product page was never publishable, because the evidence threshold also requires award descriptions to name the product explicitly and to carry complete fields, and the snowplow awards did not survive contact with that rule. But the match was real, and it would have taken one threshold change to put a page of snow-removal purchases in front of a reader as data-platform pricing.

dbt — dbt Labs' data transformation tool — is three letters that federal agencies use for other things. Dialectical Behavior Treatment. Digital Breast Tomosynthesis. Both appear in contracting records, and both were collected as evidence for a data product. The same threshold rule kept them off the site; the same one change would have surfaced them.

Neither page was ever published. That is the fail-closed rule doing its job — but it is luck, not method. The matcher was capable of a serious error and only an unrelated check was preventing it.

The fix, and why it is general

The tempting repair is to add two special cases: never match "snowplow", never match "dbt". That fixes two products and leaves the class of error untouched. The next ambiguous name — and there will be one, because vendors pick short evocative words — would sail through.

So instead, an alias can be marked as requiring corroboration. A bare ambiguous name no longer matches on its own; it matches only when the same description carries a second signal that places it in the right context. "Snowplow" needs a data-platform word nearby. "dbt" needs one too. Where corroboration is absent, the award is dropped and counted under its own rejection reason — ambiguous_no_context — rather than quietly ignored.

We then re-read every award in the affected files. Where a product's name is genuinely ambiguous in the corpus, the file now shows zero awards instead of a fabricated history:

Snowplow — no awards. It has no page.
dbt — no awards. It has no page.
StarBurst — nine genuine awards, down from 27. The awards that remain name the data platform; the ones removed named satellites, electrodes and confectionery.
Snowflake — the agricultural-lime guards were tightened as well; it keeps 17 awards.

The other half of the problem: awards that name more than one product

The same review turned up a quieter error, going the other way. Some awards name several vendors' products in one line — a reseller's quote for a whole stack. Counting such an award as evidence of one product's price inflates that product with money that paid for something else.

The pipeline now compares an award against every configured product's aliases and classifies anything naming an unrelated vendor as a bundle, publishes it, and excludes it from that product's headline figures. Products that legitimately sit inside larger purchases — a platform sold with its vendor's own suite — carry a curated list of companions instead, because a generic rule flagged innocent phrases like "Teradata on Azure" as bundles and would have thrown away real records.

What this says about the numbers

Nothing about this failure was visible in the published pages. The pages were thin, the threshold held, and the site looked fine. It took reading the underlying award descriptions line by line to see that a product's evidence was about snow removal.

That is the work that a "pricing data" site does not usually show, and it is the reason the methods page exists. Of the 33 products published here, 15 more were configured and refused: two had awards that never named the product, and the rest had fewer than the three distinct awards a page requires — Looker and Fivetran each stopped at two. None of them has a page. When the count of refusals stops being published, the count of pages stops meaning anything.

A contract total is not a price

2026-10-03T00:00:00Z

If you take one thing from this site, take this: the amount in a federal contract award is not the price of the software named in it.

It is a number, it is public, and it is about the product. It is still not a price. Almost every confident claim you have read about "what government pays for X" is built by ignoring that distinction, and it is why those claims disagree with each other so violently.

What the number actually is

An award record says: this agency, on this date, obligated this much money to this recipient for the work described here. That is all. The amount is a contract total — the value of one purchase, for the scope of work the description sets out.

"Scope of work" is doing a lot of work in that sentence. The same platform, bought by two agencies in the same month, can appear as:

a licit per-seat subscription for a 200-person team;
an enterprise-wide agreement covering every employee of a department;
a three-year renewal of an entitlement bought before the current contracting officer arrived;
a bundle where the platform is one line among several vendors' products;
an implementation project in which the licences are almost incidental to the labour;
a modification that adds funding to a contract signed two years ago without buying anything new at all.

Six different things, six wildly different amounts, one product name. Average them and you get a number that describes none of them. Publish that average as "the government price" and you have published a fiction with a citation attached, which is worse than publishing nothing.

How to tell which kind of number you are looking at

This is where a site either does the work or pretends the problem away. Our method is to label every figure with what it is, and to refuse to compare across labels:

contract_total — the value of one award. What most of this site's records are.
unit_price — a price per unit stated in the award text, e.g. (10) @ $2013.00 EA. Quoted, not computed.
derived_unit_price — no rate is stated, but the text states a quantity, so the total is divided by it. We show the quantity we used and call the result derived, because it depends on our reading of the text.
list_price — the vendor's own published price. Not what anyone paid. Comparable only with other list prices.

The fourth is the one people get wrong most often. A vendor's list price is a starting position in a negotiation, and public agencies rarely pay it — they buy through schedule contracts and reseller agreements at a discount that is itself commercially confidential. So when you see a page that sets a list price next to a contract total and invites you to compare them, you are looking at two numbers that were never meant to meet.

