The Rival Pricing-Edit Reference: Proof, Mimics, and Effective Dates

A rival's pricing page changed and a monitor flagged it. This reference is for strategy teams, talent-intelligence analysts, and operators who have to say what that edit actually proves before it goes into a competitor brief. Find the edit type, read what it proves and what can fake it, run the seven-step check, and file the finding with a bounded first-seen window, a first-party effective date, and a review date.

What a pricing-page edit can and cannot prove

On its own, a pricing-page diff proves only that one fetch of one page returned different content at one moment. It becomes a strategy finding when the new value holds across repeated fetches, a first-party source confirms it, it has a date, and it is tied to the buyers who actually pay it.

Most writing on competitor pricing page analysis stops at detection and offers one-line reads such as "price removed means moving upmarket." That read skips four questions a brief has to answer:

1. Is the change real, or an artifact?
2. When did it take effect?
3. Who does it apply to?
4. Is it this rival's move, or the whole category's?

Terms used throughout:

• First-seen date: the first time your monitor or the archive saw the new value. It tells you when the change was detected, not what the rival decided.
• Effective date: the date the vendor says the new price or package applies. It comes from a first-party notice, never from a diff.
• Bounding window: the interval between the last archived capture with the old content and the first capture with the new content. The change happened somewhere inside it.
• Digest: the content hash the Internet Archive stores for each capture. A new digest means the bytes changed, not necessarily the price.
• Grandfathering: terms that let existing customers keep the old price for a period, so the list price is not what everyone pays.
• Mimic: any artifact that produces a diff without a real pricing change, such as an A/B variant or currency localization.

Pricing-page edit lookup: what each change proves

Each row gives one edit type: what it proves, what it does not, the artifacts that mimic it, where to confirm it, and how long the read holds. Jump to the row that matches your alert.

List-price move

• Proves: new buyers in that region and billing period are quoted a different list price.
• Does not prove: that existing customers pay it, or that every currency moved by the same amount.
• Mimics: A/B variant, geo or currency localization, promo banner, grandfathered pricing.
• Confirm in: the vendor's announcement or blog post, billing help articles, and the terms.
• Read stays valid: until the next change. Re-verify quarterly.

Slack's Pro increase on 1 September 2022 is the reference case. Monthly went from $8 to $8.75 (up 9%), and annual from $6.67 to $7.25 per month (up 8%). Existing annual subscribers could lock in the old rate by renewing before 1 September, and Business Plus and custom enterprise plans were not affected. A brief that said only "Slack raised prices 9%" would have been wrong for a large share of its customers.

Annual-discount shift

• Proves: the rival changed how hard it pushes buyers toward an annual commitment.
• Does not prove: a change in discount strategy if monthly and annual prices moved together.
• Mimics: a monthly/annual toggle whose default state the capture did not record, a JS-rendered table that was not captured, a promo banner offering a limited-time annual deal.
• Confirm in: the announcement, which usually states both billing periods.
• Read stays valid: until the next change.

Compare the gap between the monthly price and the annual per-month price, before and after. In the Slack case both rose by about the same proportion (9% monthly, 8% annual), so the gap barely moved. That was a list-price move, not a discount shift. Billing toggles are often built in JavaScript, so check that the archive captured both states before you compare them.

Tier added or removed

• Proves: the packaging boundary moved. An added tier shows where the rival wants a new entry point or step-up point. A removed tier shows a package it no longer sells to new buyers.
• Does not prove: that existing customers on a removed tier were moved off it.
• Mimics: an A/B test of packaging, a JS-rendered table that was not captured (a tier card missing from one capture), grandfathered tiers still shown to logged-in customers.
• Confirm in: the changelog, the docs, and help-center migration articles.
• Read stays valid: months to years. A tier added to replace a cut one tends to stick.

Heroku's Eco tier is the case to know. Eco charges began on 1 December 2022, three days after free dynos ended. Eco now costs $5 per month for 1,000 dyno hours shared across an account's Eco apps, below the $7 starting price announced in 2022.

