Direct Answer
The Blood in the Water SEO technique starts with evidence in the search results, not a keyword-difficulty score. Find an authoritative website that ranks well with an individual page that appears weak, generic, outdated, poorly targeted, or lightly linked. Then mine that domain for more of the same pattern and build a page that serves the query better.
The insight is useful because Google is already showing that a less-than-perfect page can satisfy the result set. That is a stronger clue than a tool labeling a keyword "easy." It is still only a clue. The competing domain may be carrying the page, and your site may not have the same authority, relevance, or link support.
A weak page in the results creates a hypothesis. Live SERP data, page-level links, your own Search Console history, and a genuinely better answer determine whether the hypothesis deserves production time.
Watch the Method
Credit and evidence: the method and creator demonstration come from Vasco's SEO Tips, published on 24 September 2026. The creator uses his own product, Arvow Chat, in the demonstration and explicitly acknowledges that bias. The reported ranking speed is a creator result, not a guarantee.
The Shift: From Keyword Difficulty to Result Weakness
Traditional research commonly starts with volume, difficulty, cost per click, and intent. This technique reverses the sequence:
- Find a result where Google already ranks a page with visible weaknesses.
- Identify the strong domain carrying that page.
- Inspect the domain's other ranking URLs for the same pattern.
- Group related queries into one useful page or a coherent topic cluster.
- Validate the opening against your own site's ability to compete.
The creator calls these accidental rankings: pages that may rank more because of the host domain than because the page was deliberately built for the query. The label is memorable, but the cause is an inference. We cannot see Google's ranking calculation. Treat "accidental" as shorthand for ranking despite observable page-level weaknesses.
A Six-Step Blood in the Water Workflow
1. Start with a seed market, not a giant keyword list
Choose one commercial problem, product category, service, or audience. Search several representative queries manually. Record the result type, intent, freshness, location, and recurring domains.
2. Find the strong-domain, weak-page pattern
Look for broad category pages, forum threads, LinkedIn posts, Reddit discussions, thin listicles, stale articles, and pages whose title and H1 do not closely match the query. One weak result is interesting; the same domain repeating the pattern is a research lead.
3. Mine the domain's ranking URLs
Export the domain's ranking keywords and URLs. Collect current position, search volume, estimated traffic, cost per click, page-level referring domains, backlinks, title, heading, content date, page type, and likely intent.
4. Detect and cluster the openings
Filter for useful positions with weak page-level support. Group queries by the same user need, not by superficial wording. One complete page should cover close variants when they share intent; creating near-duplicate pages invites cannibalization and low-value scale.
5. Score fit for your site
Add business value, topical relevance, evidence you can contribute, and your site's current impressions. Subtract authority mismatch, dominant brands, unstable results, legal or medical risk, content cost, and queries that need a product or tool you cannot provide.
6. Publish, connect, and review
Build the page, link it from relevant existing pages, ensure it is crawlable, and submit a sitemap through your normal process. Review the live result for usefulness and accuracy before measuring impressions, clicks, assisted conversions, and qualified leads.
A Practical Weak-Page Scorecard
This is an implementation heuristic, not a Google ranking formula. Score each candidate consistently so the team can compare opportunities instead of choosing whichever keyword looks exciting.
| Signal | What to inspect | Suggested effect |
|---|---|---|
| Weak intent match | The page answers a neighboring question, not the actual task | +20 |
| Generic or UGC result | Forum, profile, category, social post, or broad directory | +12 |
| Low page-level link support | Few relevant referring domains to the exact URL | +15 |
| Poor query targeting | Main concept absent from title, H1, URL, or opening answer | +10 |
| Outdated or incomplete coverage | Old facts, broken steps, missing examples, no first-hand evidence | +15 |
| Your topical fit | You already have related pages, data, experience, or products | +15 |
| Existing Search Console signal | Your site already earns impressions for the topic | +13 |
| Authority mismatch | Every result is a dominant specialist or official source | -20 |
| High evidence or production cost | The useful answer needs testing, data, video, a tool, or expert review | -15 |
| Low business relevance | Traffic is unlikely to create a subscriber, lead, sale, or product action | -20 |
Use the score to create a review queue. Manually inspect the top candidates before anyone commissions content.
