SERP Competitor Teardown: How to Reverse-Engineer the #1 Page

Most SEO teams open the page that outranks them, count the words, copy the headings into a brief, and call it competitor research. They ship a longer version of the same thing and watch it not rank. The problem is not effort. The page they studied and the page that is actually ranking are two different things. The text is what the competitor wrote. The ranking is what Google is rewarding across a specific set of signals, and those signals are almost never visible from the URL itself.

A real SERP competitor teardown is forensic. You are diagnosing why a page is winning, isolating which signals matter for this exact query, and writing a counter-brief your team can ship. Single-query, structured, produces a fight plan, not a slide deck. This is the workflow we run for every Tier-1 query on a client roadmap, the same workflow that sits inside a wider 8-area SEO audit when the engagement scales beyond one keyword.

If you can run this against one of your own target queries by the end of the day, the post worked.

Why most SERP teardowns are just vibes (and what a real one looks like)

The lazy teardown: open the URL ranking one in incognito, note it has "around 2,400 words and 11 H2s," paste it into Surfer, brief a longer version, ship 3,100 words, watch nothing happen. Three months later the post sits at position 18 and the team blames link velocity.

The forensic teardown starts from a different assumption: the page you are looking at and the signals it is ranking on are distinct. A page can rank because of one Wikipedia citation. Because the domain published on the topic first, five years ago. Because its parent cluster has 40 internal links from authority URLs on the same site. The text is often the smallest factor. If your teardown only describes the text, you are matching the wrong layer.

The output is a counter-brief, not a content audit. It specifies the title format, the opening answer block, the entities to name, the schema to ship, the internal links in and out of the page, and the one differentiator the competitor cannot cheaply replicate. Nothing more.

Three pre-conditions. Confirm intent match: a "how to choose a CRM" guide will not beat product comparison pages no matter how well it is written. Confirm SERP stability over 90 days: if position one has churned through four URLs in a quarter, the algorithm is unsure and your timing is wrong. Confirm you can plausibly win in two quarters given your domain authority. If not, route the budget elsewhere.

Step 1: Lock the query and read the SERP, not the page

Open incognito, set location to the target market with a VPN, set device to what your buyers actually use. Pull the live SERP from this session, not your rank tracker, not your office IP. Google personalises and your tools cache.

Inventory every SERP feature above the organic results: AI Overview, People Also Ask, video carousel, image pack, Top Stories, sitelinks, knowledge panel. Each steals real estate and shifts the brief. The AI Overview tells you the structural format AI engines want. A video carousel means text alone will not capture every click. PAA hands you free H3s. A knowledge panel means a brand already owns the entity.

Classify dominant intent across the top ten: informational, commercial, transactional, navigational. Note any intent split. Half guides and half product pages means mixed intent, which is much harder because you have to satisfy both reader types from a single URL. Mixed-intent SERPs go to the page that resolves the conflict best.

Flag the content type pattern. Eight listicles in the top ten means "ultimate guide" pages will not break in. Six comparison pages means a definitional explainer is not going to win. The pattern is the price of entry.

If Reddit, Quora, or YouTube hold more than two spots, the playbook changes entirely. The engine wants peer-to-peer or video answers. The play is participation, not a page. We cover that in the how to rank on Perplexity breakdown and the Reddit SEO playbook, because both engines weight community-led answers higher than blog content for these query patterns.

Step 2: Profile the #1 page across nine signals

The heart of the teardown. Nine signals, scored independently, no narrative until all nine are filled in.

Topical depth. Count distinct sub-topics, not word-count. A 1,400-word page covering 12 sub-topics out-ranks a 4,000-word page covering four.

Entity coverage. Extract every named entity: tools, frameworks, people, products, methodologies. Check each for a Wikidata or Knowledge Graph entry. A competitor naming 15 Knowledge Graph entities beats your page that mentions four unnamed "tools."

Structural clarity. H2/H3 hierarchy, definition blocks, comparison tables, ordered step lists, FAQ blocks. AI engines lift these almost verbatim. Five clean definition blocks get cited more than thirty paragraphs of running prose.

Internal link inflow. Run Screaming Frog. Export the Inlinks report for the target URL. A page with 28 internal links from pages two clicks deep with descriptive anchors is in a different position than one with three footer links. The single most under-discussed ranking factor on most sites.

Backlink quality. Topical relevance, anchor distribution, dofollow ratio. 40 topically relevant editorial mentions beat 200 directory listings. Volume is a distraction.

Schema in source. View source. Look for Article with full Person markup including sameAs to LinkedIn and X, FAQPage, HowTo if steps exist, BreadcrumbList. Missing schema is a free win; most competitors ship only Article.

