Case Studies

Overview of Our Case Study Process

A client case study requires a real before-and-after number and dates you could verify. We don't have one to publish yet. Instead, we run the same audit engine on well-known public pages and publish the raw findings - so you can evaluate the tool against pages you already know.

Public Teardowns

Detailed Audits of Popular Websites

These companies are not Nebula customers. We run the same 9-signal engine on their public pages and publish the raw findings - evidence of what the audit produces on pages you can verify yourself.

carrd.co

Carrd

4.2/10

Grade D

5 findings · Audited July 30, 2026

H1 contains only the brand name: <h1>Carrd</h1>. Zero keyword signal, zero value prop, zero audience signal. Every major

Read teardown →

hotjar.com

Hotjar

4.5/10

Grade D

4 findings · Audited July 30, 2026

H1 reads "Hotjar has evolved into something more powerful" - corporate announcement copy. First-time visitors see a tran

Read teardown →

kit.com

Kit

4.8/10

Grade D

4 findings · Audited July 30, 2026

Meta description is only 35 characters: "Kit: The creator marketing platform". Leaves ~120 chars of SERP real estate emp

Read teardown →

postdew.com

PostDew

4.8/10

Grade D

5 findings · Audited August 3, 2026

No <h1> tag was found in the fetched document. If the primary promise is rendered only through client-side code, non-Jav

Read teardown →

beehiiv.com

Beehiiv

5.1/10

Grade C

5 findings · Audited July 30, 2026

Four competing CTAs appear before the hero section: (1) "Claim 20% off" banner, (2) "Sign up" nav link, (3) "Sign up wit

Read teardown →

unbounce.com

Unbounce

5.2/10

Grade C

4 findings · Audited July 30, 2026

H1 reads "Launch faster. Convert more." - this exact promise appears on Leadpages, Instapage, Swipe Pages, and Carrd. A

Read teardown →

basecamp.com

Basecamp

5.4/10

Grade C

5 findings · Audited July 29, 2026

<title> tag is only 8 characters - "Basecamp" - too short to signal topic relevance. Meta description is 185 chars, trun

Read teardown →

drift.com

Drift

5.4/10

Grade C

4 findings · Audited August 2026

The transition notice, "We've transitioned from Drift to 1mind", appears mid-page in a callout section, not above the fo

Read teardown →

www.folioverse.app

Folioverse

5.6/10

Grade C

5 findings · Audited August 3, 2026

No <h1> tag was found in the fetched document. If the primary promise is rendered only through client-side code, some re

Read teardown →

canva.com

Canva

5.7/10

Grade C

5 findings · Audited August 5, 2026

The served document is the "unsupported browser" fallback: title "Unsupported client – Canva" (26 chars, below the 120-c

Read teardown →

calendly.com

Calendly

5.8/10

Grade C

5 findings · Audited July 29, 2026

No H1 tag detected in static HTML. The primary headline is rendered via JavaScript, meaning search engines and link prev

Read teardown →

webflow.com

Webflow

5.8/10

Grade C

4 findings · Audited July 30, 2026

Two H1 tags are simultaneously present in the live DOM via Webflow Optimize: "Make your website a growth engine" and "Ma

Read teardown →

knallhart.dev

knallhart.dev

5.8/10

Grade C

3 findings · Audited July 31, 2026

The page has zero Open Graph tags (og:title, og:description, og:image) and zero Twitter Card tags. When the founder shar

Read teardown →

clickup.com

ClickUp

5.8/10

Grade C

4 findings · Audited August 2026

"Software to replace all software." names no product category, no audience, and no measurable outcome. A first-time visi

Read teardown →

notion.so

Notion

6.1/10

Grade C

4 findings · Audited July 29, 2026

Homepage HTML payload is 417KB - 3.4× the 120KB heuristic for a marketing page. Excessive HTML inflates parse time, dela

Read teardown →

typeform.com

Typeform

6.2/10

Grade C

5 findings · Audited August 5, 2026

No <h1> tag found in the document at all. Search engines, screen readers, and AI crawlers get no primary headline signal

Read teardown →

amplitude.com

Amplitude

6.2/10

Grade C

4 findings · Audited August 2026

Title is 38 characters, the shortest in this teardown batch. "A new era for product teams" contains no product category

Read teardown →

mixpanel.com

Mixpanel

6.4/10

Grade C

4 findings · Audited August 2026

"Build faster, with direction" is slogan copy, not a product claim. It names no category (analytics, product intelligenc

Read teardown →

slack.com

Slack

6.6/10

Grade B

4 findings · Audited August 5, 2026

meta[name=description] is 173 chars - SERPs show ~155. The snippet truncates mid-word at "…the AI work platform for mana

Read teardown →

postmint.de

Postmint

6.8/10

Grade B

2 findings · Audited July 31, 2026

The JSON-LD block's context key is not "@context" - it is a leaked Blade/PHP template literal: "<?php $__contextArgs = [

Read teardown →

raycast.com

Raycast

6.8/10

Grade B

3 findings · Audited September 17, 2026

The audit source proxy did not find a CTA in the first 2,000 characters.

