# Beyond the monthly price: what’s changing in SaaS billing

> From queries to credits and conversations to resolutions: 38 SaaS companies changed their public pricing, usage definitions or the package around paid work.


How much does a SaaS product cost? Increasingly, the answer depends on what happens inside it. A conversation can qualify for billing after a few messages. A single workflow can consume several kinds of credits. An allowance can cover a whole team—or grow with every paid seat.

Over the last 90 days, Competitor Tracker & Co. reviewed **70,595 website changes across 2,486 companies and products** to see how those terms are changing. We compared earlier and later versions of pricing pages, product pages and billing documentation, checked the current terms, and removed promotions and repeated examples.

This report brings together **38 companies** whose public offers changed: the work they count, the results they charge for, the rates attached to individual activities, and the usage included in a plan. Some introduced a different billing unit. Others published clearer limits, exclusions or unlimited-use packages.

Together, the examples show how much of a SaaS offer sits beneath the monthly price. The changes fall into four groups.

## 1. The unit on the pricing page is changing

A named unit tells a buyer what makes usage grow. Customer accounts, candidates and monthly tracked users connect the price to the size of a business or audience. Actions, credits and content created connect it to the work done inside the product.

Several companies replaced one of those descriptions with another. The visible price sometimes stayed exactly where it was. Elsewhere, both packages and units changed, preventing a direct comparison of headline prices.

| Company | Category | What changed |
| --- | --- | --- |
| Cust | Customer success | Foundation previously offered 100 protected customer accounts for $500/month; Growth offered 500 for $1,500. The new monthly-billed offer uses shared AI-work credits: Foundation $480 for 1,000, Growth $1,800 for 4,500, Scale $3,600 for 10,000 and Enterprise $7,500 for 25,000. Unlimited seats remain. |
| Workable | Recruiting | “1 credit = 1 candidate worked by the agent” became separate AI actions: Evaluate uses 1 credit per screening; Source uses 2 per passive candidate identified and added; Chat uses 10 “each time a candidate interacts with the chat.” |
| Fibery | Work management | A June product update announced AI billing in credits instead of calls. Pricing replaced Smart Agent call allowances with monthly AI credits. The announcement lists 250 credits per paid seat on Standard, 500 on Pro and 1,000 on Enterprise, plus €15 per extra 1,000 credits and a configurable monthly spending cap. |
| ThoughtSpot | Analytics | Analytics Pro changed its starting usage rate label from “$0.10 per query” to “$0.10 per credit” and added unlimited-LLM-token wording. The evidence does not establish that one query equals one credit; unlimited tokens also do not make the entire subscription unlimited. |
| [Amplitude](/companies/amplitude-analytics/) | Product analytics | The Guides and Surveys FAQ previously tied pricing to monthly tracked users, expressly excluding the number created. It now ties pricing to guides and surveys created, expressly excluding monthly tracked users and event volume. |
| Kibo Commerce | Commerce | KIBO removed published AI conversation-volume tiers and added a FAQ describing pricing per function, based on order lines. Order lines already appeared in the old annual subscription alongside conversation pricing. New per-function rates are unstated. |
| Adjust | Mobile measurement | Plans relabeled attributions as conversions. Enterprise changed from over 250,000 annual attributions to over 2,500,000 annual conversions. Their equivalence is unestablished, so the larger number cannot be read as a like-for-like increase in included capacity. |

![Workable’s pricing, captured September 7. Evaluate uses 1 AI credit per screening, Source uses 2 per passive candidate identified and added, and Chat uses 10 each time a candidate interacts with the chat.](/img/reports/saas-paid-work/workable-action-credits-sep-07.png)

*Workable’s pricing, captured September 7, 2026*


A company can have the same customer base while doing more analysis, screening or content creation. Once activity becomes the counted unit, that distinction becomes part of the public offer.

Credits add another layer: the allowance has a common name while the underlying work can have different weights. Two offers saying “1,000 credits” tell us little about their relative capacity without those definitions.

This extends the story in our [free-plan report](/blog/saas-free-plans-are-changing/). There, access and usable features were moving. Here, the definition of paid use moves after a customer has already entered the product.

## 2. Vendors are spelling out which results count

Support and sales-data products make the distinctions especially visible. A handled ticket, a completed conversation, a resolved issue and a found contact each mark a different stopping point in the work.

