Reconstructing a Competitor's Hidden Price From Public Evidence

You need to know what a competitor charges and how they package it, but their site says "contact sales" and nothing else. This guide is for strategy, research, and talent-intelligence teams who have to produce a real answer anyway. It carries one hidden-price competitor all the way through - the archived pages, the procurement awards, the review mentions, the arithmetic that reconciles them, and the dead ends - so you can produce a defensible price range and a tier-and-packaging map with a confidence label on every figure.

Most competitor teardowns stop at reviews, partners, or roadmap. Pricing gets skipped, and the public best answers end at "use a tool." That gap is not an accident. The skill to reconstruct hidden prices lives in a different seat than the one usually asked to do it.

Why hidden pricing is a reconstruction problem, not a lookup

A "contact sales" page is not the absence of a price. It is a price that has been moved off the public web and split across archives, procurement records, and buyer reports. Your job is to pull those fragments back together into one defended range.

The reason this is hard is structural. Vendors moved to multi-tier, module-bundled architectures, so a single seat became a variable construct and the list price became a starting point rather than an anchor. Any one archived number is nearly meaningless on its own. It only becomes useful once you layer on the discount band the category actually pays, and once you know which packaging tier it belonged to.

There is a second reason teardowns skip pricing: the people who can do it are rarely the people asked. In Refolk's index of professional profiles, the US shows about 31 times more pricing analysts and managers than competitive-intelligence analysts. Reconstruction expertise sits in pricing and finance functions, not the CI teams usually handed a competitor brief.

So treat this as an evidence problem with a method, not a search you either win or lose. The output is not a number. It is a range, a packaging map, and a confidence label you can defend line by line.

The public sources that reveal hidden prices, and what each proves

Six source types recur, and each proves something different while misleading in its own way. The table below is the map; the paragraphs after it tell you where each one lies.

Archived pricing pages show list prices from before a vendor hid them. The catch is coverage. Crawlers prioritize popular sites, so a competitor's pricing page may be captured every few months, or never, and JavaScript-rendered price tables often archive as blank. Absence of a number is not evidence a price was removed. Memento Time Travel federates more than 20 archives to return the snapshot closest to a target date, which widens your odds.

Government procurement portals expose actual prices paid. USAspending tracks every prime federal award above the $10,000 micro-purchase threshold, and each award record carries around 40 fields including award amount, potential value, period of performance, awarding agency, recipient, NAICS, and PSC codes. Coverage runs from FY2008, with custom award data back to FY2001. What it misleads on: an award is one negotiated deal, not list, and public-buyer terms differ from commercial ones.

GSA schedules publish list-level ceiling prices through GSA Advantage and eLibrary, plus historical price-paid information. These are ceiling or public-sector-specific rates, not commercial net. Treating them as "what everyone pays" overstates.

EU TED award notices disclose winners, contract values, and award dates, searchable by CPV code, NUTS region, and procedure type. TED publishes over 700,000 notices a year worth more than €700 billion, with directive thresholds around €140,000 for services. There is a CSV subset covering who bought what from whom, for how much.

Review sites and marketplaces - G2, Capterra, TrustRadius, AWS Marketplace, and buyer-guide benchmarks drawn from billions in real transactions - carry buyer-reported seat and total figures. They are self-reported and often stale, sometimes predating a repackaging.

A worked teardown: from "contact sales" to a defended range

Follow one product through. The scenario is 100 seats, one region, a three-year term. Freeze that first; every figure below maps to it.

The list anchor. The Wayback Machine returns a pricing snapshot from before the vendor moved to contact-sales. Two adjacent snapshots render blank - the JS table archived empty - but a third holds real numbers in the DOM: a mid-tier edition listed at $150 per seat per month. At 100 seats over 36 months that is $540,000 list for the scenario. Confidence on the anchor alone: medium, one source, one dated snapshot.

The procurement points. USAspending returns a three-year federal award to the vendor at an effective $118 per seat per month for a comparable seat count. TED shows an EU award at a similar effective rate. Both are net, both public sector. These bracket the high end of commercial net, because public buyers pay ceiling-schedule terms.

The buyer-reported points. Two G2 mentions quote $95 and $102 per seat per month at roughly 100-seat deals, both dated after the last repackaging. A buyer-guide benchmark cites $88 for a similar enterprise scenario. These sit below the procurement figures, as expected for commercial net.

Now the arithmetic. The list anchor is $150 per seat. The category discount band for this vendor class is 29 to 42 percent off list, so net brackets to $87 to $107 per seat. That band overlaps the commercial buyer-reported cluster ($88 to $102) and sits below the public-sector figures ($118 and up), exactly as the bias predicts. Three independent source types - archived list plus discount math, commercial reviews, and public-sector awards - converge on a commercial net range of roughly $90 to $105 per seat per month for the mid tier at this scale.

That is the mid tier. The divergence between commercial ($90 to $105) and public-sector ($118+) is not noise to average away. It is the second data point: it tells you the vendor holds firmer terms with public buyers and discounts commercial enterprise deals more deeply. Keep both, labeled.

The dead ends I hit

Two forks went nowhere, and both are worth naming because you will hit them too.

The first: the archived pricing page looked captured but rendered blank on the first two tries. Reading "no price shown" as "price removed" would have been a false positive. Opening adjacent snapshots and confirming the DOM actually held numbers rescued the anchor.

