Meta Prospecting: the funnel doesn't break before search — it breaks at search
Meta prospecting spent $4,238 to produce 1,306 "leads" and 1 qualified buyer. The leads are real — they deliberately navigate from the ad's single home to the search page. But once there, they go passive: only ~1% run a real search, and almost none save, tour, or qualify.
The bottleneck isn't audience quality alone. It's that a cold visitor lands on the search page and is given no reason to act. That's a landing-experience problem, and it's fixable.
What actually happens
The old assumption was "they land on one property and bounce." That's true for the 87% who never become leads — but it's wrong for the leads. The leads take a real second step:
The property→search move is genuine, not an app redirect — only 1% made the jump in under 3 seconds; the median gap is 67 seconds of real time on the home first. So they do the hard part: leave a single listing to go find more.
Then the funnel collapses at the search page. Of the whole cohort:
- 1% refined a search (adjusted filters)
- 3% used real search (AI query or autocomplete)
- 0.6% saved a search · 0.2% requested a tour · 0.2% qualified
They arrive at a page full of homes and simply… stop.
The replays confirm it
Session replay only records real humans (bots can't run the recorder), so this is a clean, bot-free view. Even among the leads who reached the search page, the median session runs ~16 seconds, and 11 of 14 sampled sessions were under 30 seconds. A rare power-browser (one ran 13 minutes / 619 clicks) carries almost all the real engagement; the typical visitor glances and leaves.
A secondary finding worth an eng note: the landing experience is inconsistent — some prospecting ads drop users on a property page, some on a raw /search page, and some on the marketing homepage. A cold click deserves a deliberate, consistent first screen.
Prospecting vs. retargeting (per lead)
The same behavior split shows why warm audiences outperform: retargeting reaches people who already have intent, and it converts them 3–6× better on every high-intent action.
The prospecting campaigns, ranked
| Campaign (UTM) | Spend | Leads | CPL | Qual |
|---|---|---|---|---|
| prospecting-v1 (DPA full catalog) | $1,359 | 380 | $4 | 0 |
| ddcold-v1 (Cold Speedeon)* | $677 | 296 | $2 | 0 |
| fl-intent-v1 (Speedeon FL) | $672 | 228 | $3 | 1 |
| moneyflow-v1 | $448 | 166 | $3 | 0 |
| tx-intent-v1 (Speedeon TX) | $572 | 141 | $4 | 0 |
| conquest-v1 | $189 | 48 | $4 | 0 |
| Prospecting total (16 lines) | $4,238 | 1,306 | $3 | 1 |
*ddcold (Cold Speedeon) was paused Aug 3 — it optimized a cold audience toward a deep onboarding event, failed to exit Meta's learning phase, and was throttled to ~$1/day. Cheap CPL, zero downstream.
Watch it yourself
Representative replays (PostHog login required):
Recommendation
Fix the search landing before spending more on prospecting
Prospecting successfully gets cold visitors to the search page — and loses them there. The highest-leverage move is not more budget; it's what a cold visitor sees when they hit /search: relevant pre-filled results, an obvious save-search prompt, and an AI-search nudge, so arriving at search leads somewhere. A consistent first screen for every cold click would help too.
- Reallocate toward warm/retargeting — it converts 3–6× better per lead; money should keep flowing from cold prospecting to warm audiences (especially Active Savers, which qualifies well but is budget-starved).
- Treat prospecting as top-funnel, measured on cost-per-real-search, not raw "leads" — the lead count is a /search pageview and overstates intent.
- Retire lines that reach search but never search — high raw leads, zero downstream (ddcold done; watch prospecting-v1 and the Speedeon intent lines).