Free-tier cut or shrink

• Proves: the rival decided the free plan cost more than it brought in, and it usually replaces it with a cheap paid entry tier.
• Does not prove: a retreat from self-serve. Both documented cases kept a low-priced way in.
• Mimics: a renamed product page, an archive capture gap, aggregator listings that still show the old free plan.
• Confirm in: the changelog and the vendor's own post explaining the change.
• Read stays valid: long. Heroku's cut has never been reversed.

Heroku's changelog states that from 28 November 2022 it no longer offered free dynos. Its stated reason was fraud and abuse of the free plans, and the current pricing page still has no free dyno. Stack Overflow's Teams free plan launched in March 2021 for up to 50 users. The renamed product's Starter plan is now free for up to 3 collaborators. The date of that shrink is not established publicly. If you need it, bound it from the product's own archive history.

Price replaced by "contact sales"

• Proves: at minimum, the rival stopped publishing a number for that tier.
• Does not prove: a move upmarket. I found no documented case of a developer-platform top tier switching to "contact sales", so there is no precedent to lean on. Treat the upmarket reading as a hypothesis.
• Mimics: a JS-rendered table that was not captured. This is the strongest mimic, because an empty price cell looks exactly like a removed price. Also A/B variants and geo flicker.
• Confirm in: the enterprise page, where seat minimums tend to appear, and the terms.
• Read stays valid: short until something confirms it. Re-check within the quarter.

When a competitor replaces a price with "contact sales", first prove the capture is complete (step 4 of the procedure). Then look for a seat minimum or new enterprise packaging that explains the change.

Seat minimum added

• Proves: the rival set a floor on deal size for that tier.
• Does not prove: that small existing accounts were pushed out. They may be grandfathered.
• Mimics: grandfathered pricing, a promo banner advertising a team bundle.
• Confirm in: the enterprise page and the terms. One practitioner source notes that seat minimums land on the enterprise page, not in the main pricing table.
• Read stays valid: until the next packaging change.

Value-metric swap

A value metric is the unit a price is charged against: seats, API calls, compute hours, or outcomes.

• Proves: the rival changed what it charges for. This is the strongest packaging signal on this list.
• Does not prove: a strategy specific to this rival if the swap is from seats to hybrid. That mostly follows the category (see base rates below).
• Mimics: an A/B test of a new page layout, a promo banner, copy edits that rename the unit without changing billing.
• Confirm in: the docs, where metering rates appear, and the terms, where overage policy appears.
• Read stays valid: long once the docs confirm it, because the billing system has to change too.

If a competitor switched to usage-based pricing, the docs page that lists metering rates is the proof. Until the docs agree, a pricing page that says "pay for what you use" without published rates is marketing copy.

Credits or AI add-on

• Proves: the rival charges for AI separately rather than folding it into existing tiers.
• Does not prove: a distinctive strategy. Bain reported that roughly 65% of companies introduced a hybrid approach that adds an AI meter on top of seat-based pricing.
• Mimics: trial credits in a promo banner, an A/B variant.
• Confirm in: the docs for credit consumption rates, and the terms for overage.
• Read stays valid: short, because credit rates are easy to adjust. Re-verify quarterly.

Feature moved between tiers

• Proves: the rival moved a fence, meaning the feature boundary that pushes buyers from one tier up to the next.
• Does not prove: a change if the feature was only renamed.
• Mimics: cosmetic copy edits, an A/B variant, a JS-rendered comparison table that was not captured.
• Confirm in: docs and changelog entries that name the feature and the tiers.
• Read stays valid: until the next packaging review.

I have no documented public case for this edit type, so judge it on confirmation alone. A feature move with no changelog or docs entry stays unconfirmed.

Who sits behind a pricing change at a rival

In Refolk's index of professional profiles, pricing authority in US SaaS sits with a few senior people. Of 16 US profiles with pricing titles and "SaaS pricing" in the headline, 8 are Director of Pricing or Head of Pricing.

The title bands are my grouping. The employers in the set include SailPoint, Boomi, Ping Identity, Deltek, Salsify, and Versapay. The largest metro cluster is Atlanta, with 3 profiles.

