Australia: the plan, and how we work it

We appear in search results about 6.5 million times every three months, and people click through 9,864 times. That is 0.15%. For every thousand people who see us, one and a half click.

That is not a ranking problem. We are on page one for the terms that matter. They are seeing us and choosing something else.

Think about how you use Google. You scan, you read three or four headlines, and you pick one. You are not evaluating websites, you are picking a headline. That headline is the title tag.

The nine are already written. Your job is to ship them and then find out whether they worked. That second part is the real lesson. Anyone can change a title and say traffic went up. Traffic goes up and down on its own.

1. Record the before state, all nine, first

   Current title, current description, and clicks, impressions, click-through and average position for the last 28 days. Before you touch anything. You cannot recreate it afterwards.

2. Open the page in the HubSpot page editor

   Settings, then the SEO panel. Never the Design Manager.

3. Swap in the specced title and description

   From title-tag-rewrites-phase1.md.

4. Publish, and write down the date

   Per page. Everything we measure later hangs off it.

5. Annotate each one in SEOtesting

   What changed, which page, what date.

6. Re-measure at 14 days and 28 days

   Same metrics, same length of window.

• Open the live page, right click, View Page Source, search for <title>. Your new text should be there. If the old text shows, it did not publish or it is cached.
• Check the length. Roughly 60 characters before Google truncates the title, about 155 for the description.
• A few days later, google the page title in quotes and see whether the new one is showing.

• Change nothing else on those nine pages during the measurement window.
• Compare equal windows. 28 days against 28 days, never against "since we shipped".
• Watch position as well as click-through. If position moved too, the title may not be the cause.
• A title change can lose you clicks. That is normal, and exactly why we measure. If it drops we revert and we have still learned something.

One page makes the point. /seo-services/link-building/ gets about 84,000 impressions and 9 clicks. Google shows that page to a lot of people and essentially none of them want it.

Nothing is broken. It ranks. But when someone sees it, they choose something else. Every one of those 84,000 impressions is a chance we already earned and threw away.

The skill here is writing for the search result rather than for the page. Stop thinking "what is this page about" and start thinking "what did the person type, and what would make them feel this page answers it".

• For each page, look at the top three search terms it actually appears for. Write for those, not for what you assume the page is about.
• Record the before state, same as L1.
• Aim for about 60 characters on the title, 155 on the description.
• Ship in batches of five, so if something goes wrong you know which batch.
• Annotate each batch. Measure at 14 and 28 days.

• Do not keyword-stuff. "SEO Agency Melbourne | SEO Services Melbourne" reads as spam to a person, and people are who click.
• Every title must be different. Two pages with the same title compete with each other.
• A page at position 25 with low click-through is normal. A page at position 6 with low click-through is a real defect. Those are the priority.

Headings are not decoration. They are how a page explains its own structure.

Google does not see your design. It sees an outline: this is the subject (H1), these are its sections (H2s), these are the parts within them (H3s). If that outline is a mess, the page is harder to understand and harder to rank.

Three problems come up constantly, and you will find all three on our site: more than one H1, skipped levels (an H2 followed by an H4), and headings chosen for how big the text looked rather than what the content is.

• List every heading in order with its level, for each priority page.
• Check: exactly one H1, no skipped levels, every heading describes what is beneath it.
• Fix by changing the heading level, not the styling. If something needs to look smaller, that is a CSS job.

If fixing a heading level changes how the page looks, stop and flag it. That means styling is attached to the heading level, which is a template issue and Lawrence's side.

This is the least understood thing in SEO and it has the best return for the effort.

Every link on our site passes value to the page it points at. The way we link between our own pages is us telling Google which of our pages matter. Most websites never decide this on purpose, so all the value pools on the homepage and whatever the menu links to.

Direction matters. Blog posts and case studies are what other websites link to. Service pages make money. So the flow goes from blog posts and case studies towards service and location pages. Not the reverse.

A link inside body text is worth more than a link in the menu. The menu links to everything, so it says nothing about what is important. A link inside a sentence is a choice.

