AI implementation for ecommerce and trades · Brisbane

Know if AI will pay off before you build it.

I'm Andy Fox-Hulme. I help ecommerce brands and trades businesses work out which jobs AI can take off their team, then build the ones that pay for themselves. Everything customer or money facing waits for a person to say yes.

The problem

Most AI projects die before they earn a cent

RAND's research puts the failure rate above 80%, about twice that of ordinary IT projects. The technology is rarely the culprit.

In small businesses it looks the same every time: nobody owns the job, the process is a mess, the data is scattered across inboxes and spreadsheets, and nobody measured how long it took before. So the build gets paid for, used for a fortnight, then quietly ignored.

I score the work first. It's like a sound check before the gig: five minutes that saves the whole night.

The method

Six questions, one verdict

The AI Viability Score rates any job out of 100 on the six things that decide whether automating it will pay.

  1. How big is the job? Hours a year, at what it costs you now.
  2. How rule-based is it? Could you write the steps on one page?
  3. Where do the inputs live? One clean system, or people's heads.
  4. What happens when it's wrong? Trivial, or legal and money damage.
  5. Who owns it? A keen owner and a team that wants it gone.
  6. Can the systems connect? Open API, Zapier, exports, or locked.
FIX · under 45Sort the process before you build. Building now means paying to automate a mess.
TEST · 45 to 69Run a 4-week supervised pilot. AI drafts, a person approves.
SCALE · 70+Build it properly, with monitoring and an owner.

Proof

Built inside a working ecommerce business

A sports apparel brand selling across Australia and New Zealand. Names changed, numbers real.

Every change to the ad account waits for a person to type YES. AI does the legwork. People keep the keys.

  • Weekly ads monitorWrites and emails the management report for two countries every Monday, ranks each product line on ROAS, MER and CPA, and ends with the one decision that needs attention. Scored SCALE.
  • Ad test builderTurns a creative brief into tracked Meta ad sets, created paused. Scored TEST, so a person approves every launch.
  • Test-to-scale pipelinePromotes winning tests into scaling campaigns, again only after a human sign-off.

Work with me

Score it, build what earns its place, keep it running

For ecommerce brands and trades or service businesses, roughly 10 to 100 staff, where at least one team is drowning in repeat work: reporting, inbox triage, quoting, order admin or ad testing.

Step 1 · Diagnose

AI Viability Audit

$2,500 fixed, ex GST
  • Half-day workshop with your department heads
  • Up to 10 workflows scored and ranked
  • Top 3 builds scoped with budget and payback
  • Fix list for everything marked FIX
  • Fee credited against your first build within 60 days
Step 2 · Build

Fixed-scope build

$8k to $25k per workflow
  • Priced from the score, never above the payback budget
  • TEST workflows start as a 4-week supervised pilot
  • Human approval on anything customer or money facing
  • Staff training and a written runbook
  • Before and after measures agreed up front
Step 3 · Run

Run and improve

$750 to $2,500 a month
  • Monitoring for errors and drift
  • Fixes when your tools or suppliers change
  • Monthly hours-saved and accuracy report
  • Next workflow scored and queued
  • Month to month after the first 3 months

Straight up: the score is a readiness check, not a promise of savings. Hours saved only become cash when the time is redeployed or a hire is avoided, and the pilot replaces my estimate with your real number.

The Case Files

Latest write-ups

Real builds told as case files: the victim (usually time or money), the suspects, the evidence and the verdict. Spoiler: AI is the prime suspect. It's rarely guilty.

About

15 years of paid media, now pointed at AI

I'm a Brisbane-based AI implementation consultant and paid media strategist. I have run Brisbane Web Marketing since 2011 and spent 15 years running Meta, Google and LinkedIn ads for trades businesses and SMBs, including senior roles at LION Digital and Fractal.

Along the way I built the AdSpend Viability Score™, which tells owners whether their ads will work before they spend a dollar. The AI Viability Score applies the same FIX, TEST, SCALE logic to automation.

Before marketing I ran a recording studio. Same lesson both times: get the levels right before you hit record.

Full CV · LinkedIn

Start with one job

Score one workflow free. If it comes back TEST or SCALE, book a 20-minute call and I'll tell you honestly whether it's worth paying anyone to build.