What this suggests, labelled clearly as a hypothesis: if pricing decisions sit with a few senior people, a rival hiring a new pricing leader may come before a pricing-page change. I have not tested that lead time, and 16 profiles is a small sample. Treat a new pricing-leader hire as a reason to watch the page more closely, not as evidence of a change.

One practical trap: adding a "Computer Software" industry filter to the same search returned 0 results in both the US and the UK. The industry tag is unreliable for this role. Search by title and by named employers instead. Refolk takes the request in plain English and matches on title and employer, so you can name the companies you track instead of trusting an industry tag.

How a pricing diff lies: artifacts and failure modes

Six artifacts produce pricing diffs without any pricing change: A/B variants, geo or currency localization, archive capture gaps, JavaScript tables that were not captured, promo banners, and grandfathered pricing. Rule out each one before you write a read.

Noise is the normal state of a pricing monitor, not the exception.

One vendor guide suggests tuning filters until false positives fall below roughly ten percent of alerts. Vendors say element-level selectors that watch only the pricing table or tier cards reduce false positives, but none publishes by how much.

Localization deserves its own check because the same change can land unevenly across currencies:

These are Slack Pro figures for 1 September 2022, with the percentages derived from the old and new prices. The GBP figures may be mismatched. If they are right, the UK move far outran the USD move, and a monitor watching from a single region would have understated it.

The eight ways this work goes wrong, and the check for each:

1. A/B flicker read as a price move. A single fetch shows a new price that later reverts. Check: fetch again from a fixed region and require the value to hold over time.
2. Edge-location currency mistaken for a price change. $49 and £39 alternate between fetches. Check: did the currency symbol change along with the number?
3. JS shell read as "pricing removed." The capture shows an empty table. Check: compare the CDX length field with earlier captures, then open the replay.
4. Misdating through collapse. A timestamp:6 query returns the first capture of the month, so the change looks up to a month off. Check: use collapse=digest.
5. Your evidence capture vanishes. It sat only in the Save Page Now near-real-time index. Check: keep a local screenshot and HTML copy as well.
6. List price treated as what everyone pays. Existing customers may be locked in and higher tiers exempt, as at Slack. Check: read the announcement for lock-in terms.
7. Aggregator sites used as confirmation. Third-party listings contradict each other. For Stack Overflow's Teams product, one major review site listed free for up to 50 teammates while another said there was no free plan. Check: only first-party pages count.
8. Survey base rates compared across definitions. "85% usage-based" and "15% seat" do not measure the same thing. Check: cite the definition next to the figure.

How to date a competitor price change

Bound when the page changed using the archive's CDX index, then take the effective date from a first-party notice. These are two different facts, and a brief needs both.

How the CDX index works for this job:

• Fields. The public CDX fields are urlkey, timestamp, original, mimetype, statuscode, digest, and length.
• collapse=digest shows only captures whose content digest differs from the previous line. It collapses adjacent lines only, so a duplicate digest elsewhere in the results stays. A page that flips between two versions shows up as several lines, which is itself a sign of an A/B test.
• collapse=timestamp:6 keeps the first capture of each month, not the most recent. It can misdate a change by up to a month.
• collapse=timestamp:10 shows at most one capture per hour. It is useful for thinning a heavily captured page once you know the rough window.

https://web.archive.org/cdx/search/cdx?url=example.com/pricing&collapse=digest&fl=timestamp,digest,statuscode,length

Record:
Last capture with old digest:  <timestamp>  length <n>
First capture with new digest: <timestamp>  length <n>
Length comparable to earlier captures? <yes/no>
Replay shows prices? <yes/no>

Save Page Now pins the current state, with three limits to plan around:

• A third-party script reports daily limits of 8,000 captures for logged-out users and 100,000 for logged-in users.
• A capture that sits only in the near-real-time index can be dropped before it reaches long-term storage.
• Save Page Now does not add the URL to any future crawls, so it will not watch the page for you.

How long the gap between real captures of a typical B2B pricing page runs is not established publicly. No median or 90th-percentile figure exists. Measure it for your own watch list by running the CDX query across the pricing URLs you track.

Then the effective date. Vendors announce ahead, and a page edit can mark either the announcement or the effective date.

