CASE STUDY — OutPost · GROWTH / DEMAND GENERATION · PROGRAMMATIC SEO & LEAD ROUTING

OutPost: A $41 Lead, Down to $11 in Four Cycles.

OutPost runs ten thousand local business sites, and the buyer takes the enquiries. The hard part was never generating ten thousand of them. It was that every one has to be judged and killed on a ninety-day clock, and no team does that by hand.

BY Priya, Research Editor | September 6, 2026 · SEPTEMBER 6, 2026

$11

BLENDED COST PER LEAD, DOWN 73% OVER FOUR CYCLES

52K QUALIFIED LEADS IN A ROLLING 12 MONTHS, ALL MARKETS — 4.1M IMPRESSIONS · 918K CLICKS · 71K FORM FILLS
MEDIAN TIME TO PAGE ONE 46 DAYS → 14 DAYS · 71% OF FLEET ON PAGE ONE · 44% IN POSITIONS 1–3
1,860 SITES CULLED AND THEIR DOMAINS RELEASED · 0 MANUAL SEO HOURS
412K SESSIONS ACROSS THE LIVE FLEET IN THE LAST 30 DAYS

01 — THE PROBLEM

Ten thousand domains, and one classifier standing in front of them

Ten thousand domains. One classifier. No template that survives repetition. Local lead generation is decided in the first three positions. Positions 1–3 take a 31.4% click-through rate. Positions 4–10 take 6.2%. Position 11 and below takes 0.4%. A site off page one is not underperforming; it is not earning. Hand-build the sites and you never reach ten thousand. Stamp one template ten thousand times and you reach it, then get filtered, because near-identical pages across thousands of domains is the exact signature the classifier was built to catch. Both routes end the same way. Either you never get there, or you get there and get caught.

02 — INSIDE THE PIPELINE

Nine steps, no hands, and no two sites alike

There is no template, because the template is the signature. We write each site's copy against its keyword's intent, then run it through a humanisation pass until it clears AI detection. Nothing crosses between two sites: not a layout, not an image, not a sentence. The pipeline runs nine steps and no hands: keywords, generate, copy, imagery, domain, deploy, index, capture, leads. We score keywords for intent against live SERP data rather than volume alone. Domains get registered, configured and secured without a person in the loop. A new site goes live within minutes, submits itself for indexing, and reports its own rank from day one. Form fills land in the buyer's CRM or ours, and the follow-up sequence starts on its own.

Unique design isn't polish here. It's the survival condition.

03 — CULLING WHAT DOESN'T EARN

Deleting sites turned out to be the skill that mattered

Building is the easy half. Every site is instrumented from the day it indexes, and rank, click-through and lead attribution are read daily, per site and per keyword. One deadline, two rules: a site not on page one by day 90 is re-targeted, and a site with no leads by day 90 is deleted and its domain released. Position 11 earns nothing, so nothing off page one is worth optimising. Last cycle's ledger reads one way. "ac repair austin" produced zero leads and came back as "emergency ac repair south austin". "dental implants" sat at rank 34; its replacement is "same day dental implants plano". "roof repair" cost $148 a lead before it became "hail damage roof repair tulsa". "lawyer near me" was simply too broad, and now reads "workers comp lawyer mesa az". Broad head terms lose, and the loop narrows intent until a lead arrives. 1,860 sites have been culled so far. Each one taught the engine which intent patterns Google rewards in that market, and that feeds what gets built next. The sites are the visible part. The part that compounds sits underneath them.

04 — WHAT IT BECAME

From $41 a lead to $11, and faster to page one every cycle

Cycle one ran at $41 a lead and took a median 46 days to reach page one. Cycle four runs at $11 and takes 14, with $26 and $17 in between. The fall did not come from adding sites. Leads climb faster than sessions because the fleet keeps growing while the losers come out and the keyword set narrows on what converts. Over a rolling twelve months, across all markets, the fleet has produced 4.1M impressions, 918K clicks at 22.4%, 71K form fills at 7.7%, and 52K qualified leads. 71% of it sits on page one and 44% in positions 1–3, on zero manual SEO hours. Two limits worth naming. Qualified means a form fill that cleared our filters, 52K out of 71K; what happens after the handoff is the buyer's ledger, not ours, and we publish no close rates. And these are our numbers, blended across the markets we already run. A new vertical and geography starts at cycle one, at cycle one's price.

Want a fleet pointed
at your market?

Talk to the team Agentic AI service →

FAQ

Questions about this product

Ten thousand sites in one market sounds like spam. Why doesn't Google treat it that way?

Because a classifier built to find repetition needs something to cluster on, and there is nothing to find. No two sites share a layout, a stylesheet, an image or a sentence, and the copy is humanised until it clears AI detection. The honest version of this answer is that we do not get to decide whether it worked. Google does, and it re-decides every day.

What happens when the algorithm changes?

We can't promise immunity, and we won't. What we can promise is that nothing in the fleet gets defended. A ranking change surfaces as a lead change within days rather than at the end of a quarter, and the day-90 rules fire on it without a meeting. Any single site is disposable by design, and that is the hedge. What we cannot hedge is a shift that moves an entire market's SERP, and if that happens you will hear it from us in the same week we see it.

Do I get the $11 cost per lead, or the $41 one?

Cycle one, which on our own curve started at $41. A new vertical and geography enters at the top of the ladder, not at the bottom of ours. Before anything is built you get the keyword set, its search volume, and a cost-per-lead target for cycle one, so what you agree to is a number rather than a promise. There is no retainer until the first leads land.

Your numbers are blended across all markets. How do I know they hold in mine?

You don't yet, and neither do we. What we publish is a blend across the markets we already run, not a per-vertical guarantee, and we won't pretend a market we haven't run behaves like one we have. What does carry across is the mechanism: intent-scored keywords, one distinct site per keyword, daily instrumentation, and a ninety-day deadline that either re-targets a site or deletes it. Bring the vertical and the geography, and cycle one is where you find out.