Intent Signals in the Noise: How to Audit Keyword Relevance with Entity Graphs

The days of treating keywords as isolated query strings are over. Any intermediate marketer who has survived a Google core update knows that ranking hinges on satisfying the searcher’s underlying need, not merely matching the lexical surface. Yet too many audits still stop at volume, difficulty, and a quick glance at the SERP snippet. To evaluate target keyword relevance and intent at a level that actually moves metrics, you need to decompose the query into its constituent intent signals—what I call the “intent triad”: search behavior, SERP feature composition, and entity co‑occurrence. Only then can you decide whether a keyword is genuinely relevant to your content or just a vanity match.

Start by accepting that exact‑match keyword placement is a legacy tactic. Modern search engines parse the latent semantic structure of a query. A phrase like “best running shoes” may look purely transactional, but the SERP often surfaces buyer’s guides, listicles, and comparison videos. The presence of a featured snippet or a “People Also Ask” block signals that Google recognizes a mix of informational and commercial intent. If you blindly target this keyword with a product page that lacks educational context, you are fighting the algorithm’s own intent classification. The smart approach is to build a small entity graph around your core keyword. Map the named entities—brands, models, materials, body parts, injuries—that co‑occur in top‑ranking content. The frequency and arrangement of those entities reveal the topic’s true search intent boundaries.

For example, take “plantar fasciitis shoes.“ An entity graph will show strong co‑occurrence with “arch support,“ “heel pain,“ “orthotic inserts,“ and “podiatrist recommended.“ The intent is predominantly informational and health‑related, despite the commercial overtone. A page that merely sells shoes without explaining the biomechanics of plantar fasciitis fails the relevance test. Google’s passage‑ranking algorithms and BERT‑based semantic understanding will demote it in favor of content that addresses the condition first and the product second. By auditing your target keywords through this entity lens, you stop wasting effort on high‑volume terms where your content’s relevance score is inherently low.

Now layer in search behavior signals. You don’t need a premium tool to infer intent; you can proxy it with click‑through rate patterns from your own Search Console data. Keywords with a high impression volume but low CTR often indicate a mismatch between the searcher’s intent and your page’s promise. If your page title promises a “complete guide” but the snippet shows a price‑list, users bounce. More subtly, a keyword may have a high CTR but poor dwell time and high bounce rate. That signals that you captured the click via a compelling title but failed to deliver the exact intent the user expected. When performing a relevance audit, segment your keyword set by these behavioral metrics. A keyword that drives traffic but kills session duration is less relevant than a lower‑volume term that keeps users reading to the bottom of the page.

SERP feature analysis is the third pillar. Google has become remarkably consistent in how it formats results for different intents. “How to” queries almost always spawn video carousels and step‑by‑step guides. “Buy vs. rent” queries trigger comparison tables and review aggregators. “Symptoms of” queries generate health condition panels. When evaluating whether a keyword is relevant to your site, ask: does the dominant SERP format align with what your content can actually deliver? If the SERP is saturated with YouTube videos and your site has zero video content, the keyword is likely not a good fit regardless of relevance. Conversely, if the SERP shows mostly text‑based listicles and your site excels at that format, the intent relevance is high—even if the keyword has a lower volume than a flashier alternative.

One nuance often missed by intermediate marketers is that intent is not static. A keyword like “AI writing tools” a year ago leaned commercial; today it leans informational as the technology matures. An audit that doesn’t track intent drift over time will systematically misallocate resources. Build a quarterly review process where you re‑evaluate the top 20% of your target keywords by re‑running the entity graph, checking SERP feature changes, and refreshing your own click‑behavior data. This is not busywork—it’s the difference between content that remains relevant and content that slowly rots in the index.

Finally, relevance must be measured against your site’s topical authority. A keyword may fall precisely within the informational/commercial hybrid intent you serve, but if your site lacks the supporting entity signals to be viewed as a subject matter expert by Google, you will not rank. This is where the concept of “intent eligibility” comes in. Before adding a keyword to your content plan, score it on three dimensions: its own intent purity, the structural feasibility of your content to match the SERP format, and your site’s existing entity coverage for the broader topic. Only keywords that pass all three deserve a spot in your editorial calendar.

The takeaway is simple: relevance is not a binary state but a probability derived from multiple latent signals. Stop treating keyword intent as a label you assign at the top of a spreadsheet. Start treating it as a continuous analysis of entities, behaviors, and SERP patterns. Your content—and your rankings—will reflect the difference.