How an Australian energy efficiency consultancy cut 800+ admin hours a year.

The Client

The shape of the business.

An Australian residential energy efficiency consultancy operating across NatHERS, BASIX, and Section J. Roughly 1,000 jobs per year. A small in-house team — director, two assessors — supported by a five-person offshore contractor pool handling distribution and drafting tasks. Profitable. Cashed-up. And drowning in admin nobody could solve by hiring another person.

The director had tried the obvious moves. Hired more admin. Bought more software. Wrote more checklists. Each one moved the problem; none of them solved it. The repetitive work was structural — embedded in the way assessments, quotes, plans and reports flowed between people. You can't checklist your way out of a workflow problem.

The diagnosis.

Two hours of discovery surfaced five repeating fires:

• R-value typos. A wrong value on a summary voids a certification. One typo, one day chasing it.
• Quote-acceptance lag. Client accepts. Three days pass before the brief reaches the assessor. The SLA's already burning before the work starts.
• Plan-intake gaps. Plans arrive missing information the assessment needs. The gap surfaces a month in, when the assessor opens the file.
• Adobe friction. Every report ends with someone stamping, copying, formatting, exporting.
• Tacit knowledge silos. The senior assessor knows everything. The offshore contractors know what they were told last week. The same question gets answered seven times.

These weren't one assessor's problems. They were the structural pattern that any consultancy in the 500–2,000 jobs/year band ran into. Solve them once, the playbook applies to the next firm.

What we built.

The MVP shipped in two weeks. It was deliberately the smallest thing that could prove value — a wiki-bot, loaded with the firm's R-value tables, solar absorptance charts, NCC clauses, BASIX certificate requirements, and product specs. Queryable from the team's existing chat tool. No new login, no new software, no training.

Demo moment that closed the engagement: a live query during the discovery call. "What's the solar absorptance for Monument?" Bot returned 0.73, Dark D classification in under five seconds. The senior assessor — fifteen years in the trade — said "we have that information." The director said "this is gold."

From there, Phase 1 rolled out over the following weeks:

• Quote-acceptance auto-forward. Client accepts → assessor receives brief, booking, folder and checklist within seconds. The three-day lag stopped existing.
• Booking confirmation automation. Replaces a manual email the admin lead used to type three times a week.
• Offshore distribution. Job folders sync to the contractor team with persistent access. No more 3-day-expiry WeTransfer re-uploads.
• Online client checklist. One form, conditional logic, auto-creates the SharePoint folder structure. Replaces the JotForm the team had outgrown.
• Wiki capture on the fly. A /save command turns chat answers into permanent wiki pages. Every gap surfaced compounds the bot's knowledge.

The numbers.

Modelled at Year-2 saturation, based on the firm's actual job volume and loaded labour cost of $120/hr:

Total: ~800–950 hours/year saved. At the firm's loaded cost, that's roughly $96K–$114K/year in operational capacity unlocked. The retainer cost is $21K/year. Year 2 returns roughly 3x — and that's before counting the cost of the typo errors and SLA breaches we prevented, which are real money the firm wasn't accounting for at all.

What it cost.

Founding-client rate. $3,000 setup, $1,750/month retainer covering up to 10 hours of work — weekly on-site session, one new knowledge source ingest per week, bot maintenance. Additional automations (quote forward, plan gap finder, energy summary) priced individually as separate engagements.

Year 1 floor: $24,000. The firm's first-year operational savings on the MVP alone covered it twice over.

Why it worked.

Three things this firm had that most engagements don't:

• A director who understood ROI math. He could see 5 hours saved per week × billable rate and do the multiplication. We didn't have to sell automation as a concept.
• A team that already used chat. The bot lived where the team already lived. No new login. No training. The senior assessor was running queries within an hour of go-live.
• Honest scope ceilings. Hour caps in the retainer. Phase 2 priced separately. Nothing committed beyond what was sold. The director never got a surprise invoice — which is the surprise most consultants think they can get away with.

The playbook is portable.

The wiki-bot pattern is not specific to NatHERS. Any consultancy with tacit knowledge in senior heads and admin friction at every handoff is the same shape underneath. Quantity surveyors. Town planners. Building surveyors. Trade-specific compliance consultancies. The same MVP. The same compounding wiki. The same numbers.