The year on one page
- Where we are: ~$2.9M revenue run-rate; ~$1.3M/yr contribution after ad cost (before staff costs). 86% of intros come from a paid engine whose weekly GP margin swung +87% → −4% → +66% in June. Organic intros are worth ~2× paid intros and cost nothing, but organic is only 14% of volume and softening. An "intro" = a seller introduced to agents (CRM meeting-agents); revenue is the referral fee expectation per intro (~$732 blended).
- The market won't help: rates at 4.35% with no cuts expected; post-Budget forecasts say −20% housing turnover. Growth must come from market share and yield per seller.
- The plan — six levers: ① ship the property estimate tool and take the valuation search shelf; ② turn the 42k-agent / 1M-sold-row data platform into public agent-evidence and suburb pages; ③ switch on email against the dormant 14k-contact database; ④ re-plumb Google to bid on gross profit with hard budget guardrails; ⑤ ride the one-off pre-July-2027 investor-exit wave (WREA already owns the CGT search terms, with zero competitor presence) and lift yield per intro; ⑥ convert more of the leads we already have — speed-to-lead on the sales floor and a monthly CRO cadence on the site.
- The number: plan $2.1M contribution after ad cost (+60%); base case if SEO and email ramp slowly: ~$1.8M (+38%). Paid grows only ~12% (flat budget, better bidding); nearly all growth is owned channels going from ~550 to ~1,400–1,600 intros/yr. That mix shift is the whole game — and the main risk.
- Two hard dates: estimate tool, agent-evidence pages, and email rescue live by 1 September (spring selling season); investor-exit campaign live January 2027.
- Current Q1 launch sequence (refined 27 July): the next major customer-facing launch is controlled self-serve agent connection. It uses the journey already built and goes to a small production cohort only after three gates pass: the agent shortlist is accepted, source filtering reveals no agent, agency or data unique to REA, and the scraper/data-confidence evidence is accepted. Deeper Property Value Form call-attempt analysis sits later in the Property Value Hub rollout checklist; it is not the current CEO focus.
- What we're not doing: more paid budget, all-17k-suburb page spam, landlord leads, embedded finance, brand advertising, more dashboards.
1. Where the business actually is
From the live 12-week CRM pull (6 Apr – 28 Jun 2026): 3,295 leads → 908 intros, $361k ad spend, $398 blended cost/intro, $732 revenue/intro, $334 GP/intro. Annualised: roughly a $2.9M revenue run-rate with ~$1.3M/yr contribution after ad cost (staff and other opex sit below this line — every "GP" figure in this document means contribution after ad cost only). Intro mix: 86% paid (Google ~70%, Microsoft ~16%), 14% organic.
908
Intros / 12 weeks (~76/wk)
$398
Blended cost per intro
$732
Revenue per intro
28%
Lead → intro conversion (12wk avg)
~$715–895
GP per organic intro (zero ad cost)
0
Nurture emails sent to 14k emailable contacts
Six structural facts the strategy is built on
- The paid engine is fragile, not just expensive. Weekly GP margin swung +87% → −3.8% → +66% across three June weeks. Google weekly cost/intro ranged $410–$692 in June; in the worst week (15–21 June) the Google channel netted $16 GP per intro ($708 revenue − $692 cost). Serving budget exposure is $39.2k/week against a $25k cap (open P0, two weeks running), and 30 unlogged account changes hit in one week. This engine funds the company and it is one bad month from funding nothing.
- Organic intros are worth ~2× paid intros and cost nothing — but organic is small and softening. ~10–13 intros/week at $715–895 GP each; organic lead CVR slid 1.5% → 1.4% and organic form-completion (Clarity SubmitForm) is 2.04% vs 5.07% on paid landing pages. A single WREA page (
/australian-house-prices-graph…) gets ~880 organic users/week and produces zero leads. - Lead→intro conversion moves 26–39% week to week, and nobody is managing it as a number. WREA has 573k call-log rows and 173k call recordings (no transcripts) — the raw material for a sales-conversion system exists and is untouched. A 5-point lift on ~3,300 leads/yr is ~165 intros — as big as the entire email opportunity.
- WREA is sitting on an unmonetised proprietary data asset none of the competitors can cheaply copy: the agent-intelligence platform — ~17.2k suburbs, 42k agents, 599k listings, ~1.05M sold-history rows, per-agent sold counts / median price / days-on-market, with a public-safe rankings API already feeding WordPress/Laravel. RateMyAgent built an 85k-visit organic machine from agent pages with weaker per-visit economics; WREA has the data and ~9k visits.
- The database is dormant. ~30.5k property-owner contacts (14k emailable), zero nurture sends, while OpenAgent runs a 160-thread suburb-personalised email machine against the same sellers. Both build specs to fix this (nurture OS + estimate tool as capture engine) were completed this week — the plans exist; FY27 is execution.
- The team is small and named: Thomas (CEO), David (paid media), Felix (dev), Mark, sales + support. Everything below is sequenced for one dev plus AI-operator leverage (the weekly growth pack, Codex/Claude lanes) — with a pre-declared descope order when Q1 gets tight (§7).