The second: an early single TED award came in far higher than everything else, and it was tempting to treat it as the list ceiling. It was one negotiated public-sector outcome, not list. Generalizing it would have inflated the whole range. Requiring three or more independent points before fixing any tier is what caught it.

A single award is a negotiated outcome, not a list price, and generalizing from one deal inflates the entire range.

Normalizing across per-seat, tiered, and consumption models

Before any comparison, convert every vendor to the same monthly or annual cost for your frozen scenario. Comparing headline monthly prices is the single most common mistake, and it can distort real cost by three to five times.

Per-seat is the most straightforward model to benchmark, because the unit is clear: price per seat per year at a comparable feature tier, with total seat count as the most impactful driver of per-unit price. Tiered and consumption models take more work - you have to model the scenario's actual usage and read the cost off the vendor's curve at that point.

Then correct for utilization, the hidden multiplier. In real deployments, 20 to 40 percent of seat licenses go unused. Divide total annual spend by total licensed users, then compare to active users: if you pay for 500 licenses but only 350 log in monthly, your effective cost per user is 43 percent higher than quoted. Quoted per-seat prices understate true unit economics, so apply an active-user haircut wherever you know utilization.

This is also where the reconstruction most often stalls, and it is worth being honest about it. The binding constraint is usually not analysis but archive coverage. Niche pricing pages may be captured every few months or never, and JS tables archive blank, so the work can fail at the archive step rather than the final reconciliation. That is exactly why procurement-first practitioners invert the order and lead with hard award data, treating archives as corroboration.

Estimating discount depth from list to net

Anchor on a list figure, then apply the published category discount band to bracket net. Enterprise SaaS discounts average 29 to 42 percent off list, and the vendor-specific bands below are the reconciling multipliers.

Discounts stack rather than apply as one flat number. A typical stack is 15 to 25 percent volume by seat count, 5 to 10 percent term for a three-year versus one-year commit, plus a strategic-account layer. If you apply a 40 percent "total" discount and then add a separate volume and term stack, you double-discount and understate net.

Context on realism: analysis of a large SaaS proposal population found discounts of 1 to 20 percent produced the best outcomes, while steeper discounts over 40 percent led to smaller deals and slower closings. So a competitor quoting deep discounts is a signal about deal size and urgency, not just price. And startup-program rates - 30 to 80 percent off list for the first 12 to 24 months before they reset - are your low bound, not your typical.

Refolk removes the friction the last two sections describe: the reconstruction is only as good as the people you can interview to confirm packaging tiers and old SKUs. Instead of guessing which ex-employees held that knowledge, ask for them in plain English.

The procedure, end to end

This is the sequence the teardown followed. Procurement-first analysts start at Step 3 with hard prices and treat archives as corroboration; pricing-strategy sources argue you must fix the value metric and packaging map first, because no number means anything until you know which tier it belongs to. Both orders reach the same place if you finish every step.

The confidence label is the deliverable, not an afterthought. High confidence means three or more independent source types converge on a range. Medium means two, or three that partly diverge. Low means one source, or an anchor you could not corroborate. Attach it to every figure and write the one-line audit note next to it.

Tier: <edition/tier name>
Scenario: 100 seats, one region, 3-year term
List anchor: $<n>/seat/mo (archive snapshot <date>)
Discount band applied: <low%>-<high%> off list
Net range (commercial): $<low>-$<high>/seat/mo
Public-sector bound: $<n>/seat/mo (USAspending/TED, <date>)
Utilization haircut: <n>% unused, effective +<n>%
Confidence: High | Medium | Low
Audit note: <which independent source types converged or why they diverged>

How this goes wrong: the failure modes

Every reconstruction fails in one of a handful of predictable ways. Each one has a false positive and a check. This section is the part to reread before you ship.

Two more deserve their own paragraphs because they are the subtle ones.

The effective-versus-quoted gap. Using licensed-seat price ignores the 20 to 40 percent of seats that go unused, understating true unit cost. Where you know utilization, apply the active-user haircut. Without it, you will report a competitor's per-seat economics as healthier than they are.

Forced convergence. Averaging three conflicting points into one tidy number hides a real bimodal split, usually SMB versus enterprise packaging. Triangulation does not imply seeking complete agreement; the goal is to explore the contradictions, not force a singular interpretation. Keep the divergence and label it. The spread between your commercial and public-sector bounds is itself a finding about how the vendor packages and discounts.

Convergence, not agreement, is the confidence signal. A range that stays wide because independent sources genuinely diverge is more honest than a narrow averaged point - and the divergence maps the tier structure for free.

Before you ship the reconstruction

Run this checklist before the range leaves your desk. It is the difference between a defensible document and a guess with a decimal point.

Keeping the reconstruction current

A price reconstruction decays the moment a competitor repackages, so treat it as a living document with two re-check triggers. First, re-run the archive query on a schedule, because coverage is unpredictable and a new snapshot can appear months after you last looked. Second, re-run the procurement search when a new award period opens, since a fresh TED or USAspending record is the hardest evidence you will ever get and it dates itself.

The scarce input is not the data. It is the person who can confirm which tier an old number belonged to. That knowledge sits with ex-employees now in RevOps or pricing-strategy seats, and with the procurement officers who signed the contracts. Refolk finds them by description - ask for ex-employees of a target competitor now in pricing-strategy roles, or the public-sector officers listed on TED and USAspending awards - so the human corroboration that upgrades a figure from medium to high confidence is a query, not a hunt. When the range and the packaging map both hold up against a person who lived them, you are done.