Key Plus AEO Assessment Report
by Novastacks AI

Request any English product URL on keyplus.my — the homepage, the car insurance page, the e-hailing page, the FAQ — and the server returns the same 2,789-byte file. Byte-identical, verified by checksum. It carries the same title, “Key Plus - Trusted Insurance Providers in Malaysia”, on all four; it carries no description, no main heading, no sub-headings and 8 readable words. Everything a customer sees is assembled afterwards, in the browser. Systems that read the page without running a browser — which includes most AI retrieval — see the 8 words.

Two further rules compound it. The instruction file that tells crawlers where the content is declares one sitemap of 12 URLs; the 43 guides that earn every AI citation live in a second sitemap that is never declared — roughly 47 of the site's 59 content URLs go undeclared. And a blanket rule blocks the entire loan section, which takes down /en/loan/estimation — a page that today holds positions 68 to 86 across seven car-loan calculator terms, five of them at 49,500 searches a month. It is the closest thing Key Plus has to the calculator assets BJAK uses to win this category, and it is switched off.

The rendering gap is not universal in this market: BJAK serves 694,978 bytes of finished HTML with 846 readable words. PolicyStreet has the same problem as Key Plus and is arguably worse — 98,308 bytes containing one readable word.

Because every product URL is the same file, every product URL carries the same title and an empty description. A search engine choosing what to display for “renew car insurance online” and a model choosing what to quote both work from those signals first. Right now they are identical across the whole commercial site, so nothing distinguishes the e-hailing page from the FAQ page from the homepage.

The blog is the opposite: a distinct title, a written description, one main heading and 18 sub-headings across 1,017 words, with a comparison table and four question headings. That structure is precisely why all nine AI-cited pages come from there. But it too has a gap — the guide contains zero internal links, so a reader (or a crawler) who lands on the e-hailing guide has no path to the page that sells e-hailing cover. The content and the commerce never touch.

BJAK links 129 times from its homepage and 25 times from a single blog post. That is how it converts a 4,071-post library into rankings on the pages that take money.

Structured data is the machine-readable summary a page ships about itself. On the Key Plus blog it is well done: organisation, site, breadcrumb and article records, three blocks per page. On every product page it collapses to a single organisation block — no site record, no breadcrumb, no service or product record, and no question-and-answer record anywhere on the domain despite the guides already being written in question-and-answer form. BJAK ships eight blocks on its homepage including a service record, and question-and-answer markup on its help page.

Page experience is the second half of this. The keyplus.my homepage starts painting quickly — FCP 1.1 s — but its main content is not finished until LCP 4.5 s, and the layout keeps moving while it loads: CLS 0.508, five times the 0.1 threshold Google treats as acceptable. That combination is the visible symptom of browser-side assembly, and it produces an overall performance score of 51 out of 100. BJAK scores 95 with its largest element finished in 1.1 seconds and effectively no movement. PolicyStreet scores 55 and is slower still on paint, though its layout is stable. One bright note: on search-fundamentals Key Plus scores 92, ahead of BJAK's 85 — the basics are configured correctly, there is simply nothing behind them.

We tested eight live Malaysian queries — six diagnostic prompts plus the three commercial terms that define the category. The pattern is sharp. On the question-shaped queries Key Plus ranks: #9 on checking an NCD, #5 on checking road tax, #5 on “best website to compare quotes and renew road tax”, #8 on the three-way comparison. On the three commercial terms — “car insurance malaysia”, “compare car insurance malaysia online”, “e-hailing insurance malaysia” — Key Plus is outside the top 20 on all three.

And on every question query where it does rank, the page that ranks is a blog guide. Not one commercial page places anywhere. The brand that sells insurance comparison is visible as an explainer and invisible as a place to buy.

One more detail worth naming: on the NCD query, all four results above Key Plus — fincrew.my, MyEG, paultan.org and BJAK — answer with an interactive checker. Key Plus answers with an article. Eight queries is a diagnostic signal, not a census, but it points the same way each time.

Of the 150 keywords driving the most traffic to keyplus.my, 88 land on blog guides and 62 land on the product stack. But 34 of those 62 — more than half the commercial keyword footprint — point at one URL: the loan estimation page, which robots.txt blocks. The brand's second-largest keyword cluster by search volume is car loan calculators, held at positions 68 to 86 on a page crawlers are told to ignore.

So the identity a machine assembles from the visible evidence is: an explainer site about NCD checks, road tax and insurer comparisons. The identity the business is actually building — a place to compare, renew and finance — sits behind pages that return eight words, or behind a rule that says do not look. The strongest single asset in the entire dataset, “ncd check” at #4 with 12,100 monthly searches, is a blog post about how to check your NCD, not a tool that checks it.

Across the highest-volume keywords Key Plus shares with each competitor, the split runs along one clean line. Where the query is a checking task and the answer is a guide, Key Plus wins — “ncd check” at #4 against BJAK's own checker tool at #8 and PolicyStreet at #48, worth 797 of the domain's 2,297 estimated monthly visits from a single keyword. Where the query is a renewal or a quote, Key Plus loses to PolicyStreet by 20 to 50 positions: “car insurance renewal” #30 versus #5, “motor insurance renewal” #61 versus #8, “car insurance quotes” #25 versus #6.

That is not a content-quality gap. Every one of the renewal and quote queries should be answered by a commercial page — and Key Plus's commercial pages carry no title of their own, no description and no readable copy. PolicyStreet, whose product site has the same rendering problem, wins these anyway on the strength of a 289-URL blog and 401 declared URLs. The gap against BJAK is wider still: 1,848 ranked keywords against 208, and 71,935 estimated monthly visits against 2,297.

A repositioning programme here does not start from zero. Key Plus outranks BJAK's purpose-built NCD checker on the biggest keyword in the set. It owns its brand term at #1. It beats BJAK head-to-head on road tax. Google's AI Overview has already quoted a Key Plus product page — /en/e-hailing-insurance — alongside Grab, Allianz and ringgitplus.com, which proves the machines will use commercial pages when they can read them. The site already ships llms.txt at two locations, ahead of most competitors on that signal. And 165 of the 208 ranked keywords are new, with 34 improving against 7 declining: the trajectory is upward.

The through-line is that the hard part — writing content good enough to be quoted — is done. What is left is plumbing.