Last updated: October 23, 2026
Three in four software companies changed their pricing or packaging within the last year, according to a survey of more than 230 B2B software and AI companies run in April and May 2026 and published on May 13, 2026. The respondents ranged from startups under $1M in annual recurring revenue to companies above $150M, and the write-up doesn't say where they were based.
Read that figure as a signal of how common pricing changes are, with two limits. The survey comes from Kyle Poyar's Growth Unhinged newsletter, which is sponsored by Metronome, a billing software company. And the post doesn't say how respondents were found, so treat the sample as self-selected. Still, pricing and packaging that change this often go stale fast in other people's copy. Review sites, old launch posts and competitor comparison pages keep repeating last year's numbers, and AI answers pick them up.
My verdict: yes, publish pricing, even if you sell to enterprises. You don't have to publish every number. But give search engines and AI assistants at least one current, dated, plain-text statement of what your product costs, or they'll build the answer from whatever else they can find. Below are the options between a full price list and "contact sales," how to structure a page that gets quoted correctly, and when keeping numbers private still makes sense.
What happens to your price in AI answers when you hide it
When a buyer asks an AI assistant "how much does [your product] cost," a "talk to us" page doesn't end the search. The assistant looks for a number somewhere. If your own site has none, the candidates are a review profile you last updated two years ago, a Reddit thread where someone mentioned their renewal quote, a competitor's comparison page, or a roundup written by an affiliate site.
Google is explicit that its AI features run on the same index as regular results. Its guide to generative AI features in Search says features such as AI Overviews and AI Mode rely on "our core Search ranking systems to retrieve relevant, up-to-date web pages," and that a page must be indexed and eligible to show with a snippet to appear in them at all. A pricing page that hides its numbers behind a form or a sales call gives those systems no price to retrieve.
Buyers are using AI early. In interview research G2 published in June 2026, 52% of 335 B2B professionals described starting vendor discovery and shortlisting with AI. The same research found 76% still expect to negotiate price and terms with a person, and only 10% said upfront pricing transparency was what they mainly wanted (another 14% wanted both visibility and room to negotiate). G2 didn't say when the interviews took place, and it coded the transcripts with AI, so read the percentages as directional.
That 10% is the honest counterpoint to my verdict: most buyers in G2's interviews didn't ask for a full price list. They wanted an early signal to build a shortlist, and they still expected a negotiation later. A published starting point serves the first need and leaves the negotiation with your sales team.
A note on the research you'll see quoted about AI and software prices. Most of the public tests of how often assistants misquote SaaS prices come from agencies and AI-visibility vendors that sell fixes for the problem, so I haven't used their figures here. Your own check takes ten minutes and tells you more about your product. Ask ChatGPT, Gemini, Perplexity and Google's AI Mode what your product costs, note every number, and trace each one back to the page it came from.
The options between a full price list and "contact sales"
Between those two ends there are at least five positions, and each gives AI answers something different to work with.
Option | What you publish | What an AI answer can say | Best for |
|---|---|---|---|
Full price list | Every plan, price, limit and billing term | Your actual prices, current as of your last edit | Self-serve and product-led products |
Starting price | "From $X per month," plus what that plan includes | A real floor, so you aren't filtered out of budget questions | Mid-market products with add-ons |
Price ranges | A band per tier, such as $15,000 to $40,000 a year | A realistic budget range, without your rate card | Sales-led products with negotiated deals |
Pricing logic | The pricing unit (seats, usage, locations) and what drives cost up | How cost is calculated, though not the final number | Usage-based products with wide variance |
Contact sales only | Nothing beyond a form | Whatever third parties have guessed | Very few companies |
The first three put a number you control on the page. The fourth at least stops an assistant from inventing your pricing model, and the last hands the answer to whoever wrote about you most recently.
If reseller agreements or regulated tenders stop you from publishing numbers, choose option four and write it well. "Priced per active location, with volume discounts above 50 locations and an annual minimum" tells a buyer, and an AI summary, far more than "flexible plans for every team."
How to build a pricing page that gets quoted correctly
A pricing page has two jobs now: persuade a human, and give machines a version of the facts that's hard to misread. These steps cover the second job without hurting the first.
Put every price in HTML text. Prices baked into images, or loaded only after a currency selector or slider runs, are easy for a crawler to miss. If you use a slider, render a default value in the HTML.
Name each plan once and use that name everywhere. If the page says "Growth" and the docs say "Pro," an answer engine has to guess whether they're the same plan.
Label the billing term next to each number. "$40 per user per month, billed annually" and "$48 billed monthly" are both true. A bare "$40" invites someone to quote the annual price as the monthly one.
