Six commerce properties and a placement engine that improves itself
Aus.Tools, Aus.Deals and four state properties aggregate live pricing across Australian suppliers and retailers, with an AI negotiation advisor on top and a self-learning placement engine underneath. Every one of them already carries “Finance by Local Knowledge Finance” — the commercialisation loop closes back to the practice.
Challenge
Most business websites are brochures with a contact form. They cost money, produce nothing measurable, and their owners have no way to tell whether any part of them is working.
We wanted to build the opposite: properties that generate revenue as a designed function rather than as a hopeful side effect, and that improve on measured performance rather than on whoever argues loudest in the meeting.
The test case was industrial tools — a market with fragmented pricing across hundreds of suppliers, genuine buyer confusion, and a natural adjacency to equipment finance, which is a service the practice already provides.
Approach
The foundation is live aggregation. Aus.Tools publishes pricing and stock from more than 100 Australian tool suppliers; Aus.Deals publishes live pricing from more than 80 retailers. That data has to arrive continuously and stay trustworthy, which is an operating problem rather than a build problem — and operating it is the part that taught us the most.
Aus.Deals adds an AI deal advisor: negotiation scripts, BATNA analysis and pricing context, given away free. The user pays the store directly. The advice is genuinely useful and genuinely unpaid, which is what makes the rest of the model defensible.
Monetisation runs through a self-learning placement engine. Offers are scored using a Wilson lower-bound estimate, which is deliberately conservative about small samples, combined with decaying epsilon-greedy exploration so new offers still get tested without a promising early run being over-rewarded. Every placement carries its own statistics. An offer holds its slot because it performed, and loses it when it stops.
Chat intent detection surfaces relevant offers inside conversation rather than as banner clutter, and offer rails are shared network-wide so all six properties draw on the same commercial layer.
The four state properties — New South Wales, Victoria, Queensland and Western Australia — reuse the same architecture with programmatic suburb and category pages generated incrementally. Expanding into a new market is configuration, not a rebuild.
And the loop closes: qualified equipment finance demand routes back to the practice, which is why each site carries “Finance by Local Knowledge Finance”.
Outcome
All six properties are live. Aus.Tools publishes 100+ supplier sources; Aus.Deals publishes 80+ retailer sources. The four state properties operate on the shared platform, and the finance attribution is visible on each of them — a public, checkable link between the network and the practice that predates this write-up.
What this cluster proves is the verb clients find hardest to buy: commercialisation. Anyone can build a site. Designing the revenue model in at the architecture stage, instrumenting it, and letting measured performance decide what gets shown is a different discipline, and it is the one that determines whether a digital asset earns or merely exists.
When we tell a trades business, a retailer or an equipment-heavy operator that their website should be generating revenue, we are not repeating a principle. We are describing something we run.
Public references
Every substantive claim on this page links back to a publicly accessible record. Follow the links to verify the facts yourself.
Facing something similar?
If you recognise your own situation in this story, a 15‑minute Fit Call is the quickest way to find out whether we’re the right desk for it.
Book a Fit Call