How to Nurture a Database of Several Thousand Contacts Without Hiring an Assistant

The national median hold period for Australian resales is 8.8 years, per Cotality's Pain and Gain Report covering the March 2025 quarter, in which 86,000 dwellings were resold nationally. That single number should drive every frequency decision you make about a past-client database. Here is the arithmetic, presented as our reasoning rather than as a sourced statistic. An 8.8-year median hold implies roughly 11% of owners transact in a given year, about 1 in 9. Quarterly contact across an average ownership cycle is roughly 35 touches before the average owner is ready to transact with anyone. That reframes the whole exercise. You are not running a campaign, you are running a schedule that has to survive eight or nine years without the recipient resenting it. Thirty-five messages that each carry something the recipient actually wanted is a relationship. Thirty-five newsletters is a slow unsubscribe. Note also that most of those transactions end well, which affects the tone of anything you send. Cotality found 94.9% of resales in the March 2025 quarter were profit-making, with a median nominal gain of A$305,000. Loss-making resales had a shorter median hold, 7.5 years against 8.9 years for profitable ones, so the owners most likely to need you soonest are also the ones most likely to be in a difficult position. That is a reason for a phone call, not a template.

Segmentation only helps if the segments differ in what they receive, not just in how they are labelled. A list split into eleven tags that all get the same monthly newsletter is one list with extra admin attached. Josh Phegan made the point in Elite Agent on 14 February 2017: "Without categorising your buyers, sellers, landlords, past clients and others, you will continually be forced to prospect as you did when you first started your real estate career." In the same piece he wrote that "databases can't be bought; they must be built, and it's up to you to do it." He is a coach with no product to sell in this space, which makes the observation cleaner than most. The split that works is by the value and recency of the relationship. Four tiers is usually enough, and only the top tier gets human time. | Tier | Who is in it | Contact pattern | Who approves | |---|---|---|---| | 1 | Past clients who would refer you | Personal message on real events, plus a call | Human writes or heavily edits | | 2 | Past clients who have gone quiet | Event-triggered message, low frequency | Human approves the draft | | 3 | Enquired, did not proceed | Segment broadcast, clearly a broadcast | Human approves the batch | | 4 | Form and event sign-ups, no relationship | Low-frequency broadcast only | Approve the campaign, not each message | Assign every contact to exactly one tier and make the tier a field the automation reads. If the tier lives in someone's head, the automation cannot use it. If a contact sits in three tiers at once, the frequency rules conflict and somebody receives three emails in a week. Review the tiers quarterly, because a system that never demotes anyone eventually treats the whole database as tier one again.

The trigger for contact should be an event in the data, not a recurring reminder for a human. Calendar reminders decay. They fire when you are busy, they get dismissed, and dismissing one leaves no record that the relationship went untouched. Event triggers do not decay. A settlement anniversary, a change of address, a price change on a property someone was watching, a form submission, an event registration, a tag added by a colleague. Each is a fact in a database with a timestamp, and each can start a workflow without anyone remembering anything. The other half is a query that finds neglect directly. In Rex, `contact.last_contacted_date` is one of 437 searchable Contact fields, and it combines with the nested property tree, so one call returns every contact who owns a property in a chosen area and has not heard from you since a chosen date. Most decent CRMs expose an equivalent. If yours does not, that is a serious argument against it, because without that field the neglected part of your database is invisible to software. Rex's Zapier app publishes 20 triggers, including Contact Updated, Tag Added to Contact, Listing Sold or Leased and Open Home Attendee Added, alongside 11 write actions such as Create Reminder, Create Note, Add Tag, Send Email and Send SMS. That is enough for a stay-in-touch system with no custom code. Check the trigger list before you design, not after, because the gaps decide the shape.

The best Australian benchmark available comes from ActivePipe's Email Marketing Benchmark Report, an anonymised twelve-month analysis of its own platform reported in Elite Agent on 14 February 2022: a 27% open rate, a 5.27% click-through rate, and an average of 2,823 email contacts per user per month. The same report put average leads at 34 per agent per year and average commission per active user at A$446,556 annually, with 68% of emails containing property listings. It is vendor data about the vendor's own platform, which is a real limitation, but it is Australian and it is specific, which beats a generic global industry average. One caveat has to travel with every open rate published after 2021. Apple Mail Privacy Protection pre-loads images on the recipient's behalf, and Apple Mail accounts for roughly 46% of email clients. Every post-2021 open rate is inflated by that, including the 27% above. Click-to-open is the more reliable measure, and reply rate is better still. Ash Farrugia, CEO and founder of ActivePipe, framed the underlying behaviour in that Elite Agent piece: "The report clearly shows how most buyers or sellers sign up to an agents database when they are active in the market, but once they have successfully bought or sold their property, their need for information changes." That is the whole nurture problem in one sentence. People join your list at their moment of highest interest and you keep sending them the content that suited that moment for the next eight years.