2. FY27 market reality — and the one-off opportunity inside it
| Fact (verified July 2026) | Strategic consequence |
|---|---|
| RBA cash rate back at 4.35% after three 2026 hikes; big-four consensus: no cuts in FY27 (Westpac sees two more hikes). | No rate-cut demand wave is coming to save lead volume. Plan for flat-to-down seller demand. |
| May 2026 Budget: negative-gearing quarantine on new established-property investments + CGT discount replaced by indexation from July 2027 (existing holdings grandfathered). Westpac: investor activity −34%, total turnover −20%. | The addressable pool of transactions shrinks ~20%. Growth must come from share and yield. But grandfathering creates a time-boxed investor sell-decision wave before July 2027 — a genuine one-off demand spike in H2 FY27. |
| WREA already ranks with zero competitor presence on: capital gains tax calculator (9,900/mo), determine capital gains (5,400), ATO clearance certificate (5,400). | WREA is — accidentally — the best-positioned player in the country for investor-exit search demand. Nobody else owns this cluster. This is the single cheapest large campaign available in FY27 (§6, Lever 5a). |
| Divergent capitals: Perth +13%, Brisbane +9%, Adelaide +7% forecast; Sydney −3%, Melbourne −4%; clearance ~50%. | Two seller psychologies to serve: hot-market sellers (WA/QLD/SA — speed and yield) and anxious-market sellers (NSW/VIC — exactly the "which agent can defend my price?" positioning). Geographic budget and content weighting follows this split. |
| AI Overviews: informational queries lose 35–64% of clicks; transactional/local real-estate queries are among the least affected; being cited inside AIOs lifts CTR. | Doubles the case for tools + local/transactional pages over generic articles, and for structured, citable first-party data (fee ranges, agent stats, suburb medians) that AI engines quote — with WREA's name on it. |
| Spring selling season: September–November 2026. | Hard deadline: every FY27 asset that touches seller acquisition must be live by 1 September 2026. Q1 is a build quarter with a date, not a vibe. |
3. Competitor read
| Competitor | What they actually are (verified) | FY27 posture toward them |
|---|---|---|
| OpenAgent | The real full-funnel competitor: 924 suburb pages, OpenEstimates (valuation volume growing 28.5k→34.5k in June alone), paid valuation ads, 160-thread suburb-personalised email machine. A$23.6M raised; ~2.1M visits/yr; no visible 2025–26 product news (possibly consolidating). | Copy the system (local page → valuation capture → nurture → comparison CTA), beat them on data specificity and fee/strategy honesty. Their sitemap added zero new URLs in late June — they are not sprinting; the window to close the suburb/valuation gap is open. |
| LocalAgentFinder | Paid-heavy (Semrush-estimated ~$28k/mo — directional), disciplined conversion funnel, strong property-value page (~89k sampled vol), bought a conveyancer (2024) — monetisation deepening. ~49 staff. | Their valuation tool is the benchmark to beat on credibility (the V3 tool spec already does). Their conveyancing move validates Lever 5's ancillary-revenue path. |
| RateMyAgent (ASX:RMY) | 85k organic visits from 12k agent-directory pages, but a subscale, loss-making review/awards business (FY25 rev A$21.5M, FY26 forecast cut; market cap ~A$20M). Not a direct seller-lead competitor. | Proof-of-demand, not a threat: their traffic shows agent-evidence pages are the biggest organic pool in the category. WREA has better per-agent data economics (each visit can become a $732 intro, not an ad impression). Take their traffic playbook, attach it to a referral P&L. |
| Global pattern | HomeLight → embedded finance (equity unlock via 40+ lenders, then AI closing agent); Clever → white-label agent-matching for mortgage servicers; LAF → conveyancing. The funded winners all deepened revenue per seller rather than buying more top-of-funnel. | FY27 pilots ancillary monetisation (broker/conveyancer referrals) small and late (Q3–Q4), after the core levers are live. Direction, not distraction. |
4. The strategy: six levers
One sentence: stop renting all your demand from a volatile auction — convert WREA's four dormant assets (the valuation moment, the agent-performance data, the CRM database, and the leads you already pay for but under-convert) into owned demand and higher yield, while re-plumbing paid to bid on gross profit instead of leads.
Lever 1Own the "what's my home worth?" moment. Ship the property estimate tool as the acquisition anchor (spec done), build the /property-value hub, and take the 49.5k+/mo valuation cluster where WREA is currently absent. The single biggest demand pool adjacent to the business, both competitors are farming it, and the AIO shift favours tools over articles.
Lever 2Weaponise the agent-evidence data (the moat). Turn the 42k-agent / 1.05M-sold-row intelligence platform into public agent-evidence and suburb-market pages, sales scripts, and estimate-tool proof. This is RateMyAgent's traffic engine attached to WREA's referral economics — and the "confidence-building" thesis from your own staff brief made concrete.
Lever 3Monetise the database. Execute the nurture operating system (spec done): rescue lane for the ~137/month stalling leads (SMS decision for the two-thirds without email), then value-refresh and local-market utility lanes across the 14k emailable contacts. OpenAgent proves the category; WREA's current send volume is zero.