The Data Stack the Method Actually Needs
An LLM can classify and summarize the evidence. It should not invent the evidence. The minimum useful stack is:
| Data | Why it matters | Minimum check |
|---|---|---|
| Live SERP snapshot | Confirms what ranks now, result types, intent, and volatility | Inspect the target country, language, and device |
| Ranking keywords and URLs | Turns a competitor into a candidate database | Keep the URL, position, query, and capture date together |
| Page-level backlinks | Separates domain strength from support for the exact page | Review relevance and quality, not only the count |
| Search volume and commercial data | Helps estimate demand and business value | Use ranges and compare more than one signal |
| Google Search Console | Shows where your own site already earns impressions and clicks | Compare query and page trends, not one average position |
| Manual page review | Tests intent, originality, accuracy, freshness, and usefulness | Read the result before calling it weak |
In the demonstration, Arvow Chat combines project context with live SEO tools, SERP research, backlinks, and connected Search Console data. Its current product page says Search Console access is read-only, chat messages themselves do not consume credits, and tool-backed article or Autoblog actions remain subject to plan limits. It also states that Chat is not a fully autonomous agent.
You can reproduce the research with another stack, including OpenSEO, DataForSEO, Semrush, Ahrefs, Search Console exports, or custom APIs. The requirement is current, attributable data, not a particular interface.
The Official Blood in the Water Prompt
Arvow publishes the complete prompt at Blood in the Water SEO Technique. The version below preserves its structure while formatting it for practical use.
Build an SEO keyword-research tool that finds large, authoritative websites ranking for many keywords with weak, under-optimized individual pages. These are called "accidental rankings." A typical example is a DR 70 domain ranking third while the exact URL has zero to three referring domains, poor keyword targeting, weak content, and few or no optimized backlinks.
HypothesisIf a generic page on a strong domain can rank without deliberately targeting a keyword, a purpose-built page may be able to outrank it.
1. Find a weak SERP
Begin with a seed keyword or niche. Look for forums, Reddit or other user-generated content, broad category pages, URLs with very few referring domains, pages missing the keyword from the title or H1, outdated or thin content, and weak intent alignment.
2. Find a weak broad competitor
Identify a domain with strong overall authority and broad visibility in the niche, but repeated page-level weaknesses. The opportunity appears when a strong domain ranks through weak individual pages.
3. Mine its organic rankings
Retrieve as many ranking keywords and pages as possible. For each result, collect:
- keyword, ranking URL, position, volume, estimated traffic, KD, and CPC;
- referring domains and backlinks to the exact URL;
- domain rating or authority, title, H1, URL, intent, and SERP competitors.
The competitor becomes the keyword database.
4. Detect accidental rankings
Find queries where the competitor ranks well even though the URL has few referring domains, poor keyword targeting, weak intent fit, and weak neighboring results. Bonus signals include Reddit, forums, or Quora near the top, old or thin pages, and several low-link URLs ranking together.
5. Expand into clusters
When one opening appears, find adjacent opportunities such as reviews, alternatives, X-versus-Y comparisons, best-of-category pages, category-by-use-case pages, related entities, long-tail variants, and more rankings from the same competitor. The goal is a coherent cluster rather than isolated keywords.
6. Create an explainable opportunity score
Score candidates from 0 to 100. Positive factors include meaningful traffic, commercial intent, a top-10 competitor result, few URL-level referring domains, poor title or H1 alignment, weak neighboring results, and a large related cluster. Negative factors include strong page-level links, dedicated optimized results, specialist sites, difficult intent, and limited organic click opportunity.
Do not rely only on keyword difficulty. The prompt's example makes every contribution visible:
- +20: URL has zero referring domains
- +15: Keyword is absent from the title
- +10: A forum ranks in the top five
- +12: The query has 2,400 monthly searches
- +15: The query has commercial intent
- +14: The research found twelve related opportunities
LLMs can help with clustering, intent, and classification, but the core score should remain deterministic and inspectable.
Required output
Return a table with Score, Keyword, Volume, Position, Domain, Ranking URL, URL Referring Domains, DR, Intent, Why Weak, and Cluster. Then provide the top opportunities, keyword clusters, and recommended pages.
For every recommended page, include the primary keyword, supporting keywords, title, URL slug, intent, competitor to beat, reason the SERP appears weak, and internal-link suggestions.