The largest measured performance gap in Australian agency work is not talent, it is whether anyone followed up. OpenAgent analysed 2,049 Australian agents and found appraisal-to-listing conversion of 63.7% among the top 10% of agents against 10% for average agents, with lead connection rates of 75.4% against 42.7% for the bottom 10%. That needs an honest caveat. The analysis measures how agents handled portal-referral leads, not their whole appraisal book, so the 10% average is almost certainly well below a typical agent's overall conversion rate. The striking finding is the spread, not the absolute numbers. Top performers contacted 100% of their leads. The bottom decile contacted 80% and converted none. Nothing in that spread requires a better closer. It requires the second, third and fourth contact to actually happen, which is exactly the work a system does better than a person holding it in their head. Realtair, an Australian sales platform, claims that "up to 1 in 10 Market Updates sent convert to an appraisal", where a Market Update is a one-to-one property snapshot built from comparable data. That is a vendor claim with no published methodology or sample and the words "up to" doing real work, so it is a hypothesis to test on your own list rather than a number to build a business case on. Measure your own conversion for a quarter and you will have something better than any figure in this section.

A merge field inserts a name. Personalisation means the message would not make sense sent to anyone else. Recipients tell the difference immediately, which is why databases go cold despite high send volume. The test is simple. Read the message with the merge fields filled in and ask whether a different contact could receive the identical message with only the name swapped. If the answer is yes, it is a broadcast in a costume. That is fine for something genuinely broadcast, such as a quarterly market update, provided it is honest about being one. It is not a substitute for relationship contact. What makes a message specific is a fact the recipient knows you would only have if you were paying attention: the street they bought on, the year they settled, the school zone they asked about, the thing they said at the appraisal. Those facts exist in the CRM only if somebody put them there, which makes note-taking discipline the real input to nurture quality. No amount of automation compensates for an empty notes field. The workable model for a solo operator is a small number of genuinely specific messages each week, drafted from CRM facts and approved by a human, plus a low-frequency broadcast that is openly a broadcast. Trying to make several thousand messages feel personal produces several thousand messages that feel automated.

A review queue works when a human approves outbound in batches rather than one message at a time. The system proposes an action, pushes it to a queue and waits for a decision before anything leaves the building. Design for batch review from the first version. A workflow that drafts 60 messages overnight needs a screen with grouping and bulk approve or reject, not 60 individual notifications. One sitting of twenty minutes clears the batch. Sixty interruptions spread across a day is worse than doing nothing, because it trains you to approve without reading. Show the reasoning and the source data next to the draft. If a message references a settlement date or a comparable sale, the reviewer needs to see the record it came from. A queue that shows only prose turns the reviewer into a rubber stamp, and a rubber stamp is how confidently wrong messages get sent at volume. Confidence-based routing, where high-confidence drafts skip the queue, only works if the score is calibrated so that a 0.9 is right about nine times in ten. If you cannot measure that against real outcomes, do not use the score as a gate. And never use post-action review for outbound email or SMS, because neither can be un-sent. Log every decision: who approved what, when, and what the message said. The full case is in human approval before an automated send.

Measure reply rate and conversations started, because open rate has been unreliable since mail clients began pre-fetching images and it never measured anything you actually wanted. A reply is a human choosing to spend time on you. | Measure | What it tells you | Why it beats the obvious alternative | |---|---|---| | Replies per 100 messages sent | Whether the message was worth answering | Open rate is inflated by image pre-fetching in roughly 46% of clients | | Conversations that reached a call or meeting | Whether replies turn into pipeline | Click rate counts curiosity, not intent | | Unsubscribes per 100 messages sent | Whether frequency has outrun relevance | Complaint volume arrives far too late to correct | | Listings won divided by appraisals conducted | The end of the chain, for a property list | Revenue lags by months and hides the cause | Three numbers are enough for a solo operator: replies, conversations that reached a call or meeting, and unsubscribes. The third is the guardrail. A rising unsubscribe rate next to a flat reply rate is the clearest possible signal that you are sending more and saying less. The consumer evidence says poor contact costs real business, though the sample is small. ActivePipe's Trust Report, published 23 July 2021 and based on only 124 Australian property owners, buyers and sellers, found 61% of past clients would prefer to use the same agent again, 27.3% switched agents because of poor contact, and 63.3% wanted immediate notification of new listings. Treat 124 respondents as directional, not decisive. Track by segment rather than in aggregate. A 4% reply rate across the whole database tells you nothing. A 12% reply rate from past clients and 0.5% from cold form fills tells you exactly where the next hour should go. More on choosing the right measure in monitor the outcome, not the workflow.