Lever 4Re-plumb paid to bid on gross profit. Value-based bidding with CRM GP uploaded as offline conversion values, automated budget-exposure guardrails, change-log discipline, weekly leak-kills. Not uncontrolled spend — keep the same ~$30k conditional weekly boundary working harder and no longer able to silently lose money.
Lever 5Deepen yield per seller. (a) The investor-exit campaign riding the pre-July-2027 CGT deadline through WREA's unopposed CGT/ATO content; (b) route intros to agents that actually convert (yield routing from own data); (c) pilot broker/conveyancer referral revenue. The proven global pathway, sized for a small team.
Lever 6Convert more of what we already have. Sales-floor speed-to-lead (SLA measurement, attempt cadence, AI mining of the 173k call recordings) plus a monthly CRO experiment cadence on the site (organic form completion 2.04% vs paid 5.07% — the gap is fixable). Lead→intro from 28% toward 33%+. The cheapest intros in the company are the ones already paid for.
5. FY27 targets, decomposition, and the GP bridge
5.1 Headline targets (phased H1 / H2 — most levers are H2-weighted; don't misread October as failure)
| Metric | FY26 exit run-rate | Dec 2026 exit | Jun 2027 exit (FY27 target) | Note |
|---|---|---|---|---|
| Intros / year (run-rate) | ~3,935 | ~4,400 | ~5,400 (+37%) | Decomposition in §5.2 — against a −20% market ⇒ implied share ~+70% |
| Contribution after ad cost | ~$1.31M | ~$1.55M | ~$2.1M plan / ~$1.8M base | The number the year is judged on |
| Owned-channel intro share (organic + email + tool) | ~14% | ~20% | 25–30% | The fragility fix — compounds into FY28 |
| Lead → intro conversion | 28% | 30% | ≥33% | Lever 6; measured weekly in the pack |
| Blended cost / intro | $398 | $370 | ≤$330 | Mix shift + value bidding, not spend cuts |
| Revenue / intro | $732 | $750 | ≥$790 | Yield routing + hot-market coverage |
| Emailable, consented contacts | ~14,050 | ~15,500 | ≥18,000 | Tool capture ramp: 150/mo (Q2) → 350–500/mo (Q3–Q4) |
5.2 Intro decomposition — where +1,450 intros actually come from
| Source | FY26 run-rate | FY27 plan | How |
|---|---|---|---|
| Paid (Google + Microsoft) | ~3,385 | ~3,780 (+12%) | Flat budget; value bidding + estimate-tool landing CVR + leak-kills + geographic re-weighting. Deliberately modest — the auction is not the growth story. |
| Organic (existing + new pages) | ~550 | ~800 | Agent-evidence + suburb pages + /property-value hub + CRO lifting organic CVR 2.0% → 3.0% (that CVR lift alone ≈ +40–60 intros/yr on current traffic). |
| Estimate tool (direct + paid-assisted, net of cannibalisation) | 0 | ~350–450 | Ramp: 2k sessions/mo by Dec → 4k+/mo by Jun; ~8% contact capture; ~30% lead→intro. H2-weighted. |
| Email/nurture (lift vs holdout) | 0 | ~150–250 | Rescue (~137/mo qualified) + value-refresh + utility lanes; every claim holdout-measured. |
| Investor-exit campaign | 0 | ~100–150 | H2 only; content + email + modest exact-match paid on an unopposed cluster. |
| Total | ~3,935 | ~5,200–5,400 | Owned share lands 25–30% depending on tool/SEO ramp |
Sales-conversion lift (Lever 6) is embedded in the per-source rates above rather than double-counted as a separate line: +5pts lead→intro on ~3,300 paid+organic leads is worth ~165 intros and is the main mechanism behind the paid and organic lines improving without extra spend.
5.3 GP bridge (annualised contribution after ad cost)
| Component | GP impact | Basis |
|---|---|---|
| Market headwind on baseline | −$130k to −$250k | Westpac −20% turnover scenario, partially offset by hot WA/QLD/SA markets and the investor wave |
| L4 Paid efficiency (value bidding + leak-kills + guardrails) | +$150k to +$250k | Offline-value bidding typically improves CPA/value 15–30% (treat as industry heuristic, not a promise); June's named leaks (−$858, −$1,165 ad groups) show the recoverable floor |
| L6 Sales speed-to-lead + site CRO | +$100k to +$180k | +5pts lead→intro ≈ 165 intros × ~$700; organic CVR 2→3% ≈ +40–60 intros |
| L3 Email/nurture (lift vs holdout) | +$110k to +$180k | 150–250 incremental intros at ~$700+ GP; assumes ~1.5–2.5% annual intro conversion of the emailable base — OpenAgent's program implies the category supports this |
| L1 Estimate tool | +$150k to +$300k | §5.2 ramp math; both competitors pay for this intent today, which prices the demand as real |
| L2 Agent-evidence + suburb pages (H2-weighted) | +$80k to +$200k | ~250 incremental organic intros at ~2× GP; SEO lag is why the low end is low |
| L5 Investor-exit + yield routing + partner pilot | +$80k to +$150k | Time-boxed CGT wave + revenue/intro $732→$790 from routing |
| Component span | $1.85M – $2.6M (sum on baseline $1.31M) | Plan = $2.1M (below midpoint); base case = $1.8M if SEO and tool ramp slowly. Be honest that the ranges are planning judgment, not arithmetic certainty — the weekly pack exists precisely to correct course monthly. |
Every component is measured in the existing weekly pack (CRM source economics is the truth layer); email and investor-campaign components carry mandatory holdouts so lift is real, not attributed noise. If GP tracks below ~$1.55M run-rate at the December gate, §10's cash rule applies.
