How big should a real estate farm area be

How many homes should a farm area hold?

Between 1,000 and 2,000, and the reason is arithmetic rather than taste. Here is the working.

Australia turns over about 564,483 homes a year across a housing stock of 11.4 million dwellings, on Cotality's March 2026 chart pack. Divide one by the other and roughly one dwelling in twenty changes hands in a year. That single rate sets the size of a farm area before any question of skill or script comes into it. A patch of 1,000 homes has about 50 sales a year in it. A patch of 1,500 has about 74. Those sales exist whether you call the street or not, and they are the only listings your patch can ever produce, so the first job is choosing a patch big enough to be worth a year of your time and small enough that you can get to all of it.

The Australian Bureau of Statistics counts the housing stock slightly differently and puts it at 11,495,200 dwellings in the March quarter of 2026, which moves the rate to 4.91 per cent. The two agree closely enough that nothing below turns on which one you use.

The floor and the ceiling on the size come from different places. A patch under 1,000 homes holds fewer than 50 sales a year, which is thin ground for a year of work.

The ceiling is your week. Contacting 1,500 homes four times a year is about 6,000 calls, or 125 a week, which is 25 a working day. We put most agents at 10 to 20 attempted conversations a day, a view from the desk, with no sample published, so a patch past 2,000 homes is one somebody else is calling for you.

What caps a farm area, the calls or the sales?

The sales, once the patch is small. This is the part of farming that published prospecting rates get wrong, because those rates were measured on lists with no boundary around them.

Ray White runs a network phone day twice a year and publishes the counts: 141,131 calls and 12,304 appraisals in February 2026, then 126,773 calls and 10,021 appraisals in June, which is 11.5 to 12.7 dials per appraisal booked on lists the agents already owned. Applying the 6.5 per cent appraisal to listing conversion its managing director reported for 2025 gives 177 to 195 dials per listing won. We work through that chain in how many calls it takes to win one listing.

Now run it inside a boundary. Contact 1,500 homes once a quarter and you place about 6,000 calls in a year. At 190 dials per listing that would imply 31 listings, out of roughly 74 sales in the whole patch. That is a 42 per cent market share, which nobody holds. The rate breaks before the patch does.

So treat the dialling rate as a cost, not as a yield. It tells you what winning a listing consumes in calls. It does not tell you how many listings a fixed number of houses can hand you, and an agent who plans off the first number alone will budget for a year that the suburb cannot supply.

How do you choose which streets?

On two published numbers and one you can count yourself. Turnover decides how many sales are in play, median value decides what each one is worth, and the boards in the street tell you who you are taking share from.

What does a year in the patch actually produce?

It depends entirely on the share you take, so here it is at three illustrative shares rather than at one number we would have to invent. The share column is illustrative: it is chosen to bracket a plausible range, not measured from any published dataset.

Read the third column rather than the fourth. At a tenth of the patch you need about 1,406 dials in the year, which is a quarter of what contacting 1,500 homes four times already costs you. The calling is not the scarce input. The reason to make all 6,000 calls is that you cannot know in advance which 74 houses are the ones that sell.

How often should the patch hear from you?

Four times a year is the cadence we build campaigns around, which is a view from the desk, with no sample published. What is published is the window you are allowed to do it in, and it is narrower than most agencies roster.

Section 8 of the Telecommunications (Telemarketing and Research Calls) Industry Standard 2017 prohibits a telemarketing call on a weekday before 9am or after 8pm, on a Saturday before 9am or after 5pm, and on a Sunday at any hour, measured at the recipient's address rather than yours. The state by state arithmetic is in telemarketing calling hours in Australia.

Inside that window the hour still matters. Baylor University's study of 6,264 real estate cold calls found the 10am to 2pm block took 31 per cent of the dials and produced 53 per cent of the appointments and referrals, while calls after 5pm were 42 per cent of dials for 21 per cent of the outcomes. We take that apart in the best time to cold call homeowners.

The last question is the whole point of farming. You are not trying to book an appraisal from a stranger on a Tuesday. You are trying to know which of the 1,500 are close to moving, because roughly 74 of them will sell in the next twelve months and none of them is wearing a sign.

What breaks a farm area plan?

Almost always the same three things, and none of them is the script.

1. The patch is too big to finish. An agent who picks 4,000 homes contacts the first 900 twice and the rest never, which is not a farm area, it is a list.
2. The cadence dies in a busy month. Four contacts a year survives listing season only when the calls are somebody's actual job rather than the thing done last.
3. The measurement is boards instead of conversations. Share moves slowly, so a plan judged on listings at ninety days gets abandoned two quarters before it works.

The second one is the reason this article exists. Working a 1,500 home patch properly is about 6,000 calls a year, and the person expected to make them is also the only person who can run the appraisal they produce. Every serious answer to that buys the calls from somebody else: a junior, an offshore assistant, a portal lead that is the same call made by a marketing budget. We compare them in door knocking, calling or letterbox drops.

ListingLine is an AI voice agent that calls an Australian agent's farm area in the agent's name, at A$2.00 per connected call. On Baylor's 28 per cent answer rate, 6,000 attempted calls is about 1,680 connected calls, so a patch that size is on the order of A$3,360 of calling in a year. Baylor is a United States study from 2011 and the closest published proxy we have, not a measurement of your streets, so treat that connected call count as modelled, not measured.

What we will not do is turn that into a cost per listing. Doing so would assume an AI books appraisals at the same rate a human does, and the channel comparison linked above labels that row rate assumed equal to a human for the reason that we have not found a published measured figure. Until somebody does, the honest output is a calling cost, not a cost per listing.

The other variable is the answer rate, which moves the connected call count and therefore the bill. We take that apart in what an AI voice agent really costs per call.