How SmallBizAI.au Gets Cited by AI 500+ Times a Day and What We’ve Learned
Posted: June 3, 2026 Filed under: Personal, smallbizai.au | Tags: smallbizai.au, technology, writing Leave a commentWe launched SmallBizAI.au on March 6, 2026. In the first week, Bing Copilot cited us 13 times. By late May, it was citing us over 500 times a day. We didn’t build an SEO strategy around AI citations. We didn’t know that was a thing yet. But after tracking 20,000+ citations across three months, some clear patterns have emerged. And they repeat. What content AI models actually pull from is pretty specific. Most sites aren’t getting cited even though they probably should be.
The short version
AI citation systems are not Google. They don’t reward age, domain authority, or backlink counts the same way. What they reward is specificity. A page that directly answers “Zeller vs Square for a café in Melbourne” beats a page titled “Best payment tools for small business” every time. Most sites are still optimising for Google. That’s the wrong target.
What actually gets cited
Here’s our top cited content as of June 2026:
Notice what’s not there. No “ultimate guide to AI for small business.” No broad overview posts. The highest-cited content is either a dedicated company profile or a direct comparison between named tools.
Why AI cites comparison posts
When someone asks Bing Copilot “should I use Zeller or Square for my business,” the AI needs a source that directly answers that question. A post called “Zeller vs Square” is an obvious candidate. A post called “Best Payment Tools” is not. Too broad to cite with confidence. This is the core difference between traditional SEO and AI citation. Google rewards comprehensive coverage. AI rewards direct answers to specific questions. The query that drives citations is usually a comparison or a company lookup. Not “what is AI” but “is Rippling worth it for a 10-person business in Australia.”
The Zeller effect
One post on Zeller has been cited across roughly 25 different query variants. Not 25 clicks, 25 different questions that all route to the same page.
Queries like:
- “zeller business account review”
- “zeller vs square australia”
- “is zeller good for small business”
- “zeller fees australia”
- “how does zeller work”
All pointing to one URL.
This happens when a post answers multiple angles of the same topic, the company overview, the pricing, the comparison, the use case. Bing learns that this page is the reliable answer for anything Zeller-related and starts routing all those queries there. We call this cluster anchoring. One strong post becomes the hub for an entire query cluster, worth more than 10 thin posts on the same topic.
What doesn’t get cited
Our grants post gets consistent human traffic, people actively searching for Australian small business grants, clicking through, reading it properly. Bing barely touches it. Maybe 60–80 citations total. Why? Because AI assistants don’t answer “where can I get a grant” by citing a directory. They either tell you to check the government website directly, or they summarise. Our page doesn’t fit the format of an answer AI can pull from. Content humans search for isn’t automatically content AI will cite. The format matters as much as the topic.
Content AI cites well:
- Direct tool comparisons (“X vs Y vs Z”)
- Company profiles with clear factual structure (what it does, what it costs, who it’s for)
- “How much does X cost in Australia” – specific country context with a real number
- “Best X for [specific use case]” – named tools, named context
Content AI cites poorly:
- Broad overviews with no specific answer
- Lists of 20+ tools without clear recommendations
- News recaps (cites the original source instead)
- Content that requires context from other pages to make sense
The format that works
Our top-cited posts share a structure. They open with the direct answer. Not “in this post we’ll explore” the actual answer in the first two paragraphs. If someone asks “is Zeller good for small business,” the page answers that in the first 100 words. They use named tools throughout. Not “payment platforms” Zeller, Square, Stripe. AI systems index on entity names. If your post discusses payment tools without naming them, it won’t get pulled for queries about those tools. They include Australian context. “Fees in Australia,” “available to Australian businesses,” “works with Xero Australia.” Bing’s AI is serving Australian users. Pages that signal Australian relevance get pulled for Australian queries. They have a clear verdict. Not “it depends”, an actual recommendation, with the caveat folded in. “Zeller is the better pick if you’re a hospitality business taking in-person payments at volume. Square makes more sense if you also sell online.”
The numbers don’t equal traffic
Flare HR has 1,548 Bing AI citations. In the same period, it had 23 page views from human visitors. Bing Copilot is citing our content to answer user questions, but those users aren’t clicking through to our site. They’re getting the answer from the AI, which pulled it from us, and moving on. Citations build brand recognition even without clicks. And some pages do both, Stripe vs Square vs Tyro has over 1,000 citations and meaningful human traffic. Those are the sweet spot posts.
But if you’re building a content strategy purely for AI citations expecting traffic to follow, you’ll be disappointed. Citations are exposure, not visits. The sites that do well publish enough citation-worthy content that AI systems start treating them as a default source, then drive human traffic through practical posts on the same topics.
The pace matters
We published consistently from day one. Not perfectly (some weeks were heavier than others), but the volume was always there.
The citations didn’t grow linearly with the post count. There was an inflection around April 13, roughly six weeks after launch, where the daily citation count jumped from 77 to 214 overnight. Nothing specific triggered it. We’d just reached a point where there was enough content surface area that Bing started treating us as a default source for Australian business AI queries.
That inflection happens faster if your content is specific and consistent. It probably doesn’t happen at all if your output is infrequent or generic.
What you can take from this
If you want AI systems to cite your site, here’s what’s actually working for us.
- Pick a topic cluster where you can own the comparison. Not “AI tools” broadly, something specific. “AI tools for Australian tradies.” “HR software for hospitality businesses.” Something you can publish 10–20 posts on without running dry.
- Write the comparison posts. Name the tools. Give verdicts. Include Australian context where relevant.
- Write the company profiles. A dedicated page for each major tool in your cluster. Structured clearly: what it is, what it costs, who it’s for, how it compares.
- Answer the cost questions. “How much does X cost in Australia” is a query type AI pulls from constantly. If you don’t have that page, someone else’s answer gets cited instead of yours.
- Do this consistently for six to eight weeks.
The inflection we hit in April, citations jumping from 77 to 214 overnight, happened without us doing anything special that day. There was just enough on the site by then.
Related reading
- 16,000 Citations and Counting: How a 10-Week-Old Site Became Bing Copilot’s Go-To Source : the data behind this post
- What Australian Small Businesses Are Asking Bing AI Right Now : the actual queries driving citations
frankarr 