6. Lever detail and plays
Lever 1 Own the valuation moment
- Ship the estimate tool per the build spec (reports/wrea-property-estimate-build-spec-2026-07-02.html): P0 rails + P1 controlled beta in Q1, public by 1 September. It is simultaneously: a paid landing page (LAF/OpenAgent both pay for this intent), an organic asset, the email-capture engine, and sales enablement (valuation summary lands on every CRM lead).
- Build the /property-value hub: one authoritative page targeting the cluster (property value 49.5k, property value estimate 14.8k where WREA sits #15, property valuation 8.1k, appraisal 3.6k + the appraisal-definition cluster) with the tool embedded, suburb medians from the OTH/REA cache, and honest "estimate vs appraisal vs valuation" content that AIOs can cite.
- Convert dead traffic first:
/australian-house-prices-graph…(~880 users/wk, 0 leads, and a broken Chart.js error in Clarity) gets the tool CTA + market-data modules in week 1–2 of the quarter. Same treatment for the calculators family (61k vol, avg pos 7.1). - Targets: tool public by 1 Sep; ≥2,000 tool sessions/mo and ≥150 verified-email captures/mo by Q2 exit, ramping to 350–500/mo by Q4; WREA top-10 on "property value estimate" by Q3.
Lever 2 Weaponise the agent-evidence data
- Agent-evidence pages v2: upgrade
/best-agents/{suburb}(45 URLs already ranking) with the rankings API that already exists — real sold counts, median sold price, median DOM, fee context per suburb. Start with the top 100 suburbs by CRM GP, not all 17.2k. The template sells confidence: "here is the evidence, here is how we choose." - Suburb-market pages v1: the 141k-volume gap vs OpenAgent. Template: suburb medians + stock/DOM + agent evidence + fee ranges + estimate-tool CTA. 50–100 suburbs in Q2 (spring), 300 by Q3 where Search Console proves traction. Data freshness is automated from the existing caches — pages that update themselves are the anti-AI-slop defence.
- WREA Agent Market Index (data PR): quarterly data story from the 1.05M sold-history rows (e.g., "the gap between suburb top agents and the median on days-on-market"). Purpose: links, brand searches, AIO citations. Owner: Thomas, AI-drafted, one media pitch cycle per quarter — explicitly time-boxed so it can't sprawl. RMA does awards; WREA does evidence.
- Internal reuse: the same data feeds sales scripts ("3 agents sold within 400m in the last 12 months") and the estimate tool's agent-evidence teaser — one asset, four surfaces.
- Targets: top-100 agent-evidence pages live by 1 Sep (page generation by content/AI lane, not Felix); organic lead users 136/28d → ≥220/28d by Q3 exit; first Index story published October.
Lever 3 Monetise the database
- Execute the nurture spec (reports/wrea-email-nurture-build-spec-2026-07-02.html) exactly as phased: P0 safety (defuse the NOW-timed workflow, suppression, eligibility) and P1 rescue pilot in Q1; value-refresh + VIC local utility in Q2; scale + event alerts in Q3.
- SMS decision (Q2): 66% of rescue candidates have no valid email; the CRM already sends SMS. After the email rescue pilot reads out, decide a 1-SMS rescue step (fast-rescue-v2) with its own consent posture. Without it the rescue lane addresses a third of its audience.
- Every lane carries a 20% holdout — email GP claims in the weekly pack are lift-vs-holdout, never opens. The estimate tool feeds this lever: verified emails with scoped consent flow straight into value-refresh watches (the two specs share one contract).
- Targets: rescue live by mid-Q1; ≥150 incremental intros attributable to email by FY27 exit; unsubscribes <0.5%/send; zero suppressed-contact sends.
Lever 4 Re-plumb paid to bid on gross profit
- Value-based bidding: upload CRM intro events with GP-proxy values as offline conversions (build: Felix; validation + bidding migration: David); migrate Agent (National) from Target CPA $200 to tROAS once signal volume clears the floor (the governance doc already roadmaps this — FY27 executes it). Until then, weekly leak-kills stay manual: the −$858 VIC and −$1,165 NSW ad groups in June are the recurring pattern to automate away.
