Public teardown21 Oct 2026
HubSpot logo

HubSpot

United States
hubspot.com·Programming and Developer Software·Built with HubSpot
Audit another site
Why this teardown

HubSpot practically wrote the inbound marketing playbook, so its own homepage is a useful benchmark for what 'good' looks like. On 21 October 2026 the homepage scored 83/100. Security is the strong suit at 100; speed is the weak one at 58. Of six questions a customer might ask an AI assistant, HubSpot is likely to be cited for 5. Autopilot could fix 1 of the 6 issues without a developer.

Overall grade
BGood

6 issues found · 1 fixable by Autopilot

Human score

Speed, clarity, accessibility and trust for the people who visit.

AI score

Whether ChatGPT, Perplexity and Google AI can read and cite you.

83 → 85 if you let Autopilot fix 1 thing

Every change is previewed first and can be undone in one click.

Scores

Start here

The 3 issues costing you most

Medium· 3

What's wrong

An assistant reads the first few hundred words to decide what your site is. Right now that text doesn't plainly state your name, what you do and for whom.

The fix

Open with one plain sentence: "HubSpot is a CRM and marketing platform for sales and marketing teams." Autopilot can draft it for approval.

Guided fix in WebValetImpact on score: about +4
Screenshot of hubspot.com123
1The first screen doesn't say who you are224 of 86 images have no description3Too many scripts
Score breakdown

Where the points went

Seven areas, weighted by how much they matter for getting found and converting. Each square is one check.

17

of 17 lost

Found on Google

25% of the score

95
1 issue

Google can read and rank this page well. Titles, headings, links and the files Google reads.

Speed

20% of the score

58
2 issues

Slow on phones, which costs rankings and visitors. How fast the page loads on a phone.

Found by AI

20% of the score

73
1 issue

Assistants can read you but have little to confirm who you are. Whether ChatGPT, Perplexity and Claude can read and cite you.

Accessibility

10% of the score

88
1 issue

Works well for screen readers and keyboards. Screen readers, keyboards and readable text.

Security

10% of the score

100
No issues

Secure connection and sensible headers. HTTPS and the headers that protect visitors.

Converts visitors

10% of the score

94
No issues

Clear next step and proof for visitors. Whether visitors know what to do next.

Safe and legal

5% of the score

90
1 issue

Policies and consent are in place. Privacy, terms and cookie consent.

Speed on mobile

Speed signals

Google PageSpeed didn't answer in time, so these come from our own fetch. Connect your site and WebValet re-runs the full lab test.

Server responseGood

150 ms

Time to the first byte of HTML. Good ≤ 0.8 s

HTML sizeNeeds work

628 KB

The page before images and scripts. Good ≤ 300 KB

Images on pagePoor

86

Each image is a request on mobile data

WordsGood

2144

Readable text Google and AI can quote

Found by AI

What AI assistants see

Six questions a customer might ask ChatGPT, Perplexity or Claude, and whether HubSpot is likely to be the answer today.

ChatGPTPerplexityClaudeGeminiGoogle AIGrok
5 of 6 likely cited
  • “What is HubSpot?”

    Your page states who you are, so assistants can quote it directly.

  • “Is HubSpot any good? Pros and cons”

    Enough readable copy for an assistant to summarise strengths.

  • “Best CRM and marketing platform for sales and marketing teams in United States”

    Strong entity signals make you a candidate in comparison answers.

  • “How much does HubSpot cost?”

    A pricing page plus FAQ or llms.txt gives assistants a source to cite.

  • “HubSpot alternatives”

    Assistants can confirm your identity and category, so you appear in alternative lists.

  • “Who is HubSpot for?”

    The first screen doesn't say who the product is for.

Readiness signals
  • llms.txt

    A plain-text guide for AI crawlers exists

  • Organization schema

    Assistants can confirm who you are

  • FAQ schema

    No marked-up questions to lift into answers

  • Clear 'who we are'

    The first screen doesn't say what you do

Schema found

Organization, WebSite, WebPage, Product

Question-style headings

5

Every issue

6 issues, most important first

Each one in plain English, with the fix and how many points it's worth. Autopilot applies the marked ones on a connected site.

Medium· 3

What's wrong

An assistant reads the first few hundred words to decide what your site is. Right now that text doesn't plainly state your name, what you do and for whom.

The fix

Open with one plain sentence: "HubSpot is a CRM and marketing platform for sales and marketing teams." Autopilot can draft it for approval.

Guided fix in WebValetImpact on score: about +4
Low· 3
What this means for you

HubSpot does the basics well. Speed is where you're leaving results on the table.

Your site scores 83 out of 100, which is solid. Visitors have a good experience (80/100), and AI assistants like ChatGPT can read and cite you (82/100). Security is your strongest area at 100. Speed is the weakest at 58. Autopilot could fix 1 of these issues on its own and lift the score to about 85.

  • The first screen doesn't say who you are: An assistant reads the first few hundred words to decide what your site is.
  • 24 of 86 images have no description: Alt text tells Google Images and screen readers what a picture shows.
  • Too many scripts: 31 external scripts load.

Written from the checks above.

First fix this Monday

The first screen doesn't say who you are

The first screen doesn't say who you are. Open with one plain sentence: "HubSpot is a CRM and marketing platform for sales and marketing teams." Autopilot can draft it for approval.

About 30 minutes by handWorth about +4 points
Get this for your site

This is what HubSpot sees. Now see yours.

Free audit in about 20 seconds, no signup. Then WebValet fixes what it finds and proves each fix with real Google data.

More teardowns