Read teardown →

asana.com

Asana

6.9/10

Grade B

4 findings · Audited August 5, 2026

Two problems in the primary signal: the H1 text reads "The OS forhuman-agentteams" (line-break collapse - missing spaces

Read teardown →

monday.com

Monday.com

6.9/10

Grade B

4 findings · Audited August 5, 2026

The document carries 2 H1 elements - the page's primary topical signal is split. H1: "People and agents working as one t

Read teardown →

intercom.com

Intercom

6.9/10

Grade B

3 findings · Audited August 5, 2026

HTML payload is 1,528.4 KiB - 12.7x the 120KB heuristic and the second-heaviest in our dataset. First paint waits on a m

Read teardown →

hubspot.com

HubSpot

7/10

Grade B

5 findings · Audited August 4, 2026

H1 and title tag share no significant keywords - messaging misalignment. The title is "HubSpot | Software & Tools for yo

Read teardown →

figma.com

Figma

7/10

Grade B

4 findings · Audited August 5, 2026

HTML payload is 1,560.7 KiB - 13x the 120KB heuristic and the heaviest in our dataset. HTML this large delays parse and

Read teardown →

cloudcostwise.io

CloudWise

7.1/10

Grade B

3 findings · Audited August 3, 2026

No recognized ad-tracking artifact was observed in the fetched source HTML. That does not prove runtime or server-side t

Read teardown →

shopify.com

Shopify

7.1/10

Grade B

4 findings · Audited August 5, 2026

H1 and title tag have weak keyword overlap. The title ("Shopify: The All-in-One Commerce Platform for Busi…" - 66 chars,

Read teardown →

mailchimp.com

Mailchimp

7.2/10

Grade B

5 findings · Audited August 4, 2026

The page contains 6 <h1> tags, each with a different value proposition: "Email & SMS marketing minus the learning curve"

Read teardown →

squarespace.com

Squarespace

7.2/10

Grade B

4 findings · Audited August 5, 2026

H1 reads "A websitemakes it real" - a line-break collapse fused "website" and "makes" in the source HTML. Title runs 62

Read teardown →

linear.app

Linear

7.3/10

Grade B

3 findings · Audited September 17, 2026

The H1 source reads "The productdevelopmentsystem for teamsand agentsThe product development system for teams and agents

Read teardown →

loom.com

Loom

7.3/10

Grade B

3 findings · Audited September 17, 2026

Title tag is "Free screen recorder for Mac and PC | Loom", the hero's actual pitch is "Unstuck your team with Loom" and

Read teardown →

miro.com

Miro

7.3/10

Grade B

3 findings · Audited September 17, 2026

H1 reads "The collaboration layer your AI tools are missing.", the primary conversion message is that the visitor curren

Read teardown →

airtable.com

Airtable

7.3/10

Grade B

3 findings · Audited September 17, 2026

The title "Airtable: Build Enterprise-ready AI Workflows, Apps & Agents | Airtable" opens and closes with the brand name

Read teardown →

zapier.com

Zapier

7.4/10

Grade B

4 findings · Audited August 5, 2026

HTML payload is 461.0 KiB - 3.8x the 120KB heuristic, the heaviest in this teardown batch. Every byte delays first paint

Read teardown →

pipedrive.com

Pipedrive

7.4/10

Grade B

3 findings · Audited August 2026

The hero presents both an email capture form ("Enter your work email → Try it free") and a button CTA ("Try it free") at

Read teardown →

smartwatermark.app

SmartWatermark

7.5/10

Grade B

2 findings · Audited August 3, 2026

No recognized ad-tracking artifact was observed in the fetched source HTML. Runtime or server-side tracking remains unkn

Read teardown →

framer.com

Framer

7.8/10

Grade B

4 findings · Audited September 17, 2026

The most visible bottom CTA reads "Start without AI" - on a page entirely positioned around AI-powered website building.

Read teardown →

deel.com

Deel

8.9/10

Grade A

3 findings · Audited September 17, 2026

The fetched HTML is approximately 966 KiB.

Read teardown →

What a client case study looks like

The first client case study we publish will name the page (or explain why it's anonymized), state the actual before/after metric, give the measurement window, and link the underlying evidence. If we can't show our work, it doesn't go here.

Until then: run the audit on a page you know well. See whether the findings match what you already know is true. That is the fastest way to evaluate the engine.

See it on your page

Same engine. Your URL. No signup required.

Find the Leak →