- Guardrails become code, not ledger items: the $39.2k-serving-vs-$25k-cap exposure (open P0) gets an automated weekly budget-exposure check with Telegram alert; the ended-but-enabled $8.4k campaign gets paused; account changes require a logged rationale — David's 30-changes week becomes a Monday review artefact, not archaeology.
- Microsoft: profitable both recent weeks ($294–371/intro) — resolve the 4-week-old platform-vs-CRM reconciliation, then scale within cap toward its best states (VIC/NSW/QLD CPAs ~$85–105).
- Geographic weighting: shift marginal budget toward Perth/Brisbane/Adelaide (rising markets, more transactions) and toward the exact-match state winners (QLD +$4.3k, Regions +$2.6k at 385% margin in late June).
- Targets: exposure P0 closed in July; value upload live by end Q1; tROAS live by mid-Q2; no negative-GP Google week in FY27 H2; blended CPI ≤$330 by Q4.
Lever 5 Deepen yield per seller
- (a) Investor-exit campaign (the FY27 one-off): the CGT-indexation change lands July 2027 with grandfathering — every property investor in the country has a 12-month decision window, and WREA already owns the search shelf (CGT calculator 9,900/mo, determine capital gains 5,400, ATO clearance 5,400 — zero competitor presence). Build: an investor-exit hub (CGT-change explainer + net-proceeds calculator chaining CGT calc → estimate tool → agent comparison), an investor email lane against CRM contacts flagged as investors, and a modest exact-match paid layer. Launch January 2027; peak into the pre-deadline autumn.
- (b) Yield routing: WREA's own data can score agents on intro→listing→fee conversion. Route intros preferentially to converters; fix coverage gaps in hot markets (WA/QLD/SA) where an unserved intro is pure lost GP. Revenue/intro $732 → $790+ without a single extra lead. (HomeLight's agent-scorecarding is the proven analogue.)
- (c) Ancillary referral pilot (small, Q3–Q4): two or three partnerships — conveyancer and mortgage broker referral fees on the existing lead flow, following LAF's conveyancing move and Clever/HomeLight's partner-distribution pattern. Success = a measurable second revenue line per seller with zero added headcount; failure = kill it in one quarter.
Lever 6 Convert more of what we already have
- Speed-to-lead measurement first (Q1, near-zero dev): add first-contact-time and attempt-count per lead to the weekly pack from the existing call logs. You cannot manage what isn't on the Monday page. Industry evidence is unambiguous that contact speed and attempt cadence dominate lead-gen conversion.
- Attempt-cadence standard (Q1–Q2): agree a sales-floor SLA (e.g., first call < 5 minutes in business hours, 6+ attempts across 72h, SMS on no-answer) and report compliance weekly. The email/SMS rescue lane (L3) is the automated safety net behind it, not a replacement for it.
- AI call-transcript pilot (Q2): transcribe a sample of the 173k recordings; mine for objection patterns, script wins, and the difference between converting and non-converting first calls. Feed findings into scripts and the decision-relief content library. Felix-light: transcription is a batch job, analysis is an AI lane.
- Site CRO cadence: one Clarity-informed funnel experiment per month (form length, step order, proof placement, estimate-tool entry points), starting with the organic form gap (2.04% vs paid 5.07%) and the known JS errors (btoa/Latin1, jQuery) already flagged in the weekly packs.
- Targets: lead→intro 28% → 30% by Dec → ≥33% by Jun; first-contact-time on the weekly pack by end July; transcript pilot readout by end Q2.
6.7 Data confidence and public-evidence dependencies
Added 2 Jul 2026. The levers above lean on WREA's proprietary data being right and defensible. Several in-flight data projects are therefore explicit FY27 dependencies with owners and gates — not background evidence. Each appears on the execution board with the ID shown.
