How FairCost prices are calculated

Modelled estimate

Modelled estimates use the existing FairCost guide assumptions and published city factors. They may also use clearly labelled suburb, size or materials assumptions. The original source documents for these assumptions have not been independently reconciled in this build. They are not verified rate cards or observations.

Low and high are model bounds. The midpoint estimate is halfway between those bounds, rounded to dollars. It is not a median or percentile of actual payments. A model has no sample size, observation window or last-verified date. Unknown GST treatment is shown as unknown rather than excluded.

Verified consumer submissions

A price is eligible only when its submission is approved by a recorded reviewer, has a private evidence attachment, a successful recorded ABR check and passed duplicate and outlier screening. Only written quotes, invoices and receipts are eligible. Pending, rejected, self-reported, incomplete, ambiguous evidence and future-dated records do not enter the price calculation. Duplicate identifiers are counted once. Existing broad aggregates are not treated as verification evidence.

At least 10 eligible submissions in the same service, city and consistent job/scope, GST, materials and quoted-versus-paid cohort are required to publish a range. This is FairCost’s publishing rule, not a Google indexing rule or a guarantee of representativeness. The largest eligible matched cohort is identified on the card. Different job types are not silently pooled. Eligible records are limited to the preceding two years.

Q1, median and Q3 use linear interpolation between sorted amounts at the 25th, 50th and 75th percentiles, rounded to cents. Each entry has equal weight; the sample size counts submissions, not unique households. We do not infer missing payment dates, GST flags or scope. Quotes are kept apart from paid-price records.

The current submission workflow screens duplicate candidates and extreme values against model bounds for human review. That can introduce selection bias. A rejected or flagged record is not proof that its price is wrong. Operator and customer concentration are not measured by the submission count.

Disclosed operator booking records

MaidForYou is owned by FairCost’s founder. Its booking summaries remain disabled until reconciled against the original export. The verification process requires a unique booking identifier, confirmed completed status, Sydney location, date, bedrooms, clean type, frequency, reconciled base amount and explicit GST treatment. It excludes cancelled, non-Sydney, deep and end-of-lease cleans, invalid amounts and duplicate identifiers.

Published cohorts show their own sample count, frequency, home size and fixed date window. Five-plus bedrooms are grouped only from records explicitly identifying those sizes. We do not construct a pooled range from different bedroom quartiles. Cohorts smaller than 30 bookings are withheld. Repeated visits count as separate bookings; this is not a count of independent customers.

The legacy generator used Final Amount while claiming extras were excluded. Reconciliation must confirm whether the amount is the base service charge after discounts, whether extras are excluded and whether GST is included. Until that is resolved no legacy amount is presented as verified. The matching script documents its percentile convention and records the export hash and check date.

One business’s prices reflect its service area, customer mix and pricing decisions. They are not an independent citywide survey. Annual changes can reflect different customers or incomplete years; no causal explanation is inferred from a price series alone.

How to cite FairCost

Include FairCost, the dataset title, disclosed source, exact cohort, date window, access date and dataset URL. The FairCost Index links to the CSV dataset. Released aggregates are available under CC BY 4.0; an empty dataset indicates verification is pending.

Independence and corrections

We don't sell placements or leads today; if that changes, paid slots will be labelled and will never affect price data or rankings.

Regulatory pointers are not a substitute for a qualified review of the exact work. Unverified thresholds have been removed, and outstanding reviewer assignments are recorded. Report a source or pricing correction.