This is not merely another keyword-difficulty tool. It should find queries where authoritative domains rank with weak pages, then use those rankings to uncover clusters of unusually attractive opportunities.
A Shorter Approval-Gated Wrapper
Use this compact version when the connected agent can also create or publish content. It stops after research so a human can approve the evidence first:
Analyze [market/site] for strong domains ranking with weak individual pages. Use only connected live SEO data and label the source and capture date for every metric.
- Find domains with broad visibility in the niche.
- Return ranking URLs in positions
[range]where the page has weak intent fit, few relevant referring domains, generic or user-generated formatting, outdated coverage, or poor title or H1 targeting. - Cluster queries by shared intent. Do not create separate recommendations for close variants that belong on one page.
- Compare each cluster with
[my domain], including existing impressions, related pages, topical fit, and cannibalization risk. - Score each opportunity from 0 to 100. Show the positive factors, negative factors, missing data, and confidence separately.
- For the top five, propose the user problem, content format, original evidence required, internal links, conversion path, and a measurable success criterion.
- Do not write or publish content. Return a review table with source URLs and reasons a human can verify.
The last line matters. Research and production should be separate approvals. Otherwise an attractive but incorrect score can create dozens of low-value pages before anyone notices.
Build a Better Page, Not a Better-Optimized Copy
A purpose-built page does not win merely because the keyword appears in its title. Google's Search Essentials recommends helpful, reliable, people-first content and says that meeting technical requirements does not guarantee crawling, indexing, or serving.
For each approved opportunity, add something the weak result does not provide:
- a direct answer that matches the actual task;
- first-hand testing, screenshots, data, or worked examples;
- clear definitions and decision criteria;
- current sources with visible dates;
- a useful tool, template, checklist, or comparison;
- limitations, failure cases, and who should not use the advice;
- internal links that help readers continue the job.
Google's people-first guidance specifically asks whether a page adds original information, substantial value, and a satisfying answer. Its current AI-search guidance also warns against generating separate pages for every query variation primarily to manipulate rankings or generative responses.
How to Validate the Technique
- Before publishing: save the SERP, positions, competing URLs, backlink data, and your opportunity score.
- After discovery: confirm the canonical URL and indexing state. Do not confuse indexing with ranking.
- Weekly for four weeks: track page-level impressions, clicks, CTR, relevant queries, and conversions.
- At 30 days: compare the result with a matched page selected through conventional keyword research.
- Keep or revise: update the page when it earns impressions without clicks, ranks for the wrong intent, or fails to create a useful business action.
Google recommends focusing on trends in impressions and clicks more than position alone. Search Console's position is an average and changes with the selected aggregation, country, device, and query.
Companion Lesson: Being Mentioned in AI Search
The requested companion video, How to Get Mentioned in AI Search, was published by Vasco's SEO Tips on 23 September 2026. It extends the discussion from ranking pages to building sources and brand information that answer the prompts people use in AI-assisted search.
Video Chapters
| Time | Topic | Time | Topic |
|---|---|---|---|
| 00:00 | The technique | 07:07 | Opportunity scoring |
| 00:47 | Beyond keyword difficulty | 07:38 | Required data |
| 01:26 | Strong domains, weak pages | 08:52 | Turn research into action |
| 02:09 | Why weak rankings matter | 09:25 | Run it on your site |
| 03:25 | The complete process | 10:07 | Highest-leverage keywords |
| 04:04 | Prompt breakdown | 10:28 | AI agents for execution |
| 05:14 | What makes a page weak | 11:32 | Real data vs AI slop |
| 06:12 | Mine competitor rankings | 11:57 | Why the strategy can work |
| 06:33 | Accidental rankings | 13:38 | Why live data matters |
Sources
- Vasco's SEO Tips: the NEW AI SEO technique I use to Rank (#1 on Google in 24 hours) (video and supplied transcript)
- Arvow Chat official product page (current tools, Search Console boundary, chat history, credits, and autonomy limits)
- Arvow: official Blood in the Water SEO prompt
- Google Search Essentials
- Google: Creating Helpful, Reliable, People-First Content
- Google's Guide to Optimizing for Generative AI Features
- Google Search Console Performance Report
- Vasco's SEO Tips: How to Get Mentioned in AI Search
SERP composition, rankings, backlinks, product features, and pricing change. Recheck live data before acting on any opportunity score.