| Dependency | FY27 operating rule | Gates / feeds | Board |
|---|---|---|---|
| Self-serve agent shortlist and source filtering | Self-serve is the next major customer-facing launch, but only as a controlled cohort. The displayed agent shortlist must pass an agreed accuracy check; no agent, agency or data unique to REA can be exposed; and current scraper/data-confidence evidence must be accepted. Total scraper-backlog completion is not required when the launch cohort itself is covered by accepted evidence. | Controls whether the existing self-serve journey can move from built to safely live; depends on the L2-6 evidence lane without waiting for every lower-priority backlog item | L2-8 |
| RMA public evidence (RateMyAgent sitemaps) | A strong positive-only public-evidence source — but only when driven by fresh sitemap snapshots with the sold-date guard (sold sitemap ≈ 507k rows rolling ~12 months; 81% coverage of recent REA sold). The stale legacy rma_listings table must not be consulted. Absence is never proof; individual RMA pages are DataDome-protected — sitemaps only. | Defensibility of listing-status claims behind L2 agent-evidence pages and sales conversations | L2-7 |
| REA listing-status reconciliation (ListingReconciler) | REA sold PDP sitemap = sold-candidate/public evidence. REA buy PDP sitemap = discovery/change detection only — presence ≠ live, absence ≠ withdrawn. Status corrections require GraphQL/agent-profile or exact-page proof, applied only through the existing safe ListingReconciler path (measure → verify → correct; no-write report first; Thomas-approved incl. legal coverage). | Hard gate for L2-2 (public agent pages must show reconciled counts) and for L5b yield routing (routing on wrong counts routes wrong) | L2-6 |
| OnTheHouse / ODIN valuation fallback | A cautious licensed/cache fallback: OTH suburb market-trends cache sits below the REA cache and above broad suburb/state defaults in the fallback order; it must never live-block a lead form; consumer-facing display approval (the DR2-style legal/commercial check) remains a standing gate. Thin-data suburbs (AVM medians with zero recent sales) never trigger "market moved" claims. | L1 estimate-tool fallback tier (N1 "fail useful") and L3 value-refresh email honesty | L1-8 |
| CloakBrowser / sidecar reliability | An execution and data-freshness dependency, not an afterthought: direct production reads have proved strong, but /fetch validation must become page-aware/adaptive before the sidecar is relied on as a broad verification path (reconciliation Step-2 checks, OTH warming). Until then, sidecar-verified corrections stay sample-gated. | Freshness of everything in L1/L2; reconciliation verify step | OPS-4 |
| CRM agent search / recommended-agent quality | Treated as part of sales conversion and trust (Lever 6), not a technical bug category: consultants sell from these surfaces and sellers see their output. Search/recommendation defects get triaged with the same weekly visibility as funnel metrics. | L6 lead→intro target; consultant trust in the data the whole strategy sells | L6-6 |
Principle: no proprietary data point goes public-facing (pages, tool, emails, routing) ahead of its confidence gate. The board makes these gates visible in the weekly review, so "data not ready" is a named blocker, never a silent assumption.
7. Quarterly implementation plan
Q1 · Jul–Sep 2026 Build the machine before spring (hard deadline: 1 September)
Felix's quarter, sequenced honestly (≈9 working weeks): the immediate customer-facing release is a controlled self-serve agent-connection cohort using the journey already built, after the L2-8 shortlist, source-filtering and scraper/data-confidence gates pass. The other committed lanes remain: guardrails + dead-traffic CTAs + speed-to-lead reporting; nurture P0 + P1 rescue; estimate tool P0 + P1; then beta hardening + agent-evidence template wiring. Agent-evidence page generation runs on the content/AI lane, not Felix. Offline-upload build is Felix (small), validation and bidding migration is David. Pre-declared descope order if the quarter slips: agent-evidence pages → tool P2 (emailed report, it's Q2 anyway) → utility email lanes. Rescue lane and tool P0/P1 do not slip; 1 September does not move.
| Lever | Deliverables | Owner |
|---|---|---|
| L4 (first!) | Close the budget-exposure P0 (July, week 1). Pause the ended-but-enabled campaign. Change-log discipline live. Offline GP-value conversion upload built and validating. Weekly leak-kill ritual in the Monday pack. | David + Felix + Thomas |
| L6 | First-contact-time + attempt-count metrics remain a reporting improvement; JS-error form fixes remain small hygiene work. General call-attempt optimisation is not the main Q1 launch. The Property Value Form 2.6% conversion readout moves into the Q2 Property Value Hub / paid-scale checklist. | Thomas + sales, Felix (metrics) |
| L1 | Estimate tool P0 rails + P1 controlled beta (per spec); DR1 listings verification (Thomas) and ODIN tier check closed; public launch 1 Sep. Dead-traffic conversion: house-prices-graph + calculators pages get tool CTAs (wk 1–2, near-zero dev). | Felix + Thomas |
| L3 | Nurture P0 (defuse workflow, suppression, eligibility, preview) + P1 rescue pilot with holdout (per spec). | Felix |
| L2 | Self-serve agent connection first: close the shortlist, source-filtering and scraper/data-confidence gates, then run a controlled production cohort. Continue the agent-evidence template v2 and top-100 suburb-page lane for 1 Sep, with public counts still gated by reconciled evidence. | Felix + Thomas (self-serve); Felix (template) + content |
| Gate | Q1 exit review: self-serve controlled cohort live with measurable selection-to-intro outcomes; tool public; rescue sending; agent pages live; value upload feeding Google; exposure P0 closed; speed-to-lead on the Monday page. Anything not live gets a named blocker and a date. | Thomas |
Q2 · Oct–Dec 2026 Spring harvest
| Lever | Deliverables | Owner |
|---|---|---|
| L4 | tROAS live on Agent (National) once signal floor met; scale Microsoft within cap; shift marginal budget to WA/QLD/SA and exact-match winners. Paid landing traffic progressively pointed at the estimate tool (A/B vs standard funnel). | David |
| L1/L2 | /property-value hub live; suburb-market pages 50–100 (spring cohort); estimate-tool embeds on suburb + calculator pages; first Agent Market Index story (October, Thomas + AI-drafted) with PR push. | Felix + content + Thomas |