The test to apply to any pricing claim

Ask three questions of any figure you are shown about public-sector software:

What kind of number is it? If the answer is not stated, assume the worst.
What was bought? If the award text says "software support renewal, per attached", the amount is not evidence about a licence, and no amount of arithmetic will make it one.
How many records is it drawn from? One award is an anecdote. Three is a floor. A band with a count next to it, and the scope of the records it was computed from, is the only shape in which this information is honestly reportable.

Everything on this site is built to survive those three questions. Where the evidence cannot survive them, we do not publish a figure — and where it can survive them only partly, we say so in the sentence that carries the number.

The federal data-platform market, in 2,368 records

2026-10-05T00:00:00Z

There are 33 data-platform products on this site, and behind them sit 2,368 public contract awards worth $2,116,618,029. Every one of those records is a real thing that happened: an agency, a date, an amount, a vendor or reseller that received it, and a link to USAspending.gov where you can read it yourself.

We have now read all 2,368 of them. This is what they say.

Most of the money is licences, and most of the rest is a mystery

Classify each award by what its description says it bought, and the picture is less messy than the raw totals suggest. Licences account for 42 percent of awards, renewals for 25 percent. Together they carry $1,140,135,172 of the $2,116,618,029 — a little over half the money, on two-thirds of the records.

Then there is the remainder, and it is worth being blunt about it. Support-only awards are 10 percent of records, services 3 percent, training 2 percent, and genuine multi-product bundles another 2 percent. Each of those is a real purchase, but none of them is a price for the product it names. An award that buys three years of implementation labour and happens to mention the platform is not evidence of what the platform costs, and any site that averages it in is quietly wrong.

The largest single group after licences is also the least satisfying: 16 percent of awards — 369 of them — cannot be classified at all from their text. They say things like "software support renewal, per attached" and nothing else. We publish them, label them unknown, and leave them out of the headline figures, because the alternative is guessing, and a guess dressed as a price is the thing this site exists to avoid.

A handful of names receive most of it

Awarding agencies are what you would expect. The Department of Defense accounts for 712 of the 2,368 awards, Homeland Security 276, Health and Human Services 219, Transportation 174, Commerce 120. Forty-two agencies in total buy from these 33 products.

The recipients are more interesting, because they are mostly not the vendors. Carahsoft Technology Corp appears on 192 awards, New Tech Solutions on 141, Software Information Resource Corp on 119, Executive Information Systems on 106, GovSmart on 103, CDW Government on 94. Read the list and you are reading how federal software is actually bought: not from Salesforce or Databricks directly, but through a small number of resellers holding a schedule contract that an agency can order against.

That has a practical consequence for anyone trying to work out what software costs. The vendor publishes a list price; the reseller sells at a discount off it; the agency records the discounted total. The number in the public record is the third of those, and it is the only one that is public.

The largest files are the largest buyers, not the biggest prices

Ranked by total obligated value across licence and renewal awards, Tableau leads at $196,359,648, then Palantir at $182,043,748, Informatica at $110,922,920, Confluent at $105,402,245 and TIBCO at $79,472,559. Databricks, Elastic and MicroStrategy follow.

It is tempting to read that as a price ranking. It is not. It is a buying-volume ranking. A product at the top of that list is one that many agencies have bought many times, which is a statement about how widely it is deployed and how long it has been sold to government — not about what a licence costs. Tableau's file alone holds 738 awards; Palantir's holds 71.

The number that should worry a buyer

The single most useful figure on this site is not a price. It is the ratio between the smallest and largest award for the same product.

Tableau's records run from $397 to $13,047,776 — a spread of more than 32,000 times. TIBCO runs from $9,783 to $26,400,272. Microsoft Power BI from $404 to $734,755. SAS from $3,119 to $5,338,670. Informatica from $6,916 to $11,083,145.

None of those pairs is a contradiction. The small award is usually one team buying one thing; the large one is an enterprise-wide deployment with services attached, often recorded as a single line. But that is exactly the problem: a buyer who searches for "what does this cost" and finds one number will find the wrong one. The honest answer is a band, and this site now prints one at the top of every product page — computed only from licence and renewal awards, with the range and the count next to it so you can see how thin the evidence is.

What we can and cannot tell you

We can tell you, for 33 products, what public agencies have been awarded, when, from whom, and what the award text says was bought. We can tell you how that has moved, and we can hand you the whole dataset as CSV.

We cannot tell you what any of it costs per seat. Federal award text states a total far more often than it states a rate; of the 2,368 records here, only a minority state a per-unit price at all. Where an award does state one, we quote it. Where the text states a total and a quantity, we can divide — and we label the result as derived, with the quantity we used. Where neither is available, we say nothing, which is the correct answer more often than vendors and analysts admit.

