Entity Scent Audits: Manual Inverse Mapping of Competitor Topical Authority

You already know that backlinks, domain rating, and on-page tags are surface-level signals. The real edge in manual competitor analysis lies in understanding the entity scent—the invisible semantic pathway a search engine uses to resolve a searcher’s query against a competitor’s content cluster. This is not about keyword stuffing or TF-IDF scores. This is about reverse-engineering how Google’s Knowledge Graph connects a competitor’s pages to entities, and how that connection creates a topical authority halo that bleeds across their entire domain.

To do this without paid tools, you need to stop looking at individual pages and start dissecting the graph representation of their site. Start with a single competitor URL that ranks for a high-value term. Pull the page into a local HTML download. Yes, manually. Strip away CSS and JavaScript until you have a raw DOM tree. Now do a text scan for entity-bearing strings: proper nouns, dates, jargon that maps to a specific industry taxonomy, and especially co-occurring terms that appear within the same paragraph or list item. This is your entity co-location matrix. Write it down in a simple spreadsheet. Column A is the exact phrase. Column B is the entity type (Person, Organization, Product, Location, Concept). Column C is the number of times it appears within three sentences of your target keyword phrase.

The pattern you are hunting is the entity depth of the page. A thin page will mention one entity and its modifier. A thick topical authority page will mention five to eight distinct entities that share a direct predicate-argument relationship with your original search query. For example, if your target is “serverless data pipeline,” a weak page names Apache Kafka and moves on. A strong page names Kafka, links it to Apache Flink, mentions AWS Lambda triggers, references event sourcing patterns, cites Martin Kleppmann’s work on streaming, and ties all of this back to a specific use case like real-time fraud detection. Each of those entities is a node. The search engine scores the relevance of the page not just on keyword density, but on how many known, verified entity nodes are connected in a coherent subgraph.

You can validate this manually using Google’s own serp features. Search for the competitor’s title exclusively with the `site:` operator and watch what knowledge panels or entity cards render. If a competitor’s page triggers a Wikipedia infobox, a schema.org Product panel, or a “People also ask” block that lists three different specialized entities, you have proof the page speaks to a richer entity graph than yours. Now reverse engineer that by building your own entity list. Take each of those entities and search for them in Google’s free Natural Language API demo or even in a simple N-Gram analysis of your competitor’s visible text. Map out which entities appear at the top of the page versus the bottom, which ones appear in H2 tags, and critically, which ones appear in the internal anchor text of the page’s outbound links.

This is where the manual grind pays off. Open the competitor’s sitemap.xml (usually at `/sitemap.xml`). Download it. Run a fuzzy text comparison between their title tags and the entity names you extracted. If you see a title tag that contains an entity name from your list but not the primary keyword, that page is an entity hub. It exists purely to reinforce the semantic relationship between your target keyword and that entity. Google uses these hub pages as weighting mechanisms. You can see this effect without any paid tool by using the `related:` operator in Google or by running a simple search for `site:competitor.com intitle:“entity”`. The number of pages they have linking back to a single entity tells you how much they are subsidizing that concept.

Do not stop at text. Extract their image alt tags and meta descriptions. Often the most fragile content on a page—the alt text—contains early-stage entity references that they are testing for ranking. If you see an alt tag that says “asymmetric encryption key exchange” on a page about SSL certification, you know they are trying to weight that entity into the page’s relevance vector. Run those alt tags through Google’s reverse image search (free) and see which other entities cluster around that visual content.

Finally, examine the structured data with a raw JSON-LD viewer. Copy the competitor’s page source, extract the `application/ld+json` blocks, and parse them manually. Look for `@type: Article`, `@type: HowTo`, or `@type: FAQPage`. The entities listed inside `mentions` or `about` arrays are the ones Google considers most authoritative for that page. If you see three high-level entities there that you do not cover, you have found a gap. Write those entities down, build a page that ties them together in a stronger, more coherent subgraph, and you have manually reverse-engineered a competitor’s authority signal without spending a cent on rank trackers or crawlers.