| L3 | Nurture P2: value-refresh lane (stage-branched) + VIC local utility monthly send; emailed estimate report + magic-link verification live (tool P2). SMS rescue decision after email-rescue readout. | Felix, Thomas (SMS call) |
| L6 | AI call-transcript pilot (sample of 173k recordings) → script/objection readout; CRO experiment cadence running (1/month); organic CVR trending toward 2.5%. | Thomas + AI lane |
| Gate | Q2 exit review: owned-channel intro share ≥20%; tool ≥2k sessions/mo, ≥150 verified captures/mo; lead→intro ≥30%; no negative-GP Google week since tROAS; suburb pages showing Search Console traction (else pause expansion). GP run-rate check vs $1.55M (§10 cash rule). | Thomas |
Q3 · Jan–Mar 2027 The investor-exit wave
| Lever | Deliverables | Owner |
|---|---|---|
| L5a | Investor-exit campaign launch (January): hub + net-proceeds calculator chain, investor email lane (CRM investor segment, holdout-measured), exact-match paid layer, February Index story angled on investor selling. The quarter's headline swing. | Thomas + all |
| L5b | Yield routing v1: agent conversion scorecard from own data; routing rules with sales team; hot-market coverage gap-fill (recruit agents where intros go unserved). | Thomas + sales |
| L2/L3 | Suburb pages → 300 where proven; NSW/QLD utility email expansion; market-event email capability (rate decisions) live behind preview gate. | Felix + content |
| L5c | Partner pilot: 2–3 conveyancer/broker referral agreements on existing flow. | Thomas |
| Gate | Q3 exit review: investor campaign CPL/intro economics vs core funnel; routing lift on revenue/intro; lead→intro ≥32%; partner pilot go/kill. | Thomas |
Q4 · Apr–Jun 2027 Compound and decide
- Scale what the gates proved; kill what they didn't (explicitly — a lever that missed two consecutive gates does not survive into FY28).
- Investor wave peak into the July 2027 deadline; second Index story cycle; suburb/agent page refresh automation.
- FY28 options memo with full-year owned-channel data: ancillary revenue expansion vs geographic deepening vs (only if the core compounds) net-new surfaces.
- Full-year review against §5 targets.
8. Operating rhythm and scorecard
- Monday growth pack (exists): unchanged as the weekly truth layer, plus four new blocks as levers ship: email lift-vs-holdout, tool funnel, owned-channel share, speed-to-lead/attempt compliance. The four-verdict paid governance stays.
- Monthly lever review (new, 60 min): each lever gets green/amber/red against its §6 targets, one decision per amber/red. No status theatre — decisions only.
- Quarterly gates (§7): pre-agreed exit criteria; the gate meeting reallocates budget and Felix-time for the next quarter.
- FY27 scorecard (top of every weekly pack): contribution after ad cost (rolling 12-wk, annualised) → $2.1M; owned-intro share → 25–30%; lead→intro → ≥33%; blended CPI → ≤$330; revenue/intro → ≥$790; emailable consented contacts → 18k; Google GP-margin floor → no negative weeks in H2.
9. What we are deliberately NOT doing
- Not increasing total paid budget until value bidding has a clean quarter. The auction is where the fragility lives; feed it efficiency, not cash.
- Not building all 17.2k suburb pages. 100 → 300, expansion gated on Search Console proof. Thin programmatic pages are an AIO-era liability.
- Not chasing property-management/landlord leads (LAF's adjacency) — dilutes the seller-intro engine and the sales team's focus.
- Not launching embedded finance (HomeLight-style equity products). Right direction, wrong balance sheet and team size; the partner pilot is the FY27-appropriate version.
- Not doing brand advertising (TV/radio/sponsorships). The Index/data-PR play is the small-team substitute; revisit only if FY27 compounds.
- Not adding new report/dashboard infrastructure. The weekly operating system already works; FY27 spends build capacity on customer-facing assets.
- Not hiring ahead of proof. One content/AI-operator contractor is the only contemplated addition, and only if the Q1 gate shows page production is the bottleneck (§11).
10. Risks and pre-agreed responses
| Risk | Pre-agreed response |
|---|---|
| Turnover falls harder than −20% (Westpac hike path materialises) | Verdict system already handles it: cash-protection mode, paid floor at proven exact-match winners, levers 1–3+6 continue (cheap and compounding). The investor-exit wave is counter-cyclical to this risk. |
| Cash rule (the base-case trigger): GP run-rate below ~$1.55M annualised at the December gate | Freeze all H2 discretionary spend (partner pilot, Index PR, suburb expansion beyond 100), hold paid at winners-only, and re-plan H2 around the two levers with the best measured GP-per-effort. No new builds start until the run-rate recovers. |
| Google volatility / AIO expansion into local queries | Owned-share target is the hedge; tools + first-party data + structured citations are the AIO-resilient assets. If organic informational clicks fall, tool and agent-evidence pages (transactional) hold. |
| Paid engine has another negative-GP month before value bidding lands | Guardrails-as-code in July; weekly leak-kill; the ~$30k conditional weekly boundary is real — a breach without a logged decision is an incident, not a surprise. |
| Execution capacity (one dev) | Sequenced plan, both Q1 builds fully specced, gates kill scope, and the Q1 descope order is pre-declared in §7 (agent pages slip first; rescue + tool P0/P1 never slip). If Felix is blocked >2 weeks, the descope order executes automatically — no re-litigating. |
| Key-person risk: David (paid) and Felix (dev) | Paid: change-log + guardrails-as-code turn tribal knowledge into system state (Q1 week 1); the weekly pack already lets Thomas read the account cold. Dev: both build specs are written to be executable by any competent Laravel dev — that's deliberate redundancy. |
| ODIN/data dependency + data rights | Handled in the estimate-tool spec (DR1/DR2 owner gates — DR1 = Thomas verifying sold-listings data is public-domain before comparables ship; DR2 = ODIN paid-tier display check; budget breaker, suburb-tier fallback, swappable source interface). |
| Investor wave doesn't materialise | Campaign is content + email + modest exact-match paid — low sunk cost; the CGT content assets retain evergreen value regardless. |
11. Decisions needed from you now
- Adopt the FY27 targets (§5) — including the explicit plan-vs-base framing — or adjust before Q1 locks.
