Which CRM vendors let AI assistants read their pricing?
24 vendor sites, 14 AI crawlers, one pricing page each, all tested on 24 September 2026 and printed in full below.
- Published in full
- No email required
- Method you can repeat
What we found
- None of the 24 CRM and sales-tool sites we checked on 24 September 2026 blocks GPTBot, ClaudeBot, PerplexityBot or Google-Extended in robots.txt.
- Three sites return a 403 error to AI-bot user agents on their pricing page: Salesforce, Zoho and Vendasta.
- Pipedrive's pricing page served a Cloudflare challenge to every client we tried, browser user agent included.
- HubSpot, Copper and ActiveCampaign render their prices with JavaScript, so a crawler reading raw HTML finds no price at all.
How we ran the audit
We picked 24 sites that buyers compare when they shop for a CRM or a sales tool. For each one we fetched robots.txt and read the rules for 14 AI crawlers: GPTBot, ChatGPT-User, OAI-SearchBot, ClaudeBot, Claude-User, Claude-SearchBot, anthropic-ai, PerplexityBot, Perplexity-User, Google-Extended, Applebot-Extended, meta-externalagent, CCBot and Bytespider. A crawler counts as blocked only when its own group, or the catch-all group when it has none, disallows the whole site.
Next we requested each pricing page four times with curl. One request used a desktop Chrome user agent. The others used the published GPTBot, ClaudeBot and PerplexityBot strings. We logged the HTTP status and checked whether a price appeared in the HTML before any script ran.
The requests came from an ordinary office connection, not from the crawlers' own networks. So a 403 here shows the site has a rule for that user agent. It's the same rule a real crawler meets whenever the site checks the name alone.
Pricing page access by user agent
HTTP status for each request, plus whether a price shows in the static HTML. 24 September 2026.
| Vendor | Pricing page tested | Browser | GPTBot | ClaudeBot | PerplexityBot | Price in static HTML |
|---|---|---|---|---|---|---|
| Salesforce | salesforce.com/sales/pricing | 200 | 403 | 403 | 403 | Yes |
| Zoho | zoho.com/crm/zohocrm-pricing.html | 200 | 403 | 403 | 200 | Partly, most prices load by region |
| Vendasta | vendasta.com/pricing | 200 | 403 | 403 | 403 | Yes |
| Pipedrive | pipedrive.com/en/pricing | 403 challenge | 403 challenge | 403 challenge | 403 challenge | Page not served |
| HubSpot | hubspot.com/pricing/sales | 200 | 200 | 200 | 200 | No, prices load with JavaScript |
| Copper | copper.com/pricing | 200 | 200 | 200 | 200 | No, prices sit in script data |
| ActiveCampaign | activecampaign.com/pricing | 200 | 200 | 200 | 200 | No, prices load in a calculator |
| monday.com | monday.com/crm/pricing | 200 | 200 | 200 | 200 | Yes |
| Close | close.com/pricing | 200 | 200 | 200 | 200 | Yes |
| Attio | attio.com/pricing | 200 | 200 | 200 | 200 | Yes |
| GoHighLevel | gohighlevel.com/pricing | 200 | 200 | 200 | 200 | Yes |
| Apollo | apollo.io/pricing | 200 | 200 | 200 | 200 | Yes |
| Clay | clay.com/pricing | 200 | 200 | 200 | 200 | Yes |
| folk | folk.app/pricing | 200 | 200 | 200 | 200 | Yes |
| Instantly | instantly.ai/pricing | 200 | 200 | 200 | 200 | Yes |
| Smartlead | smartlead.ai/pricing | 200 | 200 | 200 | 200 | Yes |
| lemlist | lemlist.com/pricing | 200 | 200 | 200 | 200 | Yes |
| SuiteDash | suitedash.com/pricing | 200 | 200 | 200 | 200 | Yes |
| Keap | keap.com/pricing | 200 | 200 | 200 | 200 | Yes |
| GigRadar | gigradar.io/pricing | 200 | 200 | 200 | 200 | Yes |
| Twenty | twenty.com/pricing | 200 | 200 | 200 | 200 | Yes |
| Day.ai | day.ai/pricing | 200 | 200 | 200 | 200 | Yes |
| Lightfield | lightfield.app/pricing | 200 | 200 | 200 | 200 | Yes |
| Streak | streak.com/pricing | 200 | 200 | 200 | 200 | Yes |
Checked 24 September 2026 with curl, following redirects. 200 = page served; 403 = refused.
