Most SEO case studies are published after the win, with the losing half edited out. This one starts in the middle, with a result that is genuinely half good: our AI chatbots page climbed onto page one for “ai chatbots canberra” in a week — and not one person clicked it.
We are writing it up as it happens, on our own site, because it is exactly the work we do for clients and it is easier to show than to describe. Part 1 is the baseline and the changes. Part 2, in three to four weeks, is whether they worked.
The baseline: page one, zero clicks
| Metric | 20–26 Aug 2026 | 27 Aug – 2 Sep 2026 | Change |
|---|---|---|---|
| Impressions | 9 | 57 | +48 |
| Average position | 12 | 6.7 | −5.3 |
| Clicks | 0 | 0 | 0 |
| CTR | 0% | 0% | 0 |
Read that carefully, because the shape of it is the whole diagnosis. Impressions up six-fold and position improving by five places means Google decided the page belonged in front of more people — the ranking half is working. Fifty-seven people saw us in the results and every single one of them chose something else.
That pattern has a specific meaning. When you are invisible, you have a ranking problem. When you are on page one and nobody clicks, you have a snippet problem. They need completely different fixes, and the common mistake is to answer the second with the first — more backlinks, more content, more of everything, when the actual failure is forty characters of title text.
At position 6.7 a decent result is a 3–6% click-through rate. We were getting 0%.
What we found on the page
So we went and looked at our own page the way a searcher would, and it was not hard to see why they scrolled past.
The title was about the wrong thing. “AI Chatbot for Business Websites | Canberra AI Integration” — someone who types “ai chatbots canberra” scans the results for those words in that order. Ours had “Canberra” buried in the second half, attached to a phrase nobody searches for, and the singular “Chatbot” where the query says “chatbots”.
The H1 said “AI Chatbots” and nothing else. No place, no audience, no offer.
The description had no hook. No location, no price, no reason to pick us over the five results above. At position 6 or 7 the snippet is doing all the work, and ours was doing none.
The local intent was thin. The page described a product; it did not read like a page for Canberra businesses specifically. “Canberra” appeared in the plumbing, not in the argument.
Two technical things also sat underneath all of it, both recently corrected: the sitemap was only submitted on 14 September, and duplicate http/www hosts were consolidated on 15 September. Neither explains a 0% CTR — a snippet problem is a snippet problem — but both affect how quickly Google picks up what we do next.
What we changed this week
Specific changes, all shipped together, so part 2 measures the package rather than any one line:
- A new title that mirrors the query. “AI Chatbots for Canberra Businesses — Try Ours Live | Connect Your Way”. The words the searcher typed come first; “Try Ours Live” is the differentiator nobody else in those results can claim.
- A new H1: “AI chatbots for Canberra businesses”.
- A new description with both hooks the old one lacked — local and commercial: what it does, that it is built in-house here, and the price.
- A local section with real local argument — who this is for in Canberra and why in-house and local support matters, instead of “Canberra” as decoration.
- A live-demo CTA above the fold. The chatbot is already on the page; now the top of the page tells you to open it. Our whole differentiator is that you can try the product before you talk to anyone, and it was previously hidden in a floating button.
- Published pricing. Free first month, then a one-off setup and A$100 a month. Pages that publish a price get clicked by people who are ready to buy and skipped by people who are not — both are wins.
- An FAQ answering the six things people actually ask before buying: cost, setup time, what it will not do, where the summaries go, updates, and what happens after the free month.
- Internal links with the anchor “AI chatbots Canberra” from the homepage services card, from a related post, and from a new supporting post — so the phrase we want to rank for is the phrase our own links use.
- A supporting post: what AI chatbots cost in Canberra, and why we build ours in-house, answering the research-stage question the product page is not the right place for.
- A
chat_openevent in GA4, so we can see whether people who land on the page actually open the chat — the on-page half of the funnel we were flying blind on. - A Google Business Profile service entry for AI chatbots, because local packs and organic results feed each other for queries with a city in them.
How we will measure it
Three sources, and we will publish all three in part 2 whether they flatter us or not.
- Search Console, query and page report, same comparison shape as the table above: impressions, average position, clicks and CTR for “ai chatbots canberra” on /products/ai-chatbots.
- GA4 events — landings on the page,
chat_open, andchat_message_sent— to see whether the clicks we win do anything once they arrive. - Enquiries, the only number that pays for anything.
What a good result looks like: at the current position, a 3–6% CTR would be normal — on 57 impressions a week that is two or three clicks, which sounds trivial and is actually the difference between a page that exists and a page that works. If the changes also move us into the top three, the range is more like 20–30%. If CTR is still 0% in four weeks, the title is not the problem and we will say so and look again.
We will re-check in three to four weeks. Search does not respond to a change in a straight line, and anyone who reports week-one numbers as a result is selling noise.
The honest caveat
We guarantee a technical score, never a ranking. Every page we build is guaranteed to hit Lighthouse SEO 100, because that is entirely within our control: it is structure, metadata and semantics, and we either deliver it or we have not finished. Performance, Accessibility and Best Practices are shipped at 90+ with the report in front of you — held to the same discipline, but not guaranteed, because they depend on things like the images you give us. You can read what the score covers in what Lighthouse SEO is and why we guarantee 100.
Where you rank is not within anyone’s control. It depends on Google, on the competition, and on things that change without notice. Any agency promising you position one for a given phrase is either misunderstanding the product or misrepresenting it, and the honest version of this case study might well end with “that did not work, here is what we tried next”.
This is the service
What you have just read is what our SEO work looks like for a client, every month: monitor the queries that matter, find the page that is under-performing, work out whether the problem is ranking or snippet or intent, change something specific, measure it, repeat. Not a report full of charts — one change you can point at, and the number it moved.
It is included in every website build we do, alongside the Lighthouse SEO 100 guarantee. If you would rather have someone doing this on your site than read about us doing it on ours, get in touch.
Part 2 lands in three to four weeks with the numbers.