- Confirm 1 September as the hard spring deadline and ratify the Q1 descope order in §7 (this is the real decision — what slips first when something slips).
- Capacity: approve in principle one content/AI-operator contractor if the Q1 gate shows page/content production (not dev) is the bottleneck — pre-approving the trigger avoids losing a month to the decision in October.
- Cash rule: ratify the December trigger (§10) — below ~$1.55M GP run-rate, H2 discretionary spend freezes automatically.
- DR1/DR2 closure (sold-listings public-domain verification; ODIN tier display check) — both are yours and gate the estimate tool's comparables and launch.
- Approve the paid re-plumb (offline GP-value upload + guardrails + change-log) with David — it changes his weekly workflow, and §10 makes his account state system-readable.
- Sales SLA: agree the speed-to-lead standard with the sales floor (Lever 6) — the metric ships in July either way; the standard needs your authority.
- Investor-exit campaign: green-light the January launch so content and the CRM investor-segment work can be scheduled in Q2.
- Partner pilot appetite: yes/no to spending your own time on 2–3 conveyancer/broker conversations in Q3.
12. Appendix: evidence and sources
- CRM economics: live 12-week pull via scripts/report_sales_funnel_condensed.sh (2026-07-02); weekly reports reports/wrea-growth-operator-weekly-2026-06-{01,08,15,22}-*.html (verdicts, campaign GP, ledger P0/P1s, Auction Insights, Clarity). Where weekly-report snapshots and the live pull differ slightly (data settling), this document uses the live pull.
- Build specs (execution-ready): reports/wrea-email-nurture-build-spec-2026-07-02.html · reports/wrea-property-estimate-build-spec-2026-07-02.html.
- Data platform: docs/rea-top-agents-architecture-2026-05-02.md (42k agents, 599k listings, 1.05M sold rows, rankings API) · docs/rma-rea-public-evidence-handoff-2026-06-23.md (2.94M-row listings corpus, RMA coverage analysis).
- Call/sales data: funnel access report (573,390 call-log rows, 173,142 recordings; transcripts not yet first-class) — the Lever 6 substrate.
- Governance: docs/weekly-paid-media-governance-review.md · docs/wrea-growth-operating-system.md.
- Market thesis: docs/wrea-staff-meeting-global-trends-2026-06-25.html (seller journey more cautious/researched; WREA = confidence-building).
- External (verified 2026-07-02): RBA cash rate 4.35% (June hold after Feb/Mar/May hikes); Westpac housing update 26 May 2026 (turnover −20%, investor activity −34%, capital-city price forecasts); Cotality May 2026 HVI (national zero growth, Sydney/Melbourne declines, Perth +25.8% y/y); May 2026 Budget CGT-indexation change effective July 2027 with grandfathering; OpenAgent A$23.6M raised / ~2.1M visits; RMA Global FY25 A$21.5M revenue, FY26 forecast cut (aggregator-sourced); LocalAgentFinder Dott & Crossitt acquisition (2024); HomeLight Buy-Before-You-Sell US$20M + EVA AI closing (Apr 2026); Clever Pro lender white-labelling; AI Overviews click-loss studies (Semrush et al., US-centric) with real-estate transactional queries least affected. Unverified flags: OpenAgent 2025–26 internal activity; Clever revenue estimate; RMY FY26 guidance.
Strategy authored by Claude (Fable 5) on 2026-07-02 for Thomas Roberts, from live CRM economics, four weekly growth reports, the July 2026 build specs, WREA's internal platform documentation, and independently verified market/competitor research. v1.1 — revised after two independent adversarial review passes (numbers reconciliation; cold board-advisor read). v1.2 (2 Jul) — added §6.7 data-confidence and public-evidence dependencies (RMA snapshots, REA reconciliation, OTH/ODIN fallback, sidecar reliability, CRM agent-search quality) with board IDs. v1.3 (27 Jul) — refined the current execution order: self-serve agent connection is the next controlled customer launch; its shortlist, source-filtering and scraper/data-confidence gates are explicit; Property Value Form call-depth analysis moved to the later Property Value Hub rollout checklist.