Find your SEO competitors by checking which pages answer the searches you want to reach, under comparable market, language and device conditions. Use keyword-overlap reports to expand that list, then inspect the actual pages. A company can compete for your customers without appearing for those searches. A publisher or forum can compete for the same query without selling your product.
The useful output is a list of pages to improve, evidence to obtain and technical problems to investigate. The six steps below cover competitor discovery, keywords, content, backlinks, site structure and an action plan. Start with a defined product line or service rather than comparing every part of two unrelated websites.
1. Find competitors for a defined set of searches
Write down the audience and task first. For an industrial-pump supplier, finding a manufacturer, choosing a pump and diagnosing a fault are different tasks. The following examples are hypothetical query directions, not live rankings or a client study.
| Example query | Question to investigate | Useful observation |
|---|---|---|
| industrial pump manufacturer | Is the searcher looking for a manufacturer, distributor or product catalogue? | The supplier identity, offering and exact landing page |
| centrifugal pump selection | Does the result help with a process, calculation or model comparison? | Inputs, selection criteria and supporting documentation |
| pump cavitation symptoms | Does the page explain symptoms, diagnosis or a repair service? | Equipment scope, evidence and the next practical step |
For real searches, record the query, time, location settings, language, device, login state, result URL and page type. Note conditions you cannot control. An incognito window does not remove every source of variation. Keep ads, ordinary organic results, local listings, video and AI answers in separate fields. Google's position documentation explains why a one-off search may differ from Search Console's average position.
Classify each recurring site by two independent questions: does it compete for customers, and does it compete for these searches? A site can do both. A marketplace may be relevant to a product query but a poor model for your installation guide. Review the specific competing URL before treating the whole domain as a benchmark.
Tool discovery adds coverage. Semrush's competitor-discovery documentation describes an organic competitor report based on shared keywords in Google's top 20 organic results. For a new site with little ranking data, a target keyword list can provide a starting point for tracking. Shared keywords still need to be filtered for the products, markets and tasks you actually serve.
Use a short working list of close comparisons and separate single-topic references. Its size should follow the research scope; there is no universal number of competitors that makes an analysis valid. A large site's entire keyword portfolio is rarely the right comparison for one small product category.
2. Compare keyword gaps on matching terms
Choose the same country database, device, date and target scope for both sites. Compare domains with domains, relevant directories with directories, or equivalent pages with pages. A competitor's estimated global site traffic and your product page's Search Console clicks are different measurements.
- Build a starting query set from your own Search Console data, product documentation, sales questions and observed results. If your site is new, begin with the customer task.
- Export the keyword, ranking URL for each site, recorded position and the tool's volume and difficulty fields. Save the report settings alongside the export.
- Open the result pages. Group terms that need the same answer, while checking whether different models, uses or buying stages need distinct pages.
- Check your existing content before commissioning anything new. Distinguish a missing answer from an incomplete answer, the wrong page type or a gap in the tool's database.
In Semrush Keyword Gap, Missing means all the selected competitors rank while the site labelled “You” does not. Untapped means at least one competitor ranks while you do not. These labels describe the report and its filters; they do not prove that your page is unindexed or that a query has no demand. Historical data and comparisons at exact-URL, subdomain or subfolder level require the applicable subscription.
Ahrefs Organic Traffic estimates search clicks using ranking positions, search volumes and estimated CTRs. It is not a competitor's measured GA4 sessions. Its free-account documentation limits the relevant free Site Explorer and Site Audit access to websites with verified ownership. Check the target and report you need before choosing an Ahrefs plan.
Search Console describes the properties you can access. Its query documentation explains anonymized queries and row limits, so a missing query is not evidence of zero demand. Likewise, a low-volume term is not automatically an easy or high-converting opportunity. Use head and long-tail keyword distinctions to organize research, then turn the findings into a query-to-page plan. An existing page may need a better answer rather than a near-duplicate article.
3. Read the pages to identify a useful content advantage
A keyword spreadsheet tells you where to look. The page itself shows whether a reader can make a decision or complete a task. Replace “their article is longer” with a specific observation: “the guide names pump types but does not tell buyers which operating conditions to supply.” Verify the missing technical answer against relevant original documentation rather than treating the competitor as the final authority.
| Review area | What to record | Possible improvement |
|---|---|---|
| Title and opening | The task promised and the answer actually given | Narrow an overbroad title; state the answer and its conditions |
| Steps and comparisons | Inputs, decision criteria, alternatives and exceptions | Add usable definitions and explain what happens next |
| Evidence | Source version, model scope and genuine testing methods | Obtain original documentation; use cases only when records exist |
| Images and tables | Whether they explain a real distinction and work on mobile | Correct labels, units, headings or confusing diagrams |
| Next action | Whether the reader can find product details or enquiry requirements | Add an appropriate destination with a descriptive link |
Google's helpful-content questions emphasize originality, completeness, accuracy and reader benefit, and explicitly reject a preferred fixed word count. A practical article-writing process starts with the question and supporting material. Whether 800 words are enough depends on the answer, not the competitor's length.
Numbers, years and “tested” claims in titles must be supported by the content. Changing “Best” to “Top” adds no new information. Examine whether a page explains a decision, supplies a usable example or removes an uncertainty. Do not copy every heading simply because several ranked pages use it.
When reviewing updates, identify what changed: a price, product version, requirement or procedure. Changing a date alone is not a substantive update. Comments and forum discussions can reveal unanswered questions, but a few remarks do not represent all readers. You normally cannot read a competitor's actual engagement time or conversion rate from a public page; even your own engagement measurements need to be interpreted against the page's task and event definitions.
