Methodology v0.3
How the Opportunity Report works.
The report turns the listing counts shown on your own website into estimates of enquiry workload, unanswered calls and appraisal opportunity, using a fixed, disclosed set of assumptions. It is not a benchmark, not an audit, not a measurement of any person, and it involves no test calls or fake enquiries.
What we read, and what we never do
If you give us your agency website, we fetch at most four pages of it: the address you typed, its contact page (only when the homepage leaves gaps), and the for-sale and for-rent index pages it links to. From those we suggest the agency name, state, contact paths, advertised hours, the listing counts your own site displays, and the website platform, which often indicates a CRM. We also make one lookup of your domain’s public mail (MX) records to suggest whether email runs on Google Workspace or Microsoft 365. Every suggestion is shown to you with its source and date, and you can correct or clear any of it before the report exists. That is the entire footprint.
We do not read realestate.com.au, Domain or any other portal for this report. We do not call your office, email your team, submit enquiry forms, or measure your response times covertly. Listing counts are reported exactly as your site displays them; we never infer rent-roll size from them.
The figures and their assumptions
The three headline figures use the same model as the response-gap calculator, computed from your active rental and sales listing counts:
- Rental workload: 12 enquiries per active rental listing per week, seven minutes per response, plus 10% for CRM updates.
- Buyer calls: four calls per active sales listing per week, with 20% going unanswered.
- Appraisal opportunity: three seller or appraisal opportunities per ten active sales listings per week, with 20% modelled as occurring when the team may be unavailable, totalled monthly. The results screen also expresses this as estimated potential commission, at an assumed average of $25,000 per appraisal; the private report presents it as a count of conversations worth checking. Both are modelled estimates — never measured missed calls, lost commission or revenue.
- Team capacity equivalents: working days per month = weekly hours × 52 ÷ 12 ÷ 7.6; working days per year = weekly hours × 52 ÷ 7.6; share of week = weekly hours ÷ 38 × 100. The 7.6-hour day is an illustrative full-time comparison and the Fair Work 38-hour week is a transparent baseline only — neither is a savings claim.
v0.3 also reads, where available: published opening hours, Google rating and review count for your office and nearby agencies (Google Places data fields), and up to eight public pages per website — yours and up to three nearby agencies’ — checked for click-to-call links, appraisal paths, booking paths, after-hours statements and seller-qualification questions. A page we could not read is reported as “Could not verify”, never as an absence. Buyer-to-seller industry context cites realestate.com.au seller-leads research and is never multiplied against modelled volume — only against counts you provide.
These are fixed assumptions applied identically to every agency: the same inputs always produce the same outputs, and nothing about your CRM, your systems or anything else we observe changes them. They are a starting point to test against your own records, which is exactly what the report’s scenario module does: it computes only from numbers you supply, shows its formula, and never substitutes an industry average for a missing value.
The nearby-agencies section
When enabled, the report lists the most-reviewed real estate agencies within roughly seven kilometres of your office, sourced from the Google Places API under licence and ranked by Google review count, with the date checked shown. Review count is the only measure we hold licensed data for, so it is the only measure named: the section makes no market-share, listing-volume or lost-listing claims, and none of it affects your figures. Your own agency is excluded from the list, and the list appears only inside your own assessment and your private report, never on any public page.
Evidence labels
- Observed. Public information from your agency’s own website, its public mail records, or the licensed Places lookup, always shown with its source and the date checked.
- Self-reported. A value you typed or corrected.
- Estimated. Computed from the counts plus the fixed assumptions above, or from numbers you supplied in the scenario module, with the formula shown.
- Measured. Data supplied with your permission from call, inbox or CRM metadata, during a validation review.
Because AgentFront does not yet hold a sufficiently large, consented dataset of Australian agency operations, the report contains no percentiles, no industry averages and no comparisons of operating performance between agencies, and it will stay that way until such a dataset exists and its method can be published.
Limitations
Estimates built from listing counts describe likely demand on your front office, not measured performance: your actual enquiry volumes, answer rates and appraisal conversion depend on your market, your team and your processes. That is why every figure carries its label, why the scenario module exists, and why the report’s strongest recommendation is to validate against 30 days of real call and inbox metadata before making decisions on it. Review counts in the nearby-agencies section measure online review activity, not service quality or market share.
Versions
v0.3 (August 2026): snapshot-based report — evidence-labelled data model, published-hours coverage window, public-page response audits, appraisal conversations replacing commission figures in the private report, capacity equivalents and a rule-based action plan. Amended 14 August 2026: the results screen again presents the appraisal opportunity as estimated potential commission, using the unchanged model and the $25,000 assumption disclosed above; the private report keeps presenting it as a count of conversations. v0.2 (August 2026): estimate-led report computed from observed listing counts, with the licensed nearby-agencies section. v0.1 (August 2026): a five-question operating-practice rubric, since replaced. Any change to the assumptions or method gets a new version number here, and every report records the version it was produced with.