The Spam Act 2003 (Cth) applies to every commercial electronic message, email or SMS, and the exposure scales with the size of your list. Three obligations apply: express or inferred consent before sending, clear identification of the sender with accurate contact details, and a functional unsubscribe honoured within five business days. ACMA enforces this against organisations with legal departments. The Commonwealth Bank paid $7,500,000 in October 2024 over more than 170 million emails sent without a functional unsubscribe and 34.8 million sent without consent. Tabcorp's TAB paid $4,003,270 in June 2025. Pizza Hut Australia paid $2,502,500 in May 2024 over more than 10 million messages. In July 2024 ACMA noted businesses had paid more than $15 million in spam and telemarketing penalties over the preceding 18 months. Two points bite hardest on a large database. Inferred consent is not permanent and requires a provable, ongoing relationship with marketing directly related to it, so a contact who enquired six years ago and never transacted is a weak basis for a send today. And unsubscribe handling has to work across every system that sends, not only the newsletter tool. A person who unsubscribes from the newsletter and then receives an automated SMS from the CRM has been failed by an integration gap, and that is still your breach. Kylie Davis, author of The Trust Report and founder of the Proptech Association Australia, framed the standard above the legal floor in Elite Agent on 23 July 2021: "The permission promise is where agents live up to the expectations of the recipient by providing information that delivers ongoing value." Consent is what the law requires. Ongoing value is what stops people withdrawing it.

From 1 July 2026, obligations under the Anti-Money Laundering and Counter-Terrorism Financing Act 2006 apply to real estate professionals, and the OAIC states that all reporting entities must comply with the Privacy Act when handling personal information for or in connection with those obligations. That expressly includes small businesses with annual turnover under $3 million. Anyone running a large contact database should treat this as the most consequential change of the year. The small business exemption has effectively gone for real estate agencies, turnover no longer shelters anyone, and it commenced four weeks ago, so most operators have changed nothing yet. One question is genuinely open, and a lawyer should settle it rather than a blog. Whether the obligation reaches a marketing database, or only the records held for anti-money-laundering purposes, is not settled on the public guidance available. Very few businesses can run two privacy regimes side by side inside one CRM, so the practical planning assumption is broader coverage even while the strict legal scope stays unresolved. Two related changes are already live. The statutory tort of serious invasions of privacy has been in effect since 10 June 2025, creating a private right of action that does not wait for a regulator. And since January 2026 the OAIC has been running its first privacy compliance sweep, targeting the privacy policies of businesses that collect personal information in person, with real estate agents named as an example. One architectural rule follows for anyone using licensed property data alongside a database. Every state property sales licence prohibits direct marketing use. South Australia's wording is the bluntest, that the data must not be used for compiling contact lists whether personalised or not, and Queensland prohibits mail merge functionality outright. Licensed data may populate the content of a message to someone who already consented. It may never build, enrich or select the recipient list.

How often should I contact past clients? Anchor frequency to how long people actually hold property. Cotality's Pain and Gain Report puts the national median hold period at 8.8 years, so quarterly contact across an average ownership cycle is roughly 35 touches before the average owner transacts. No vendor or coach publishes a tested optimal frequency, so treat quarterly as a defensible starting point and adjust against your own unsubscribe rate. What share of my database will transact this year? A median hold period of 8.8 years implies roughly 11% of owners transact in a given year, about 1 in 9. That is our arithmetic from Cotality's figure rather than a surveyed statistic. It is roughly double the 5% figure that circulates in proptech marketing, which we could not trace to any primary source and would not rely on. Do I need consent to email people already in my CRM? Yes. The Spam Act 2003 (Cth) requires express or inferred consent for every commercial electronic message, and being in your CRM is not consent by itself. Inferred consent requires a provable, ongoing relationship with marketing directly related to it, so an enquiry from six years ago that never became a transaction is a thin basis for a send today. Is open rate really useless? It is unreliable enough that you should not manage against it. Apple Mail Privacy Protection pre-loads images and Apple Mail is roughly 46% of email clients, which inflates every open rate published since 2021, including the 27% Australian real estate figure from ActivePipe. Click-to-open and reply rate both measure a decision a human actually made. Can AI draft the messages for me? It can draft them. It should not send them. Drafting from CRM facts and routing the result to a batch approval queue is the pattern that works, because an email cannot be un-sent and a wrong fact in a message to a past client costs more than the time the automation saved. Keep a log of who approved each batch and when. Does the Privacy Act apply to my business if turnover is under $3 million? If you are a real estate professional, as of 1 July 2026 it applies at least in part. AML/CTF Act obligations now cover the sector, and the OAIC states that all reporting entities must comply with the Privacy Act when handling personal information for or in connection with those obligations, expressly including small businesses under $3 million turnover. Whether that reaches a marketing database is an open question worth putting to a lawyer.