Skip to content
HubSpot

vantagepoint.io: from 81 to 100 on HubSpot Website Grader in two days

How we took vantagepoint.io from 81 to 100 on HubSpot Website Grader in two days: 74 requests down to 26, page speed 1.7 seconds, every change reversible.

Vantage Point case study cover: vantagepoint.io from 81 to 100 on HubSpot Website Grader
Vantage Point case study cover: vantagepoint.io from 81 to 100 on HubSpot Website Grader

On October 10, 2026, HubSpot's Website Grader scored vantagepoint.io 81 out of 100. Two days later it scored 100: 30 of 30 for performance, SEO and mobile, and 10 of 10 for security. We did the work on our own HubSpot site with the same process we use for clients, and every change can be rolled back.

Where we started

Website Grader runs Google Lighthouse against your homepage on a simulated mobile phone and scores four areas. Here is our first run next to our latest.

AreaOctober 10October 12
Performance21/3030/30
SEO30/3030/30
Mobile20/3030/30
Security10/1010/10
Total81100

Four checks failed on the first run. The homepage made 74 requests when the Grader wants fewer than 30. It took 6.1 seconds to become usable on a phone, against a target of 5.3. One JavaScript file wasn't minified. And the legal links in our footer were too small and too close together to tap reliably.

What we changed

We worked in rounds. Each change was tested on a lab copy of the page first, the file it touched was backed up, and only then did it go live. We re-ran the Grader after each round.

Round 1: the obvious failures (score 92)

  • Footer legal links made 48 pixels tall with spacing between them, which cleared the tap target check.
  • Our site search widget's script minified, from 81 KB to 58 KB, and loaded on the first scroll, tap or key press instead of with the page.
  • An animation library pulled from a third-party CDN switched off.
  • Partner logos and a background graphic loaded only as they come near the screen.

Rounds 2 and 3: fewer files (score 94)

  • Ten partner logos combined into one image sprite, so ten downloads became one.
  • The header logo drawn once instead of twice.
  • Six small icons drawn with CSS instead of separate image files.
  • Seven section stylesheets moved into the page head so the browser styles the page in one pass.
  • HubSpot's capture of non-HubSpot forms turned off, after we checked all 2,591 URLs in our sitemap and confirmed every real form already submits to HubSpot directly.
  • Our product analytics script set to load on first interaction or after 8 seconds.

Round 4: HubSpot's own scripts wait their turn

HubSpot's script loader brings in tracking, the cookie banner, chat, pop-up CTAs and the ads pixel on every page. We changed the site template so the loader starts on a visitor's first scroll, tap or key press, or 10 seconds after the page loads, whichever comes first. That single change took the homepage from 38 requests to 25 and roughly halved the time before the page responds to a tap.

Last fix: a right-sized logo

Lighthouse then flagged one image: our header logo was a 620 by 274 pixel PNG displayed at 118 by 52. We replaced it with two small WebP files, 160 and 240 pixels wide, and let the browser pick the one it needs. The next Grader run read 100.

Before and after on the homepage

MeasureBeforeAfter
Website Grader score81100
Page requests (Grader)7426
Page speed (Grader)6.1 seconds1.7 seconds
Page weight (Lighthouse, mobile)596 KB244 KB

The tradeoff we accepted

Holding back HubSpot's loader has a cost, and we chose it on purpose. A visitor who leaves within 10 seconds without scrolling, tapping or pressing a key is not recorded in HubSpot analytics. The cookie banner and chat appear on first interaction or at 10 seconds rather than the instant the page opens. Before we published, we confirmed that tracking, the banner and chat all load after a scroll and after 10 seconds of idle time, and that the forms on our assessment and package pages still render.

We also turned down a more aggressive option that would have cut more requests but dropped more pageviews. Firms in regulated industries care about what a change does to tracking and consent, so we put those decisions in front of the client before we touch anything.

What a 100 does and doesn't tell you

The Grader tests one page, your homepage. Lighthouse timings also move around from run to run, so we judged each change on several runs rather than one. We ran the same checks across 23 page types on our site: average requests fell from 49 to 38, and a few heavier pages, including our main conversion page, still need work. We're working through those next.

Get the same review for your site

We now offer this as the Website Grader Tune-Up. It starts with a free review: we run Website Grader and Lighthouse on your homepage and the pages that matter most to you, list what's failing and why, and mark the fixes that affect tracking, consent or chat so your team decides on those first. The fixes are then quoted and built in your HubSpot portal, tested before they go live, with a backup of every file we change.

Request a free review, or see how we work with HubSpot Content Hub.

Your project

Want results like these in your firm?

Talk with a senior consultant about your Salesforce or HubSpot goals. We'll tell you what a similar project would involve and how long it would take.

Book a free CRM assessment

Not ready to talk? Get new articles on Salesforce, HubSpot and AI for regulated firms by email.

Latest Articles

vantagepoint.io: from 81 to 100 on HubSpot Website Grader in two days

How we took vantagepoint.io from 81 to 100 on HubSpot Website Grader in two days: 74 requests down to 26, page speed 1.7 seconds, every cha...

First Lien Capital: from HubSpot trial to Marketing Hub Pro in 36 days

How First Lien Capital, HubSpot and Vantage Point ran a guided Marketing Hub trial that delivered 10,828 emails and ended in a Pro upgrade ...

How One HubSpot Workflow Created 250K Marketing Contacts Overnight

A misconfigured HubSpot workflow turned 12,000 marketing contacts into a quarter-million overnight. See how to diagnose, reverse, and preve...