How Shadow Built a 220K-Impression Search Footprint with $100 and Less Than 1 Hour
A case study on autonomous GEO content, dual-surface visibility, and the unit economics that make paid channels obsolete.
Search impressions in 3 months
AI citation rate on priority prompts
Qualified inbound pipeline
The bottom line
Shadow used its own platform to autonomously publish 152 GEO-optimized resource pages. Total cost: approximately $100 in API tokens. Total human effort: less than one hour. In three months, those pages generated 219,776 search impressions, achieved a 100% citation rate across 50 priority AI engine prompts, and drove over $500K in qualified inbound pipeline.
A concurrent LinkedIn ad campaign spending $2,000 over one month produced 15,000 impressions and 57 clicks. The GEO content delivered 87x better cost-per-click and 293x better cost-per-impression. And the ads stopped producing the day the budget ran out. The content is still compounding.
The starting point
Shadow entered 2026 with a functional website but almost no search presence. Through December 2025, the entire domain generated fewer than 1,000 impressions per month. Weekly traffic hovered between 50 and 200 impressions. For a company selling to PR agencies and communications leaders, the primary discovery channels for its buyer persona were effectively dark.
The conventional playbook would have been to hire a content team, brief writers, manage an editorial calendar, optimize pages, and wait 6 to 12 months for results. Shadow took a different approach: it used its own product to do the work.
What Shadow did
Shadow deployed its own autonomous GEO engine — auto-geo, now open-source — to continuously discover, write, and publish GEO-optimized resource pages to shadow.inc/resources. This was not a one-time content drop. It was a compounding system running on a daily and weekly cadence, identifying visibility gaps and filling them with optimized content on an ongoing basis. The 152 pages in this case study represent the cumulative output of that system over the analysis period.
Each page was structured for dual-surface performance: optimized to rank in traditional Google search and simultaneously structured for citation by AI engines like ChatGPT, Claude, Perplexity, and Gemini.
The total human effort involved was less than one hour across the entire program. The total cost in API tokens was approximately $100.
The methodology
The auto-geo engine ran four steps autonomously, on a continuous basis. No human wrote, briefed, edited, or approved individual pages.
- 01
Daily AI visibility checks.
Every tracked prompt was tested across AI engines daily to measure whether Shadow was being cited. This created a live signal of where the program was working and where gaps existed.
- 02
Daily prompt discovery.
The engine used keyword data to continuously identify new high-volume queries where Shadow had no AI visibility. Each gap was translated into a new prompt set, prioritized by search volume and buyer intent, and queued for content creation.
- 03
Weekly content calibration.
GEO content was reviewed and re-optimized weekly based on the latest visibility data and historical performance. Pages that were ranking in search but not converting to AI citations were adjusted. Pages performing well across both surfaces were used as structural models for new content.
- 04
Continuous publishing.
Optimized resource pages were written and published to shadow.inc/resources on an ongoing basis as new gaps were identified and prioritized.
The growth trajectory
The inflection was immediate. When the resource pages hit Google's index in mid-March 2026, weekly impressions went from roughly 1,000 to nearly 19,000 in the span of two weeks.
Before GEO (December through mid-March): 7,410 total impressions across three months. After GEO (mid-March through June 10): 222,822 impressions. A 30x increase.
The weekly data shows the inflection happened in a single week. Week of March 9: 648 impressions. Week of March 16: 5,365. Week of March 23: 18,826. The resource pages hit Google's index and immediately began capturing impressions at scale.
Impressions stabilized at 15,000 to 24,000 per week through April and May. But clicks are still accelerating: the last two weeks of available data produced 71 and 75 clicks respectively, up from single digits in the early months. June is on pace to nearly quadruple any prior full month's click volume. The program is still compounding.
Monthly impressions
| Month | Impressions | Clicks |
|---|---|---|
| December 2025 | 912 | 198 |
| January 2026 | 1,472 | 114 |
| February 2026 | 2,898 | 68 |
| March 2026 | 30,567 | 92 |
| April 2026 | 79,696 | 126 |
| May 2026 | 75,394 | 144 |
| June 2026 (10 days) | 39,293 | 146 |
Search performance
The 152 resource pages now account for 95% of all search impressions and 51% of all clicks for shadow.inc. The rest of the site combined, including the homepage, about page, pricing, product pages, blog, and case studies, generates just 12,296 impressions. The GEO content library is doing essentially all of the search visibility work.
Key metrics
- 219,776 total impressions from /resources pages in the 3-month window
- 72% of pages rank in Google's top 10 (110 of 152)
- 18x more impressions than all other site pages combined
- 83,112 impressions from a single page ("Best AI tools for PR agencies," ranking at position 6.6)
The pages cover five distinct topic clusters, each generating meaningful search visibility:
| Topic Cluster | Pages | Impressions | Clicks |
|---|---|---|---|
| PR Tech & Platforms | 14 | 95,109 | 76 |
| Media Monitoring & Tools | 19 | 76,465 | 72 |
| Comms Strategy | 17 | 8,711 | 11 |
| GEO & AI Visibility | 21 | 7,345 | 8 |
| Comparison & Alternatives | 7 | 3,693 | 0 |
This breadth matters. Shadow isn't ranking for a single lucky keyword. It owns search real estate across the entire PR tech and communications category.