robots.txt rules and content signals
Only sites with a whole-site AI-crawler block or a Content-Signal line are listed. The other 19 allow all 14 crawlers.
| Vendor | AI crawlers disallowed site-wide | Content-Signal line |
|---|---|---|
| Attio | CCBot, Bytespider (GPTBot, ClaudeBot, PerplexityBot and Google-Extended are allowed by name) | None |
| SuiteDash | Bytespider | ai-train=yes, search=yes, ai-input=yes |
| ActiveCampaign | None (its pricing configurator path is disallowed for all bots) | ai-train=no, search=yes, ai-input=no |
| Close | None | ai-train=no, search=yes |
| monday.com | None (path-level rules only) | ai-train=yes, search=yes, ai-input=yes |
robots.txt files fetched and parsed on 24 September 2026. HubSpot's help site, knowledge.hubspot.com, also disallows Bytespider.
Help centres are harder to read than pricing pages
Plan limits often live in the help centre, not on the pricing page. That's where you'd find automation caps, seat minimums and credit rules. Four help centres in the set served a Cloudflare challenge to plain HTTP clients when we checked in September 2026: support.monday.com, help.activecampaign.com, knowledge.apollo.io and Keap's help site, which now redirects to learn.thryv.com. help.salesforce.com returned a 403 to the ClaudeBot user agent.
When an assistant can't read the vendor's own page, it answers from whoever explains the page instead. In our September 2026 probes, every answer about these vendors cited three to eight independent explainers beside the vendor. We map those sources in which sources ChatGPT and Claude cite, and we first wrote up the pattern in what answer engines actually cite.
Check a vendor's price before you trust an AI answer
If a pricing page returns a 403 or renders its prices in a script, the figure your AI assistant quotes came from somewhere else. Open the vendor's page in a browser before you sign. Our CRM pricing index lists each price with its source URL and check date, and the Pipedrive cost page covers the one pricing page no plain client could open.
Questions
- Does blocking AI crawlers keep a CRM vendor out of ChatGPT answers?
- Not in our tests. In nine search-mode probes about Salesforce, Pipedrive, Zoho, monday, ActiveCampaign, Vendasta and HubSpot, run in September 2026, every answer still cited the vendor's own domain. Each answer also cited three to eight independent explainers, which is where the unreadable pricing detail came from.
- What is the difference between a robots.txt block and a user-agent block?
- A robots.txt rule is a published request that a named crawler stay away. A user-agent block happens at the server or firewall, which returns an error such as 403 when a request names a bot. Our audit found no robots.txt blocks on the main AI assistants and three user-agent blocks on pricing pages.
- What does a Content-Signal line in robots.txt mean?
- It's a Cloudflare convention that states how a site wants its content used. ai-train covers model training, search covers search indexing and ai-input covers use inside an AI answer. In September 2026 ActiveCampaign was the only site in the set to set ai-input=no.
- Can I repeat this test myself?
- Yes. Fetch each site's robots.txt, then request the pricing page with curl four times, once with a browser user agent and once each with the GPTBot, ClaudeBot and PerplexityBot strings. Record the HTTP status and whether a price appears in the raw HTML.
- Can AI assistants read salescrew.io's pricing?
- Yes. The robots.txt on salescrew.io allows GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended by name. Every page is prerendered to static HTML, so the $299 Core price and the unlimited-seats rule are in the page source.
A CRM whose price any assistant can read
SalesCrew publishes one price per tier in static HTML: Core is $299 a month with unlimited seats.