Slack's rise came with lock-in for existing annual customers. Heroku's cut was permanent, with a $5 Eco tier as the paid entry point. There is no systematic public study of lag by source type, so treat these as worked examples rather than norms.

The day your monitor noticed a price is a fact about your monitor, not about the rival.

Step-by-step: verify a pricing edit before it enters the brief

Run these seven steps in order on every flagged edit. Expect about half a day of hands-on work spread over a few days of repeat fetches. The output is a brief row you can defend.

Notes on the steps:

• Step 1 takes about 15 minutes. The aim is an alert that speaks in prices, not pixels.
• Step 2's duration is a judgment call. One to three days of repeat fetches is a working figure, not a tested one. Lengthen it for a rival you know runs pricing experiments.
• Step 3 takes about 30 minutes, most of it spent opening replays.
• Step 5 takes one to two hours. Metering rates appear in docs, seat minimums on the enterprise page, and overage policy in the terms.

Sources disagree on two points, and you should pick a side on purpose:

• Cadence. One page-monitoring vendor checks every 1 to 4 hours. Another practitioner guide warns that hourly monitoring raises your false-positive rate as fast as it lowers detection lag. I would rather run less often from a fixed region than drown in flicker.
• Scope. One guide says to narrow the monitor to the pricing table. Another says to also watch docs and the changelog. These are compatible: narrow the alerts, and widen the confirmation.

Not every source carries the same weight. Rank your evidence like this:

Category base rates: is the move the rival's or the market's?

Compare every pricing-model change with category base rates before calling it strategy. A move most of the category has already made tells you little about this rival. A move few companies make tells you a lot.

How to use these:

• Seat to hybrid follows the category. In the Growth Unhinged survey, hybrid rose from 27% to 41% while pure per-seat fell from 21% to 15%. Among usage-pricing operators in the PwC and m3ter survey, 72% use a hybrid model. A rival making this swap is keeping up, not breaking away.
• A move to outcome pricing is specific to the rival. Pure outcome pricing is 5% among AI monetizers in the High Alpha figures, whose sample size is unverified. A rival moving there is making a distinctive bet.
• AI meters are common. Bain's figure of roughly 65% adding an AI meter on top of seats means an AI add-on is the default, not news.
• Definitions conflict. "85% any usage-based" and "15% pure seat" come from different surveys measuring different things. Never set one against the other in a brief. Quote the survey, its sample size, and its definition next to every figure.

Keeping pricing reads current

Pricing reads go stale within months, so every row in the brief needs a review date, and the default is quarterly. One secondary source counted more than 1,800 pricing changes in a year, an average of 3.6 per company, though its underlying dataset is unverified. Either way, a read that is two quarters old is probably describing a page that no longer exists.

Set review dates by edit type:

• Within the quarter: contact-sales swaps, credits or AI add-ons, anything still unconfirmed.
• Quarterly: list-price moves, annual-discount shifts, seat minimums, feature moves.
• Twice a year: confirmed free-tier cuts and value-metric swaps that the docs back up. Heroku's cut has never been reversed, and these changes are expensive to undo.

Watch the people as well as the page. If the hypothesis about pricing leaders holds, a new monetization or packaging hire at a rival is the earliest sign you will get. A standing search in Refolk for monetization or packaging leads hired in the past 12 months at developer-platform companies flags those hires without you having to check each rival's team page.

File every finding in the same shape:

Rival:
Edit type: (list price / annual discount / tier added or removed / free-tier cut / contact sales / seat minimum / value metric / credits or AI add-on / feature moved)
What changed: (old value -> new value, currency, billing period, region fetched from)
First-seen window: last old-digest <timestamp> to first new-digest <timestamp>
Effective date: <date>  Source: <first-party page>
Applies to: (new buyers / all customers / excludes tiers ...)  Lock-in terms:
Mimics ruled out: A/B [ ]  currency/geo [ ]  capture gap [ ]  JS table [ ]  promo [ ]  grandfathering [ ]
What it proves:
What it does not prove:
Base-rate context: (survey, n, definition, figure)
Evidence pinned: Save Page Now [ ]  local screenshot [ ]  local HTML [ ]
Review date: