Website audit
Trusted on sight, hard to shortlist
An agency with enterprise-grade credentials selling to marketing leads who need a business case. Two of the five facts that buyer needs are missing from the site entirely — so the simulated visitors were briefed to go looking for them.
Cold visitor
Marketing Manager, B2B SaaS scaleup
Arrived
Clicked a LinkedIn post. No prior brand awareness.
Wants
Decide in ten seconds whether this agency is for a company her size.
Judges on
Does the hero name an industry and an offer? Do the logos back it up?
Ready to act
Head of Marketing, mid-market SAP partner
Arrived
Google search for “webflow migration agency australia”. Third result.
Wants
Migrate off a legacy CMS this quarter; needs timeline and rough cost before booking a call.
Judges on
Timeline in writing, an indicative budget, and evidence they have done migrations before.
Comparison shopper
Marketing Director, enterprise services
Arrived
Peer referral. Comparison document already open with two rival agencies in it.
Wants
Build a case her CMO signs off when someone asks “why them?”
Judges on
Quantified outcomes, results inside testimonials, anything slide-ready.
AI agent
AI assistant acting for a marketing director
Arrived
Direct fetch. Dispatched with a task and no brand context.
Wants
Return typical timeline, indicative cost and a contact route for three Australian Webflow agencies.
Judges on
Readable without JavaScript, facts extractable, a non-form contact channel, and the crawl policy.
Cold visitor
Priya Raman
/
lands on the homepage
Ten seconds. Convince me.
AI agent
Research agent
/
GET / — parsing
No JS shell, no empty container. I can actually read this one.
Ready to act
Daniel Okafor
/faqs
skips the hero, opens the FAQs
I know what I need. Where's the detail?
Comparison shopper
Meredith Shaw
/projects
opens the work listing
Don't just take your word for it — fine. Show me the numbers.
Cold visitor
Priya Raman
/
reads the hero headline
Every agency says they're more than an agency. What do you actually do?
AI agent
Research agent
/faqs
extracting fact 1 of 3 — timeline
8–10 weeks Webflow, 10–16 for complex migrations. Retrieved. High confidence.
Ready to act
Daniel Okafor
/faqs
finds FAQ 21
A real number, in writing, without a call. First site this week that's told me anything.
Comparison shopper
Meredith Shaw
/projects
scans six project cards
Good logos. Where are the results?
Cold visitor
Priya Raman
/
reads the subhead
Start-up, scale-up and enterprise? So you'll take anyone. That's the opposite of reassuring.
AI agent
Research agent
/what-we-do
searching the services page for pricing
No rates, no packages, no model. Continuing.
Ready to act
Daniel Okafor
/faqs
searches all 31 FAQs for cost
You told me how long it takes but not what it costs. I can't take a timeline to my CFO.
Comparison shopper
Meredith Shaw
/works/our-guard-mutual-insurance-website-redesign
opens the Our Guard case study
Ten pages, eight CMS collections, a four-stage buyer journey. This is good writing.
Cold visitor
Priya Raman
/
hits the client logo wall
SAP. IBM. Juniper. Football Australia. Okay — they're real.
AI agent
Research agent
/llms.txt
GET /llms.txt — crawl policy
robots.txt was permissive and pointed here. Checking what it actually says.
Comparison shopper
Meredith Shaw
/works/our-guard-mutual-insurance-website-redesign
reaches the outcome section
The quality of the conversation that follows a website visit. That is not a result.
AI agent
Research agent
/llms.txt
parsing directives
ClaudeBot: Disallow. GPTBot: Disallow. Wildcard: Disallow. The notes claim indexing is allowed.
Ready to act
Daniel Okafor
/faqs
reads the SEO migration answer
301s, metadata, Core Web Vitals. Written by someone who has actually done this.
Cold visitor
Priya Raman
/
finds the specialities list
SaaS, tech scaleups, enterprise. This is the sentence I needed at the top, not buried in a list of fourteen.
AI agent
Research agent
/marketing-assets/ai-search-readiness-checklist
cross-referencing the same domain
An AI Search Readiness Checklist — sold from behind a file that tells me to go away.
Comparison shopper
Meredith Shaw
/projects
checks the other 27 studies
Not one percentage. Anywhere. My CMO's first question is going to be: did it work?
Ready to act
Daniel Okafor
/
looks for a booking link
Book a Meeting in the nav, but the contact page is a form. Which one is faster?
AI agent
Research agent
/projects
GET a case study URL — 404
Nav sends humans to /projects, case studies live at /works/. Dead link in the sitemap too.
Cold visitor
Priya Raman
/
reads a testimonial
Head of BusinessOne at SAP. That's a real person with a real title. Good.
Comparison shopper
Meredith Shaw
/
falls back to the testimonials
Everyone's a pleasure to work with. Nobody says what changed.
Ready to act
Daniel Okafor
/contact-us
reaches the contact page
Phone and email in plain text, seven-step process laid out. Honest. I'll call.
AI agent
Research agent
/
compiling for its user
Fact 2 of 3 unavailable. Reporting timeline and contact route only.
Cold visitor
Priya Raman
/
scrolls back to the top
I'd probably enquire. But the logos convinced me, not the writing.
Comparison shopper
Meredith Shaw
/
closes the tab
I think LimeHub might be the better shop. I just can't prove it to anyone else.
Ready to act
Daniel Okafor
/contact-us
weighs it up
Going in blind on budget, and so are they. That's a wasted half hour for both of us.
Nav Detail binds here.
Primary navigation is clear and conventional, and the 31-question FAQ hub is unusually well organised for an agency site. Two things pull the score down. “Our Work” in the nav resolves to /projects, but all 28 case studies live under /works/ — so a URL built from the listing page fails. A case-study URL declared in sitemap.xml also returned 404 on this run. Resources are split across /marketing-assets/ and root-level paths with no obvious rule.
The hero could be said by any agency in Australia, and the subhead widens the audience rather than narrowing it. The precise ICP does exist on the site — thirty questions deep in the FAQ — which means the clarity problem is placement, not thinking.
Exceptionally strong, and the reason the cold visitor stayed on the page. Named enterprise logos, government supplier status, two certified platform partnerships, and eight testimonials attributed to named people with real titles at real companies. An all-onshore team is stated explicitly, and phone and email sit in plain text in the footer. Very little is missing here.
The weakest category and the one that costs deals. Across 28 case-study pages there is no conversion lift, traffic figure, pipeline number or time saving. The Our Guard study is genuinely well written and states outright that success is measured qualitatively. The testimonials are warm and specific about the working relationship, and silent on outcomes.
Genuinely good in places. The phone number and email are real tel: and mailto: links, and the seven-stage process from Intro Meeting through to “Delivery and Beyond” removes a lot of anxiety before a first call. Against that: no pricing or budget signal anywhere on the site, “Book a Meeting” in the nav resolves to a contact form rather than a calendar, and the form's own fields were not visible to text extraction.
Scored 4.0 out of 10. The working, so it can be checked:
Final: 4.0 of 10.
Ranked by likely impact. Each finding quotes the page it came from.
Critical
Machine readability
The AI agent found /llms.txt reading as a blanket block on AI crawlers — naming GPTBot, ClaudeBot, Google-Extended and PerplexityBot, plus a wildcard Disallow — while the same domain sells an AI Search Readiness Checklist. The file's own notes contradict its own directives.
Forget clicks on Google ads. The new fight is for a place in AI's answers. If your website isn't optimised for AI search and conversions, you risk being invisible.
View the pageWhat to do
Critical
Proof depth
Across 28 project pages the comparison shopper found no conversion lift, traffic figure, pipeline number or time saving — and closed the tab saying she could not prove the choice to anyone else.
the measure of success is about the quality of the conversation that follows a website visit
View the pageWhat to do
Serious
Message clarity
The cold visitor bounced off the headline and was only rescued by the logo wall four sections later. The precise version already exists on the site, thirty questions deep in the FAQ.
More than just a Marketing Agency — we move Brands Beyond the Ordinary
View the pageWhat to do
Serious
Conversion path
Both the ready-to-act buyer and the AI agent retrieved a precise timeline and then hit nothing at all on cost. Price is absent from the services page, all 31 FAQs and every case study.
Most Webflow projects run 8–10 weeks from kick-off to launch
View the pageWhat to do
Worth fixing
Navigation
The nav sends humans to /projects, but all 28 case studies live under /works/. The AI agent guessed the wrong path and 404'd, and so will anyone building a URL from the listing page — including crawlers indexing your work.
Don't just take our word for it – check out some of the projects we've worked on.
View the pageWhat to do
Worth checking
Navigation
The 100,000-page Webflow Enterprise migration case study is listed in sitemap.xml but returned 404 when fetched. It may be transient or a stale entry — verify before acting. If that study exists it is the strongest proof asset on the site.
/works/migrating-100-000-web-pages-in-webflow-enterprise
View the pageWhat to do
Worth checking
Conversion path
The contact page's headline, copy, phone and email extracted cleanly, but the form's fields, labels and submit text did not. That may be a limit of this run's method — or it may be how an AI agent acting for a buyer experiences the page.
Ready to Move Beyond Ordinary?
View the pageWhat to do
Keep doing
Proof depth
These are the most useful things on the site. The ready-to-act buyer visibly relaxed on reading them, and the AI agent retrieved them cleanly as task facts. Almost no agency in Australia commits to this publicly.
Most brand identity projects run 8–12 weeks from kick-off to final delivery
View the pageWhat to do
Keep doing
Conversion path
The phone number and email sit in the footer as real tel: and mailto: links, so the AI agent could hand its user a usable next step without submitting a form — a fallback most B2B sites no longer offer.
1300 375 336 · info@limehub.com.au
View the pageWhat to do
All findings with their tasks, the four visitor journeys and the methodology. We will email it within one business day.
Four simulated visitors — three human buyer types and one AI agent acting for a user — were run against this site's real conversion event: booking an intro meeting via the contact form, phone or email.
What this run did. Live content pulled from 7 pages and reasoned over: structure, copy, navigation, calls to action, trust signals, proof and page-to-page consistency. robots.txt, llms.txt and sitemap.xml were fetched directly to test how the site treats automated and AI traffic. That the page text returned fully formed without executing JavaScript is itself the evidence behind the AI agent's readability finding.
What this run could not do. No live browser was connected, so there are no screenshots and no visual or mobile rendering check. No load-time or Core Web Vitals data was measured — run PageSpeed Insights separately. The contact form's fields could not be inspected. The verbatim contents of llms.txt could not be retrieved in full, so that finding rests on two consistent summaries: read the raw file before acting on it. Only 7 of 217 sitemap URLs were examined, and one of two attempted case-study URLs returned 404, which may be transient.
These findings are LimeHub's opinion based on observation at a point in time, not a statement of fact about the business's performance.