Twelve of the 33 products have a published list price from the vendor. The other 21 are quote-only, so for those the only public price evidence in existence is the award record — and that is precisely why it is worth reading.

Department of Commerce Awards ARCHITECHTURE SOLUTIONS LLC $1.88M for Alteryx

2026-10-05T00:00:00Z

The Department of Commerce, specifically through its U.S. Patent and Trademark Office, awarded ARCHITECHTURE SOLUTIONS LLC a contract for Alteryx on September 11, 2020. This delivery order, classified under "Information Technology Software," involved an obligation of $1,882,075.01. The total potential value for this work, including the base and all options, was set at a ceiling of $1,976,649.85. This distinction between the obligated amount and the ceiling is important for buyers, as it shows the immediate financial commitment versus the maximum possible expenditure for the defined scope of work. The period of performance for this contract spans from September 11, 2020, to September 10, 2023, indicating a multi-year engagement for the Alteryx platform and its associated services.

The contract was procured under a "full and open competition after exclusion of sources" process, which resulted in three offers being received. This competitive environment suggests that the agency sought a range of solutions before making its selection, potentially driving better value. The award was placed against a Government-Wide Acquisition Contract (GWAC), specifically NNG15SD42B, which is a multiple-award vehicle held by the National Aeronautics and Space Administration. Utilizing a GWAC allows agencies to streamline procurement by leveraging pre-negotiated terms and conditions, potentially reducing acquisition lead times and administrative burdens. For professional buyers, understanding the use of GWACs like this one is crucial, as they represent established channels for acquiring technology and services across the government.

ARCHITECHTURE SOLUTIONS LLC, the recipient of this award, is recorded with several business designations, including Asian Pacific American owned business, limited liability corporation, minority owned business, partnership or limited liability partnership, self-certified small disadvantaged business, service disabled veteran owned business, small business, and U.S.-owned business. These classifications indicate that the company falls into several socio-economic categories often prioritized in federal contracting, which can influence procurement decisions and set-aside opportunities. This particular award contributes to ARCHITECHTURE SOLUTIONS LLC's broader federal portfolio, as the company holds 342 federal transactions totaling $42,028,159.73. This volume of past federal work suggests an established presence and experience in government contracting.

Product and Agency Context

Alteryx offers a platform designed to unify data, analytics, and AI workflows. Its capabilities include automating data-intensive business processes and scaling trusted insights and decision intelligence. Specific features include Data Prep & Blend, which helps transform spreadsheets into governed and repeatable workflows, and Agentic Analytics, which supports building AI agents and extending to preferred AI platforms. Alteryx also integrates with cloud data platforms like Databricks, Snowflake, and Google, allowing for in-place data processing with push-down capabilities. These features are relevant for agencies looking to modernize their data infrastructure and leverage advanced analytics.

Alteryx is utilized across various departments such as Data Analytics & BI, Finance, IT, and Supply Chain & Operations, and serves industries including Energy & Utilities, Financial Services, Higher Education, Manufacturing, Public Sector, and Retail & Consumer Goods. Its use cases range from accelerating operations, such as Month-End Close Automation and Reconciliation, to accelerating decisions through Demand Forecasting and Management Reporting. The platform emphasizes automated reporting, enterprise workflow automation, and self-service analytics workflows, alongside providing trusted logic for AI agents and conversational analytics. For government agencies, these capabilities can support data-driven decision-making and operational efficiency.

The Department of Commerce's mission is to create conditions for economic growth and opportunity. The U.S. Patent and Trademark Office, a sub-agency, operates within this broader mission. The acquisition of Alteryx by the U.S. Patent and Trademark Office suggests an application of these data and AI capabilities to support the agency's specific functions, potentially in areas requiring extensive data analysis for patent and trademark examination or administration. The work for this Alteryx contract is recorded as being performed in Fairfax, VA, indicating the geographical location of the service delivery or primary user base for this specific award.

PriceProof — the newsroom

How we caught our own matcher inventing two products

What went wrong

The fix, and why it is general

The other half of the problem: awards that name more than one product

What this says about the numbers

A contract total is not a price

What the number actually is

How to tell which kind of number you are looking at

The test to apply to any pricing claim

The federal data-platform market, in 2,368 records

Most of the money is licences, and most of the rest is a mystery

A handful of names receive most of it

The largest files are the largest buyers, not the biggest prices

The number that should worry a buyer

What we can and cannot tell you

Department of Commerce Awards ARCHITECHTURE SOLUTIONS LLC $1.88M for Alteryx

Product and Agency Context

Sources