4. Inspect backlink sources and keep the measurements distinct
Using the same tool and reporting period, distinguish total links, referring domains, source pages and target pages. Open representative sources to confirm that the link still exists, its text fits the context and it reaches the intended destination. A tool's first-seen or lost date is not necessarily the actual publication or removal date.
Use the correct metric names. Ahrefs DR is domain-level, while UR is page-level. Site Audit also has a separate Page Rating, computed after crawl processing. These are tool metrics, not interchangeable labels or published Google scores.
Guangsuan's backlink combination approach emphasizes quantity, a mix of resource levels, layered construction and weight transfer, with different roles for GPB, GNB and GMB. The service page illustrates a first layer of 100 links, a second of 1,000 and a third of 10,000. That is Guangsuan's stated strategy, not a Google-mandated quota. A public backlink crawl alone also does not establish a competitor's complete multi-layer arrangement.
GPB assigns one link to one original article on one top-level independent domain, without repeated domains within the same package. Deduplication across separate purchases is agreed separately. Indexing, dofollow and non-deletion commitments are company delivery responsibilities, with verification times and remedies defined in the contract; they are not Google ranking guarantees. The service does not promise a specified DA or source-site traffic figure.
A source may be a directory, contribution, media reference or forum link. Record the actual format and eligibility conditions rather than assigning a fixed value multiplier to the label. Finding a competitor's source does not establish that you qualify to join it, or that it will publish your material. Keep candidate resources, required work and verified delivery in separate columns.
5. Compare public structure without inventing backend access
Screaming Frog's free version is limited to 500 URLs per crawl, with restrictions on advanced configuration, JavaScript rendering and API integrations. A scoped crawl can inspect responses, redirects, titles, canonicals and links that it can reach. A URL limit, robots restriction or missing rendering means the report covers only part of the site.
Sorting competitor pages by actual search clicks is not a standard public crawl feature. The official GSC integration instructions require a Google account with access to the relevant property. Without that access, use third-party estimates to select pages for inspection and label them as estimates.
- Crawl depth: count the link path from the chosen starting page, not the slashes in the URL. Check whether important pages are reachable from relevant categories and articles rather than imposing a universal three-level limit.
- Orphan pages: a links-only crawl cannot reliably find every page with no internal entry. For your own site, compare the crawl with CMS records, sitemaps and authorized search data. For a competitor, qualify the conclusion as “no entry found in this sample” when the inventory is incomplete.
- Internal links: inspect the link text, surrounding sentence and destination together. Google's link guidance recommends crawlable links and says there is no magical ideal number. Decide how many links an article needs from the next useful task.
- Canonical URLs: compare redirects, canonical declarations and sitemaps using Google's canonicalization guidance. A declaration is a signal, not proof that Google selected it.
URL Inspection requires the URL to be in the current property. It cannot inspect any competitor through your own property, and a public browser does not give you their server configuration. A successful live test does not guarantee indexing or a search appearance.
On mobile, complete a real task: find a category, read the specification table and use the relevant controls. A product-page review should follow the product-to-purchase path, while landing-page navigation decisions depend on how users reach and leave the page. The number of menu items is not a substitute for that exercise.
Keep laboratory testing separate from field experience, and compare tests under matching conditions. The current good Core Web Vitals targets are LCP at or below 2.5 seconds, INP at or below 200 milliseconds and CLS at or below 0.1, assessed at the 75th percentile with mobile and desktop segmented. Missing field data does not mean a pass. A single Lighthouse score or AMP adoption percentage cannot establish a fixed traffic advantage.
6. Turn findings into an action plan and measure consistently
Address barriers to completing the user task and accessing the page, then prioritize relevant content gaps for which you can obtain the necessary material. The table below is a working template, not a client result or a ranking-score formula.
| Observed gap | Action | Completion check |
|---|---|---|
| The competing guide explains selection inputs; yours lists product names | Obtain model scope, operating conditions and original documentation | A reader knows what to supply before selection |
| Several near-identical drafts target the same task | Review intent and existing URLs before choosing the main page | No duplicate publishing for wording variants; URL changes have a separate migration plan |
| A tutorial leaves readers without a product-information route | Add a contextual model or category link | The text describes a reachable, useful destination |
| The backlink report provides only a total | Add referring domains, source pages, layers and delivery status | Separate publication, accessibility, agreed indexing checks and search outcomes |
| The mobile task has a loading or interaction problem | Identify the cause and repeat the relevant test | Record deployment, lab results and field metrics separately |
Hypothetical calculation: suppose you check the first 10 ordinary organic results for the same 20 distinct queries in one round under matching conditions. Your site appears for four queries and a competitor for eight: sample coverage is 20% and 40%, a gap of 20 percentage points. If two queries are shared, the competitor-only set contains 8−2=6 queries, not 8−4=4. This is coverage of the selected query list, not traffic market share. Record failed, missing or unperformed checks separately.
CTR is clicks divided by impressions. In a separate teaching example, two complete, non-overlapping groups using the same aggregation method have 20 clicks from 400 impressions and 32 from 1,600. Combined CTR is 52/2,000=2.6%, not the simple average of 5% and 2%, which is 3.5%. Actual query exports may omit rows, so incomplete details cannot reconstruct the property total. Compare your own segments consistently rather than assigning competitors an assumed CTR.
Give each task an owner, target URL, change date and verification method. Correct deployment, search discovery, rankings and qualified enquiries are separate outcomes. Recheck the same query set while recording new searches separately, and investigate changes before attributing them to one edit. If you need a coordinated technical, content and backlink plan, the scope can be discussed against Guangsuan's SEO service and the site's actual starting position.
For a bilingual team, the Chinese version of this competitor-analysis workflow provides the corresponding report fields and example calculations. Agree on scope before combining the two teams' findings.