Some count work that reaches a human. Others explicitly exclude it. In data enrichment, a provider returning a record is the stated success test; that is a narrower promise than a qualified lead or a sale.

| Company | Category | What changed |
| --- | --- | --- |
| Helply | Customer support | Earlier integrations copy charged per resolved ticket and made failed runs and human escalations free. Later copy says $1 per ticket with every integration, action and AI run included. Current pricing defines a ticket as any customer conversation processed through Helply. The offer is an annual commitment sized to expected volume, with a 250-ticket monthly minimum and $3,000 minimum annual contract. |
| [Featurebase](/companies/featurebase/) | Customer support | “$0.49 / resolution” became “$0.49 per successful resolution. Charged once per conversation.” Current documentation counts confirmed or assumed satisfaction and reverses the charge if the customer returns to that conversation for more help, even across billing periods. The observed change makes the public wording more explicit; it does not date the policy's introduction. |
| Dixa | Customer support | The old page said AI costs did not increase with contact volume. The new Mim FAQ publishes €0.35 per conversation, whether resolved by AI or handed to a representative, with no per-seat AI component or allowance cap. The unit rate stays fixed as volume grows; the total does not. |
| Botpress | Conversational AI | The conversation threshold changed from at least two messages to at least two messages from the end user. Voice adds one conversation per three minutes. The page also separates paid-plan AI-spend auto-recharge into $10 credit grants at 95% of the AI limit; conversation-pack purchases also trigger at 95%. |
| Nextiva | Business communications | An added XBert FAQ defines a conversation as a call longer than 30 seconds or an SMS/web-chat thread with at least three interactions. Calls under 30 seconds and single-message chats are excluded. Earlier wording already advertised $99/month for 100 interactions and $0.99 for extras; the FAQ uses conversations at those same displayed amounts. |
| [Saleshandy](/companies/saleshandy/) | Sales prospecting | Expanded Lead Finder pricing detail gives one credit for an email waterfall plus verification, six when a phone is found and seven for phone plus email. Unverified emails and missing phones cost $0. This is newly visible result-based detail on the main pricing page, rather than evidence that unsuccessful searches previously consumed credits. |
| [FullEnrich](/companies/fullenrich/) | Contact enrichment | Its reverse-email-lookup page added “Pricing on a miss” and “No match, no credit.” It does not establish that misses were charged before the addition. |
| Crescendo | Customer support | A product-page rewrite added “You only pay when it resolves, not when it routes.” Later voice copy repeats that distinction while removing an older guarantee paragraph covering dissatisfied outcomes. The removal establishes a wording change, without proving that dissatisfied resolutions are now charged. |
| Yuma | Ecommerce support | A contact-for-pricing page became full-resolution packaging: one count per fully resolved ticket, with human involvement excluded. Illustrative, forecast-based annual bundles paid upfront include $14,400 for 12,000 resolutions and $40,000 for about 48,000. Annual usage pauses at the bundle limit without overage; the separate monthly offer lists $1.50 per extra full resolution. Both carry a 12-month commitment. |
| [Regie.ai](/companies/regie-ai/) | Sales prospecting | The integrations page added an explicit enrichment rule: a credit is consumed only when a provider returns data, so misses cost nothing. Research, drafting, enrichment and dialing share credits, but this exclusion describes enrichment specifically. |
| Paligo | Technical documentation | **Pre-launch offer:** the August 29 Nexus Answers page said “Priced on successful answers,” starting at $0.60, with thumbs-down answers excluded. The September 5 capture removes that rate and feedback FAQ, replacing the wording with pricing per answer given in conversations. Both target October 2026 general availability. This is a prospective offer revision, not evidence of live charges for negative feedback. |






The support examples resist a single “pay for outcomes” label. Conversation pricing can include escalation. Resolution pricing can exclude human involvement. An assumed resolution can later be reversed. Each gives a different meaning to the short phrase next to the price.

The exclusions are equally revealing. A missed contact lookup, an unanswered greeting and a short call can fall outside a vendor's stated count. The qualifying sentence can distinguish offers whose headline unit looks similar.

## 3. More individual activities have a published price

A workflow can contain several kinds of paid work. A code review, an agent run, an article improvement and a function invocation now appear as separate lines in the offers below. Some are weighted credits inside a subscription; others are extra charges beyond an included allowance.

