You crawl your website. You add it to Customer Agent's knowledge base. You assume it works. Then a contact asks your agent a question that's clearly answered on your pricing page, and the agent fumbles it.
That's the silent failure this update solves. Before August 25, 2026, a partial or fully failed crawl gave you zero signal. Now it gives you a number, a warning icon, and a full breakdown.
What This Update Actually Is
HubSpot added a Pages column to the Imported URLs view inside Customer Agent's Knowledge tab. That column shows a ratio: pages successfully synced versus total pages discovered during the crawl.
A domain crawl might show something like 1,682 / 1,685. A single-URL import shows 1 / 1 when it's clean. If any pages fail, a warning icon appears on the count. Hovering shows a tooltip with context. Clicking opens a modal.
The modal has four tabs: All, In Queue, Complete, and Failed. Each tab shows per-page sync status and, for failures, the specific reason the page didn't load into the knowledge base.
Why HubSpot Shipped This
The external problem is simple: AI agents are only as good as the knowledge they can actually reach. If a crawl drops pages silently, the agent answers from an incomplete picture. Contacts get wrong or vague answers, and your team can't figure out why.
The internal frustration is real too. Humans who manage these portals already carry a heavy cognitive load. When an AI tool behaves strangely, debugging it feels impossible without visibility. That gap between "I set it up" and "I know it's working" is exactly where trust in AI tools breaks down.
HubSpot shipped this because observability is the foundation of a trustworthy AI layer. You can't improve what you can't see.
How to Use It Step by Step
- Go to Customer Agent in your HubSpot portal. You'll find it under the Service or AI tools navigation depending on your setup.
- Click the Knowledge tab, then open Imported URLs. Every URL or domain you've crawled appears here.
- Check the Pages column next to each imported URL. A clean sync shows a matching numerator and denominator. A mismatch means something got left out.
- Look for the warning icon. If it's there, hover to get a quick tooltip summary of what went wrong.
- Click the count or select "Learn more" in the tooltip to open the full modal. Use the Failed tab to review which specific pages didn't sync and why.
- Fix the root cause on the failed pages (crawl blocks, redirects, auth walls, etc.) and re-crawl as needed. Check the Pages column again to confirm the fix landed.
What It Touches in Your HubSpot Strategy
This update is small in interface terms but large in strategic consequence. Here's where it ripples.
Service Hub and AI Agent Quality
Customer Agent's value is directly tied to the completeness of its knowledge base. A partial crawl that drops product detail pages, FAQ sections, or policy content means the agent will deflect or hallucinate on exactly the questions it should handle confidently. This new visibility layer is your first line of quality control.
Key Takeaway
Run a Pages column audit on every imported URL in your Customer Agent knowledge base. If any ratio shows a mismatch, open the Failed tab before your next agent review call. Silent partial syncs are likely the cause of unexplained bad answers.
Content Operations and Site Health
Failed crawl pages often signal a deeper site issue: bot-blocking rules in your robots.txt, pages behind authentication, redirect chains that time out, or dynamic content that the crawler can't render. Those are content ops and technical SEO problems worth fixing regardless of your AI agent goals.
This pairs well with the Customer Agent source usage report and its source type filter. That report tells you which knowledge sources are actually resolving conversations. This update tells you which sources are fully loaded. Together, they give you a complete picture of your AI knowledge layer.
RevOps and Knowledge Governance
RevOps teams building service escalation workflows depend on the agent doing its job at tier one. If the knowledge base is leaky, the agent escalates more than it should. That inflates support ticket volume and masks the true cost of poor knowledge hygiene.
Key Takeaway
Add a monthly crawl audit to your RevOps calendar. Review the Pages column for every imported domain, triage the Failed tab, and document findings. This isn't a one-time setup task. Site content changes, and crawl health degrades over time.
Multi-Hub Availability
Customer Agent is available across Service Hub, Sales Hub, Marketing Hub, Content Hub, Data Hub, and HubSpot Smart CRM, all at Professional and Enterprise tiers, plus HubSpot Credits. That means this crawl visibility feature is accessible to nearly every team running an AI-assisted conversation workflow.
Who Should Care Most
This update matters most to four types of humans:
- Service Hub admins who own the Customer Agent setup and are accountable for answer quality in support conversations.
- RevOps leads who track deflection rates and need confidence that the AI layer is working from complete information.
- Content ops and web teams whose site architecture directly affects what the crawler can and can't reach.
- Growing companies with large websites (hundreds or thousands of pages) where a silent partial crawl could leave enormous knowledge gaps.
If your company has a website with more than 50 pages and you're using Customer Agent to handle any tier-one support, this update is directly relevant to you right now.
George's Take
We've audited a lot of portals, and one of the most common things we find is an AI agent that's been running on an incomplete knowledge base for weeks without anyone knowing. Humans set it up, it seemed to work, and then it quietly started giving answers that were just slightly off. This update is one of those quiet wins that matters far more than its size suggests. Visibility into your AI layer isn't optional. It's the foundation everything else is built on.
“You can't trust an AI agent you can't inspect. This feature doesn't just show you what failed. It shows you that HubSpot understands the difference between deploying AI and deploying AI responsibly.”
If you're thinking through where the line is between AI as a tool and AI as a decision-maker in your portal, this article on when a tool stops being a tool is worth a read before your next Customer Agent review.
If you want a second set of eyes on your Customer Agent setup, your knowledge base structure, or your overall AI layer in HubSpot, we'd love to dig in with you. Book a strategy call with the Sidekick team and we'll show you exactly what we'd look at first.
Frequently Asked Questions
What does the Pages column in HubSpot Customer Agent show?
The Pages column shows a ratio of successfully synced pages to total pages discovered during a crawl, for example 1,682 / 1,685. A warning icon appears when pages fail. Clicking the count opens a modal with tabs for All, In Queue, Complete, and Failed pages, including specific failure reasons for each page that didn't sync.
Why would pages fail to sync when I crawl a website in Customer Agent?
Pages can fail to sync due to bot-blocking rules in your robots.txt file, pages behind authentication, redirect chains that time out, or dynamic content the crawler can't render. The Failed tab in the sync modal shows the specific reason for each failed page, so you know exactly what to fix before re-crawling.
Does a partial crawl affect the quality of Customer Agent's answers?
Yes. Customer Agent only answers from what's in its knowledge base. If key pages like pricing, FAQs, or product details fail to sync, the agent will give incomplete or inaccurate answers on those topics. Running a regular crawl audit using the new Pages column is the best way to prevent silent knowledge gaps.
Which HubSpot plans include Customer Agent and this new crawl visibility feature?
Customer Agent, including the new crawl sync status feature, is available on Professional and Enterprise tiers across Service Hub, Sales Hub, Marketing Hub, Content Hub, Data Hub, and HubSpot Smart CRM. It's also available through HubSpot Credits. Free and Starter plans are not included.
How often should I audit the crawl sync status in Customer Agent?
A monthly audit is a solid baseline. Website content changes regularly, and pages that synced cleanly last quarter may fail after a site update, redirect change, or new authentication rule. Add the Pages column review to your RevOps or service ops calendar as a recurring task, not a one-time setup check.
How is this different from the Customer Agent source usage report?
The source usage report shows which knowledge sources are resolving conversations. The crawl sync status feature shows whether those sources are fully loaded in the first place. They answer different questions and work best together: one tells you what's working in production, the other tells you if your knowledge base is complete.
