The traffic fell 28% and that was the plan

Most SEO case studies open with a number going up. This one opens with organic sessions down 28% year on year, because that is what the client was paying for.

AOFS is a Melbourne commercial fitout company. I ran their SEO at Digital Next Australia between 2020 and 2022, in the same portfolio as the two Thermofilm brands and a marine dealership that had not launched yet.

The site they had, and the business they wanted

When I picked up the account the website was mostly a shop. It ran WooCommerce, it had a Buy Online section and a cart, and the page title described the company as furniture specialists. The site sold desks, screens and cable trays, one item at a time, to whoever turned up.

That was not the business the owners wanted to be in. They wanted fitouts: office, medical and educational. A fitout is a different buyer making a different search for a job worth orders of magnitude more than a desk. Nobody searching “office fitout melbourne” wants to add a monitor arm to a basket.

So the brief was not “get us more traffic”. It was closer to the opposite: stop being findable for the thing we are getting out of, and start being findable for the thing we are moving into.

What I did

I culled the ecommerce and built the business they wanted in its place.

Ten fitout service pages went up across the following year, one per distinct high-intent search rather than one per product category: office fitouts, medical fitouts, educational fitouts, office space planning, custom furniture, installation, logistics, relocation and removal, furniture hire, and maintenance and repairs. Each was written to answer a buyer’s question rather than to list a catalogue.

The build was staged rather than launched in one go. The Internet Archive first captures the office fitouts page in November 2020 and the educational fitouts page in June 2021, which are the earliest dates I can prove rather than the dates each page went live.

Then I relinked the site around them. The homepage stopped pushing people at products and started pointing at services, using the words people actually type as the anchor text, “office fit outs in Melbourne” and “medical fit outs” rather than “learn more”.

The store came off through 2021. By December the cart, the Buy Online section and the add-to-cart links were gone, and the page title had been reworded from furniture specialists to fit out specialists. One word, carrying an entire repositioning.

Why the traffic had to fall

A product catalogue generates a lot of traffic and most of it is worth very little: a URL per product, each ranking for some narrow product phrase, each pulling in a visitor who wants a price on one item. Delete the catalogue and that traffic leaves with it. It has not declined. It has been removed.

That is the shape of the April 2022 report, and it is why reading the headline metric on its own gives you exactly the wrong answer.

Net movement across the whole tracked keyword set was down 230 positions in that same month, for the same reason: the product terms left with the products. Keywords ranking finished at 37 and tracked volume at 7,020 searches a month.

The falls are the catalogue leaving. The rises are the new business arriving. Read cold, the first three look like an account in trouble. Change what a site is for and you have to change what you measure it by at the same time, and say so before the numbers land rather than after.

Where it got to

Ten fitout terms held page one in Melbourne by April 2022, across 3,310 searches a month.

Every one of those landed on a page built for it. The site also held position one, organic and local, for “australian office furniture”, which is the legacy name and the one furniture term worth keeping.

The technical side came along with it. By April 2022 the site loaded in 2.86 seconds and scored 100 on Lighthouse best practices and 88 on SEO, against a performance score of 74. I would not claim that as a separate project. A site with the shop deleted simply has a great deal less to load than the one that had it.

What I cannot show you

Money. The reports record lead actions, not revenue, and there were two of them in the surviving month. Fitouts are high value and low volume, so two enquiries is not the indictment it looks like on a services site, but I have no closed work to point at and I am not going to imply otherwise.

I also cannot show you a climb. One month of reporting survives, April 2022, recovered from a deleted folder on a backup drive along with thirty-two other client reports. The rankings above are a snapshot of where the site sat, not a graph of it getting there.

And there were Google Ads campaigns running to those same fitout pages by March 2022. They were not mine and I am not claiming them.

The part that is checkable

What makes this study stand up despite one surviving report is that the sequence is not held in my reporting at all. It is in the Internet Archive, where anyone can read it:

• August 2020, my first month on the account: WooCommerce, Buy Online, a cart, and a title calling them furniture specialists.
• November 2020 onward: the service pages start appearing.
• July 2021: both models running at once, the store still live, the ten service pages now in the navigation, and the homepage linking them with high-intent anchor text.
• December 2021: the store is gone and the title reads fit out specialists.

My favourite detail is a flaw. In December 2021, with the shop deleted and the title rewritten, the meta description still described the company as a specialist supplier of quality office furniture. The repositioning had reached the title tag and not yet the description. I would rather show you that than a tidier story, because a case study with nothing out of place in it is usually a case study that has been tidied.

One honest limit on the archive: it shows what changed and when, not who did it. The sequence matches my account of the work, and Digital Next held the SEO account and produced the reports, but the archive is corroboration rather than attribution.

What I would take from it

Traffic is a proxy. It is a good proxy most of the time, which is why it is so easy to keep reporting it after it has stopped measuring anything useful. The moment a business changes what it sells, the proxy breaks, and the job becomes explaining why the number that used to mean success now means the plan is working.