There are genuine rate changes here, alongside newly disclosed rates and estimates. Treating every added row as a newly introduced fee would exaggerate what the website evidence proves.

| Company | Category | What changed |
| --- | --- | --- |
| [Coresignal](/companies/coresignal/) | Business data | Main pricing changed contact enrichment from 10 to 20 credits. Separately, the Jobs API page replaced Search and Collect credit categories with unified credits and an entry offer of $49/month for 2,500 credits. That packaging is product-specific. |
| Greptile | Code review | July 25 copy listed $30/seat/month, 50 included credits per seat, one credit per standard review and $1 per additional credit. August 1 copy explicitly adds three credits per TREX review, labeled beta. The earlier TREX rate is unstated, so this is a disclosed weight rather than a demonstrated increase. |
| Zapier | Automation | Pricing added shared task accounting for AI, code and SDK work, with some successful actions consuming more than one task. The FAQ replaced the explicit 1.25× excess-use rate and 3× maximum with “higher rate” and maximum-limit wording. Failed-action exclusions and pauses remain; neither a higher new rate nor removal of the cap is established. |
| HelpDocs | Knowledge bases | The AI page added three free applied improvements per run, then one credit per improvement. A description costs one credit per article, tags cost one per article and a short version costs two. The page previously promoted the tools without those credit explanations. |
| Datadog | Observability | Bits Agent Builder's estimate changed from 0.3 credits per message to approximately 3 per manual or autonomous run. Bits Chat moved from feature-specific estimates of 0.5–0.7 per message to approximately 0.5. Complexity and context affect consumption. A message and a run are different denominators, ruling out a simple tenfold-price-rise reading. |
| CodeRabbit | Code review | An expanded FAQ adds eligible PR/CLI reviews beyond included limits at $0.25 per reviewed file, with Automatic, On demand or Off controls and a monthly spending cap. Cloud includes the first turn, then costs $0.40 per agent minute; Slack costs $0.40 per active agent minute. The Agent page's earlier advertised rate was $0.50; the active-runtime definition already existed. |
| Swell | Commerce | Pricing added monthly app-function allowances: Starter 100K, Basic 500K, Standard 2M and Unlimited 5M. Calls in test and live environments count. Excess use is $5 per 100,000 calls **or part thereof**, billed in the month used. API requests and storage remain separate dimensions. These newly visible rows have no verified contract start date. |
| Shopware | Commerce automation | **Nexus beta, planned pricing:** an approximate €10 per 1,000 executions starting in September became plan-tier free monthly allowances plus €12 per 1,000 excess executions starting in October. A later pricing capture extends the free introduction to October 6, 2026. The beta offer remains subject to change; these are revised plans for charging, rather than observed paid invoices. |

![CodeRabbit FAQ showing usage-based reviews at $0.25 per reviewed file, admin authorization modes and a monthly spending cap.](/img/reports/saas-paid-work/coderabbit-controls-sep-07.png)

*CodeRabbit’s FAQ, captured September 7, 2026. The extra-review price appears beside authorization choices and a spending cap. Cloud and Slack agent terms appear separately below.*

These rates describe different slices of a job. A reviewed file measures the scope of extra review work. An agent minute measures runtime. A function call can happen in testing as well as in a customer-facing transaction.

The authorization language belongs to the same commercial story. Paid continuation can be optional, capped or paused. An excess-use rate describes one path after the allowance, rather than an unavoidable bill.

## 4. The package around that usage is changing too

An allowance needs an owner and a clock. It can belong to a person, an account or a shared workspace. It can refresh daily, expire monthly or cover an annual commitment. Purchased credits may follow a different expiration rule from included credits.

The final group shows companies changing those dimensions and explaining what happens when the balance is gone. One new AI offer takes the opposite direction: unlimited conversations.