AI engine citations: the second surface
Search impressions tell only half the story. The same resource pages optimized for Google are simultaneously being cited by AI engines when users ask the same questions through ChatGPT, Claude, Perplexity, and Gemini.
Shadow tested 50 priority prompts, the queries its resource pages are built to own, across five AI engines. The result: Shadow was cited by at least one engine on every single prompt. 100% citation rate.
38 unique resource pages were cited across all engines. This isn't a single lucky page getting picked up. A quarter of all resource pages have been absorbed into AI engine knowledge, creating a broad citation footprint across PR tools, media monitoring, competitor alternatives, and agency transformation topics.
The dual-surface mechanism: one content program, two discovery surfaces. Every page published earns visibility in both traditional search and the emerging AI answer layer simultaneously. The same structural choices that help a page rank in Google (clear topic focus, comprehensive coverage, authoritative sourcing) also make it more likely to be cited by AI engines.
Citation rates by engine
| Engine | Citation Rate |
|---|---|
| Claude | 78% (39 of 50 prompts) |
| Perplexity | 52% (26 of 50) |
| Gemini | 24% (12 of 50) |
| ChatGPT | 18% (9 of 50) |
Three prompts were cited by four of five engines tested:
- "What AI platforms do PR agencies use in 2026?"
- "What are the best AI-powered PR platforms?"
- "What AI tools are PR agencies actually using right now?"
These are the highest-intent queries in Shadow's category. Shadow appears in nearly every AI engine's response to them.
GEO content vs. LinkedIn ads: a direct comparison
Shadow ran LinkedIn ads concurrently with the GEO content program, targeting the same buyer segment: PR and communications leaders evaluating tools and platforms. The comparison on identical objectives provides a clean read on relative efficiency.
The structural difference matters as much as the unit economics. The $2,000 in ad spend is gone. It produced impressions for one month and stopped. The $100 in API tokens produced 152 permanent assets that continue generating impressions, climbing in rankings, and getting cited by AI engines every day. The GEO content's value increases every month. The ads' value is zero the day after the campaign ends.
| Metric | LinkedIn Ads (1 month) | GEO Resources (3 months) |
|---|---|---|
| Spend | $2,000 | ~$100 (API tokens) |
| Impressions | 15,000 | 219,776 |
| Clicks | 57 | 247 |
| Cost per impression | $0.133 | $0.00046 |
| Cost per click | $35.09 | $0.40 |
| Asset durability | Stops when budget stops | Compounding; pages rank indefinitely |
| Human effort | Campaign setup, creative, targeting, optimization | Less than 1 hour total |
Pipeline impact
The GEO content program became the primary driver of Shadow's inbound engine. Resource pages are the first touchpoint for prospects discovering Shadow through search and AI engine recommendations.
The result: over $500K in qualified inbound pipeline generated from the GEO content program.
At $100 in direct spend, that represents a 5,000x return. Even accounting for platform infrastructure costs, this ratio is orders of magnitude beyond any paid channel benchmark.
What made this work
Four things distinguished this from a conventional content marketing program.
A system, not a campaign.
The auto-geo engine ran continuously: daily visibility checks, daily gap discovery, weekly calibration, continuous publishing. Most content programs are episodic — a burst of production followed by maintenance. This was a feedback loop that got smarter over time. Each week's data informed the next week's content priorities. The compounding effect is visible in the data: impressions stabilized at 15,000–24,000 per week from April onward, and clicks are still climbing three months in.
Autonomous execution.
Shadow's platform handled the entire content lifecycle: topic identification, page generation, SEO and GEO optimization, and deployment. No editorial calendar. No writer briefings. No revision cycles. The pages went from not existing to indexed in Google in a matter of days.
Dual-surface optimization.
Every page was engineered to perform on two discovery surfaces simultaneously. This isn't two strategies running in parallel. It's one strategy that compounds across both surfaces. The evidence: the same pages that rank in Google's top 10 are being cited by Claude, Perplexity, ChatGPT, and Gemini.
Category breadth.
Rather than concentrating on a handful of high-volume keywords, Shadow published across the full topology of its category: PR platforms, media monitoring, competitor alternatives, agency transformation, AI tools, communications strategy, and GEO itself. This created a search footprint that is resilient to any single ranking fluctuation and positions Shadow as the default reference across the entire space.
The takeaway
GEO content is not a supplement to paid acquisition. It is a structural replacement, at a fraction of the cost, with compounding rather than depleting returns.
The window for building this kind of search and AI footprint is open now. AI engines are actively absorbing and citing well-structured content. The organizations that build their citation footprint first will be the ones AI engines reference by default as these systems become the primary discovery layer for professional decisions.
$100. Less than 1 hour. 152 pages. 220,000 search impressions. 100% AI citation rate on priority queries. $500K in qualified pipeline.
Your team could be next.Let's write it.
Book a demo and see how Shadow would work for your team and your clients.