List limits and inclusions as plain lines: seats, usage caps, integrations, support level, SSO, data retention. A comparison table works well here, because each cell holds one fact.
State the currency and any regional differences. If EU or Indian prices differ, say so on the page instead of relying on geolocation alone.
Add a visible "Prices last updated" date. It shows the page is maintained, and it gives you a date to check against when an AI answer quotes an old figure.
Keep one canonical pricing URL. When you change plans, update that page instead of publishing a new one, and redirect any old pricing URLs to it.
Structured data is optional. Google's guide says there's no special schema.org markup needed for its AI features. If you want software app rich results, Google's SoftwareApplication documentation requires an offers.price value (plus a rating or reviews), which a paid product can only supply if the price is public. The visible text matters more than the markup. For the wider picture on markup and AI, see our guide to schema markup for AI search.
Enterprise and "contact sales" tiers without vague copy
Most B2B SaaS pricing pages end with an enterprise column that says "Custom" and a button. That's fine, as long as the column still says something concrete.
Write down what changes at the enterprise tier: SSO and SCIM, audit logs, a data residency choice, an uptime commitment, a named support contact, security review support. Then state the commercial facts you're comfortable sharing, such as a minimum contract length, a typical starting point, or the unit you price on.
Compare two versions of the same column:
"Enterprise: custom pricing for large teams. Contact us."
"Enterprise: annual contracts from $30,000. Adds SSO, SCIM, audit logs, EU data residency and a 99.9% uptime commitment. Priced per seat with volume tiers above 500 seats."
An assistant answering "does [product] offer SSO, and what does it cost" can quote the second version as a true answer. The first sends it looking elsewhere.
You don't have to publish your floor if deals vary wildly, but you do need to say what the tier contains. Buyers screening on security and compliance features drop vendors who make them book a call to learn whether SSO exists.
Keep the price consistent everywhere it's quoted
Your pricing page is only one source. AI answers also draw on review profiles, marketplace listings, your own help center and the comparison pages you and your competitors publish. When those disagree, the assistant has to pick one, and it won't always pick yours.
When pricing changes, update these on the same day:
Your G2, Capterra and similar review profiles, including the pricing fields, not just the description
App marketplace listings (Salesforce AppExchange, HubSpot, Shopify, Atlassian or wherever you're listed)
Help center articles that mention plan limits or prices
Your own comparison and alternatives pages. Our guide to comparison pages covers how to build them
Sales decks and PDFs that are publicly reachable, which crawlers find more often than teams expect
Old pages on your own domain are the sneakiest source of stale numbers. Search your site for the old plan names and prices after every change. A 2024 launch post announcing "new plans from $29" can keep feeding answers long after you moved to $39, so update it, add a note at the top, or redirect it.
What to do, by situation
If you're self-serve or product-led, publish the full list. Your buyers expect it, and any price that isn't on your page will be filled in by someone else's guess.
If you're sales-led in the mid-market, publish a starting price per tier and the main cost drivers. That keeps you in "affordable [category]" and budget-filtered prompts without exposing your discount structure.
If you sell mostly to enterprises, publish ranges or at least the pricing logic, plus a full feature list per tier. Your sales team still negotiates every deal, and it stops spending the first call on "roughly what does this cost?"
If your pricing is usage-based, publish the unit price and a few worked examples, such as what a team sending 50,000 emails a month would pay. A worked example is easy to quote and hard to misread.
If you're changing prices this quarter, publish the new page first, then work through the list in the previous section. Note the date the change took effect, so anyone comparing old and new numbers can tell which is current.
What to watch out for
Pricing that only exists inside a calculator. If the default state shows no number, crawlers see no number.
Two pricing pages. A regional page or an old "plans" URL that still resolves splits the signal and keeps outdated prices alive.
Annual and monthly prices without labels. This misquote is one of the easiest to cause and one of the easiest to prevent.
Hiding prices because a competitor might see them. Your competitors already know roughly what you charge from lost deals. The buyers who don't know are the ones you want.
Treating the page as done. Put a pricing review in the same checklist as the price change itself, and re-run your AI prompt check a week later.
Expecting AI answers to update overnight. Assistants that search the web can pick up a change within days. Answers that lean on older training data take longer, which is why the date on your page matters.
Who gets to describe your price
Publishing pricing used to be a conversion question. Now it also decides who describes your product when a buyer asks an AI assistant. If you don't state a price, someone else will, and their number is usually older than yours.
Start small if you have to: a starting price, a clear unit, a dated page. Then work outward through your profiles and old pages. For how pricing fits the rest of a B2B SaaS search program, read our B2B SaaS SEO guide, and if you want to track how assistants describe you over time, our guide to measuring AI search visibility covers the setup.