| Company | Category | What changed |
| --- | --- | --- |
| Walnut | Sales demos | Existing 50K/100K/200K AI-credit quantities gained annual workspace-pooling explanations, per-person controls and top-up packs: 50K for $1,000, 100K for $1,800 and 200K for $3,300, plus custom volume at $0.015/credit. The page promises no overages. Notices precede an AI pause at 100%; credit pricing itself predates this change. |
| Vellum | AI development | Documentation added different clocks: included usage expires at billing-period end without rollover; purchased PAYG credits expire 12 months after purchase. Included usage is spent first. Recurring packages also moved from dollar-named allowances to named usage bundles. A bundle's dollar price does not establish an equal prepaid credit balance. |
| Hiver | Customer support | The Voice add-on changed from 2,000 account-pooled minutes/month to 1,000 minutes per voice seat/month, pooled across voice seats. Total included team capacity now depends on the number of those seats; describing the change as an unconditional halving would miss the new denominator. |
| Prophecy | Data preparation | Starter changed from 20 credits per user to 20/month; Professional from 50/month to 50/user/month; Enterprise Express from flexible credits to 5,000/year. |
| [Unify](/companies/unifygtm/) | Sales prospecting | Earlier annual-billed offers displayed 50,000/200,000/600,000 credits. New pricing lists Base at 800 and Pro at 2,400 credits per seat/month, plus a custom annual workspace pool. A new FAQ describes hard-cap pauses and non-renewing pooled top-ups. In-flight automation can finish its current step before the hard-cap pause. |
| Piwik PRO | Analytics | The FAQ replaced end-of-cycle overage fees, with possible upgrades for frequent or significant excess use, with an automatic month-end upgrade to a tier matching usage and no overage fees. Extra usage now leads to a different stated billing mechanism; an upgrade can still increase spending. |
| Type.ai | Writing | Paid cards added Basic 500,000, Pro 1,500,000 and Max 6,000,000 credits, where relative usage multiples had appeared before. Later copy distinguishes a fixed non-expiring free grant, monthly subscription credits without rollover and non-expiring purchased credits used afterward. The free grant is not described as recurring. |
| Coassemble | Training | Build moved from 50 individual registrations/month and “No lock-in” to 600/year and a 12-month agreement; Scale changed from 200–10,000+ per month to 2,400–120,000+ per year. Build anonymous views changed from 1,000/month to 12,000/year. Annualized figures do not establish extra capacity or unrestricted pooling; AI image and narration quotas remain monthly. |
| Activepieces | Automation | Pricing made Free's 100 credits/day, Plus's 10,000/month, Team's 50,000/month and $0.007 per excess paid credit explicit. Free excess runs wait until the next morning. Daily refresh and paid excess usage were already described before these figures appeared. AI steps have different credit weights, so credits cannot universally be read as completed runs. |
| [Klaviyo](/companies/klaviyo/) | Marketing | The free mobile allowance changed from 150 SMS/MMS/mobile-message credits per month to $5 of mobile messages per month. Region and carrier affect consumption. The denomination alone establishes no change in send capacity. |
| Lindy | Work automation | August 4 individual-plan copy used “Standard usage,” “3x” and “7x.” August 15 specifies 3,000/15,000/35,000 monthly credits per user, a shared workspace pool and per-seat allocation controls. Enterprise had already mentioned shared usage and bonus credits; the newly explicit individual-plan allowances do not mark the invention of credits or sharing. |
| Olark | Customer support | A page led by $29/month per human seat and custom AI automation became a Standard AI Website Agent at $400/month, billed annually, including one human seat and unlimited AI conversations. Additional human seats cost $29/month. It is a different package from the old human-seat offer, with unlimited AI conversation use at its center. |

![Lindy’s pricing, captured September 7. The Plus, Pro and Max cards specify 3,000, 15,000 and 35,000 credits per user per month.](/img/reports/saas-paid-work/lindy-credit-allowances-sep-07.png)

*Lindy’s pricing, captured September 7, 2026*


The commercial difference can sit in the clock as much as the count. A monthly allocation and an annual commitment expose usage on different timelines. A per-seat contribution to a shared pool connects team size to available work.

Exhaustion also has several published answers: pause the AI, queue work until tomorrow, buy another pack or move to a larger tier. “No overages” can describe a stop or a plan upgrade.

## The offer is getting more specific

The 38 examples show several directions at once. Units are being replaced, successful work is being defined, individual activities are acquiring published rates and allowances are being repackaged. Unlimited-use bundles remain part of the picture.

A single price-rise narrative would erase the most useful differences. Some pages reveal a changed rate. Others reveal a new exclusion or explain an existing policy. Still others describe a proposed product before general availability. Public offers are becoming more explicit about the work they cover, even when that work is described with the same familiar word: credit, conversation or ticket.

The price on the card and the definition beneath it together describe what the customer is buying. That is why our [competitor price monitoring](/features/competitor-price-monitoring/) follows pricing pages, product details and usage definitions together. Competitor Tracker & Co. sends the changes in a weekly report.

See a [sample dossier](/demo/) or [start tracking](/#engage).

*— C. T. Lucky*


---

Source: https://competitortracker.io/blog/what-are-saas-companies-charging-for/
Updated: 2026-09-07