Freshness signals. Last-modified byline. Dated statistics. Year stamps in the title. Changelog blocks. Engines weight freshness heavily on more queries than teams realise.

E-E-A-T artefacts. Named author with credentials. Citation density. Original screenshots, original data. A "reviewed by" block from a recognised expert. Two named experts and an original chart hold off a page with none, even if the latter has more words.

Page experience. Run PageSpeed Insights. LCP under 2.5 seconds, INP under 200 ms, CLS under 0.1. A weak page experience score is a structural disadvantage ad copy will not overcome. If you are weak here, the work belongs to a technical SEO programme before any content sprint.

Step 3: Tools and the exact reports to pull

The kit is small. The skill is in reading the outputs, not running them.

Ahrefs or Semrush. Pull the organic keywords report for the competitor URL to see every query it ranks for and the traffic each drives. The sub-topics that earn traffic are usually a much smaller list than the sub-topics the page covers. Pull referring domains with DR and topical relevance columns. Export both as CSV.

Screaming Frog. Crawl the competitor site, then export the Inlinks report filtered to the target URL. The most diagnostic report in the kit, and the one most teams skip.

Google Search Console for your own site. Pull current position, impressions, and clicks for the target query. You need your starting line. Position 12 with 4,000 monthly impressions is a different game from position 89 with 30 impressions.

Surfer or Clearscope. Sanity check on entity and topical coverage. The output is raw material, not the brief.

View-source plus the Schema Markup Validator. Catches JSON-LD blocks that crawler tools sometimes miss when scripts inject them via tag managers.

PageSpeed Insights and Core Web Vitals in GSC. The page experience read takes ten minutes. Remediation takes longer and usually sits inside a wider engineering programme.

DataForSEO SERP API. Useful for historical SERP snapshots to confirm rank stability. The cheapest credible signal that a SERP is settled enough to attack is 90 days of position-one stability for the current URL.

Step 4: Score every signal as 'price of entry' or 'differentiator'

Every signal falls into one of two buckets, and the bucket determines how you respond.

Price of entry. Signals every page in the top five has: schema, dated freshness, table of contents, named author, mobile-fast rendering. Match these or you are disqualified. Matching only lets you compete.

Differentiators. Signals only number one and maybe number two have: original data, embedded calculator, video, named-expert quotes, deep internal link inflow. These decide who wins. These are what you attack.

Score each signal zero to two. Zero means the page lacks it. One means baseline. Two means strong. Score number one first, then your draft. Compare row by row. Anything where number one scores two and you score zero is your shortlist, ranked by effort-to-win. A schema gap takes a developer one hour. An internal-link-inflow gap takes a month of content publishing. An original-data gap takes a research sprint. Pick the gaps you can close in 90 days. Ignore signals where all ten pages score two; you cannot out-do the floor.

One caveat that sinks many teardowns: before drafting the counter-brief, check your own site for duplicate pages on the same intent. If three of your pages are fighting each other for the query, a fourth one will lose. We cover that in the keyword cannibalization audit playbook. Fix it first.

Step 5: Run the SERP-feature audit (the part most teams skip)

The SERP features above and beside the organic ten are where most missed leverage hides.

AI Overview. Note which sources are cited, in what order, with what passage. The cited passage is the structural format AI engines want for this query. If the AI Overview opens with a one-sentence definition then three bullets then a comparison, that is your intro contract. Match it and you become a candidate citation. This is the heart of Answer Engine Optimization.

People Also Ask. Every PAA question is a free H3 and a free FAQ schema entry. Click each to expand, then click the new questions that appear. Harvest the full set; use 12 and your page covers query variations the competitor missed.

Video carousel. Text-only pages give up the slot. The brief needs an embedded explainer video, even a simple two-minute talking-head version.

Image pack. Rename image files with descriptive kebab-case names matching keyword variations. Caption with alt text mirroring PAA questions.

Top Stories. Publisher-only slot. If your domain is not a registered News publisher in GSC, note as permanently capped and move on.

Sitelinks under number one. The engine is rewarding strong internal architecture on the competitor's site. The cluster matters more than the page. Replicate the cluster structure, not just the URL. This is a pillar-and-cluster content programme decision, and the teardown is your evidence the cluster work is required.

Step 6: Backlink delta - find the links you can actually replicate

You do not need to recreate the competitor's link profile. You need the subset of their links you can earn in 90 days.

Export referring domains for the target URL. Filter aggressively: drop nofollow, syndication networks, scraper mirrors, forum threads with no engagement, paid directories. What remains is the clean set: editorial mentions, partner blogs, journalist citations, industry roundups, podcast show notes, the occasional Reddit thread that earned real engagement.