• Work from the source-to-target plan Lawrence gives you.
• Find a place in the body where the link genuinely fits the sentence. If it does not fit naturally, do not force it. Flag it and move on.
• Vary the anchor text. First exact match, then partial, then branded or natural.
• Maximum two links to the same destination from any one page.
• Keep a record: source, target, anchor text, date.

• Do not link from dead pages. A page with no traffic has no value to pass. This is why W1 comes first.
• Do not link two different pages using the same anchor text. They will compete.
• Click every link you added on the live page. Confirm it goes where you meant and does not 404.

Everything else here is about the website. This one is not, and that is why it is worth doing.

Search "seo agency melbourne". Before the blue links there is a map with three businesses. That block is not driven by the website at all. It is driven by the Business Profile, the reviews, and whether the business details match everywhere they appear online.

We have never checked ours. Not once, for Melbourne or Sydney. We do not know whether the information is complete, whether there are photos, or whether there are old duplicate listings nobody is watching.

Website changes take weeks to show. Profile changes can show the same day.

• Inventory first. Search the business name and each office address. Look for duplicates and old listings, not just the ones we know about.
• Record everything: name, address, phone, primary and secondary categories, hours, description, photo count, video count, last post date, review count, review replies.
• Consistency check. Compare the name, address and phone character by character against the website. "St" against "Street" counts as a difference.
• Fill the gaps. Photos matter more than people expect.
• Record who currently holds the three map slots for "seo agency melbourne" and "seo agency sydney".

A page with very little on it is not neutral. It costs you. Google forms a view of a website as a whole, so a site with a lot of thin pages is judged as a thinner site, and that drags on the good pages too.

Here is the comparison. An agency ranking number one for "seo agency" in the UK has zero pages under a thousand words. Not a few. Zero. Their median page is around four thousand. Not because long pages rank better mechanically, but because they decided every page has to earn its place and held to it.

• Before writing anything, ask what the page is missing rather than how to make it longer. Usually one of: does not answer the obvious follow-up, has no examples, has no proof, does not say what happens next.
• This is body copy, so it goes through the content workflow. Lawrence walks you through it on the first one.
• If the honest answer is that nothing is missing, say so and send it back for merge or removal.
• Word count is a symptom, not a goal. A useful 700-word page beats a padded 1,500-word one.
• Never copy content from another page to fill space. That creates two pages competing.

Rankings are not permanent. A page that did well two years ago and has not been touched is usually doing worse now, because competitors updated theirs and it stood still.

We measured in April. /seo-sydney/ went from 235 clicks a month to 143. The homepage from 590 to 531. /seo-services/local-seo/ went from 14 to zero.

Nobody did anything wrong, and that is the point. Doing nothing is a decision, and this is what it costs.

• Get the terms each page currently ranks for from Lawrence. That is what you are protecting.
• Record the before state and the date.
• Goes through the content workflow. This is a copy change, it does not skip the gate.
• Update what is factually out of date, examples that have aged, statistics with old years, and questions the page does not answer.
• Do not change the URL. If it has to change, that is a redirect job and Lawrence's call.
• Measure at 28 days. Refreshes are slower to show than metadata changes, so do not judge at 14.

Structured data is a block of code that describes a page in a way machines read directly. A person reads "StudioHawk, 03 1234 5678, open 9 to 5". The code version says explicitly: this is an Organization, this is its phone number, these are its hours. No interpretation required.

It matters for two reasons. It is how you become eligible for the extra things in search results, like star ratings and FAQ dropdowns. And increasingly it is how AI systems read a page, because they read the structured version rather than guessing from the layout.

Our Australian and UK sites have this. Our US site has none at all, which is one reason it is close to invisible.

This is last on purpose. It will make far more sense after a few weeks of looking at how pages are built.

• Add the code block Lawrence gives you, per page.
• Test the live page in Google's Rich Results Test. Fix errors. Warnings are usually optional fields and fine to leave.
• View source and confirm the block appears once. Duplicates cause conflicts.