Bucket the clean list into earnable archetypes: editorial roundups, journalist mentions, quality directories, partner blogs, Reddit citations, podcast show notes. Match each bucket to your execution capacity. With a PR team, journalist mentions are winnable in a quarter. With content only, editorial roundups and partner blogs are your zone. Pick two or three buckets and ignore the rest.

Execution splits into two motions: white-hat editorial link building for editorial roundups and partner blogs, and digital PR for earned editorial coverage for journalist mentions and earned features. Different prospects, different angles, different timelines. Running both through one team usually produces neither.

Build the outreach list as a CSV: domain, contact, the angle the competitor was linked with, your differentiated angle, priority score. Hand it to the execution team. Skip one-off vanity links; chase boring repeatable links instead. Five editorial mentions per quarter from mid-tier publishers outperform one viral mention three years ago.

Step 7: Translate the teardown into a counter-brief

The counter-brief is the deliverable. Keep it tight.

Headline format. Must beat the SERP pattern, not just the number one title. If the SERP is listicles and number one is "11 Ways to X," ship "17 Ways to X" with a stronger sub-headline. Match the pattern, beat the instance.

Intro contract. A 60 to 100 word answer block in the first viewport, matching the structural format the AI Overview uses. Definition plus three bullets, definition plus three bullets.

Required entities. A pasted list of named entities the writer must mention, with internal link targets or outbound citation URLs against each. The teardown produced the list.

Required schema. Exact JSON-LD: Article with full Person markup including sameAs to LinkedIn and X, FAQPage with the harvested PAA questions, HowTo if there is a numbered process, BreadcrumbList. No back-and-forth.

Required internal links. Five to eight anchor-and-URL pairs feeding the new page from existing high-authority URLs. Eight to twelve outbound internal links from the new page. If you only spec outbound, you ship a page with no inflow and it dies.

Differentiator block. The one thing this page does that the number one cannot replicate cheaply. Original dataset. Named-expert quote from someone the engine recognises. Interactive calculator. Downloadable template. Fresh case study with real numbers. Without it, you are matching, not winning.

A counter-brief in this format runs three to five pages. No prose suggestions, only specifications. The teardown is over.

Step 8: Ship, then watch the right signals (not rankings)

Rankings lag. Watch only rankings and you will misdiagnose every post-ship outcome and retire pages that were quietly winning. Watch leading indicators instead, in three windows.

First 14 days. Indexation in GSC via URL Inspection. Crawl frequency in server logs or GSC Crawl Stats. AI Overview presence on the target query, checked manually. These move before rankings do. Clean indexation and rising crawl frequency mean the page is in the conversation, even if it has not surfaced yet.

Days 15 to 45. SERP position in your rank tracker. CTR versus SERP average in GSC. Scroll depth in GA4. Internal-link inflow growth as you backfill links from related pages.

Days 45 to 90. Referring domain growth from your outreach CSV. AI citation rate on ChatGPT and Perplexity. Conversion rate from the page if it has a downstream goal, which it should.

If position has not moved by day 60 and indexation is clean, the diagnosis was wrong. Do not edit the content first. Re-score the matrix. Find the signal you misread. Change the smallest thing that changes the score, ship, and reset the 60-day clock.

If the existing page on the topic was actually closer to ranking than the new one, the right move is a refresh, not a relaunch. We cover that in the content decay refresh framework, the cheapest source of recoverable organic traffic on most sites.

Document each teardown in a living doc, one page per query. The same workflow runs against every Tier-1 query on the roadmap, not as a one-off. After ten teardowns you will start seeing patterns in your vertical: which signals matter most, which link buckets are reliably winnable, which differentiators your team can ship fastest. The workflow gets faster as you run it more.

When the teardown says don't bother (and where to invest instead)

Sometimes the most valuable output of a teardown is the decision not to attack the query. Five disqualifying patterns.

Number one is a brand SERP for a competitor's branded query. Walk away. That SERP belongs to them by definition. Pick an adjacent generic query.

Top ten is mostly Reddit, YouTube, and forums. The engine wants community and video, not brand content. The play is participation: build a Reddit presence, ship a YouTube video.

An AI Overview answers the query completely with no click-through. Re-score as awareness-only. The metric shifts to AI citation rate. If awareness is not commercially valuable for this query, find a higher-intent variant.

Number one has 50 times your referring domains on a high-DR-gated query. Pick an adjacent long-tail and ladder up via a topical cluster. Win the cluster first, build topical authority over six to twelve months, attack the head term from a position of relevance. Head-on attacks when you are 50x under-linked are the most expensive failure mode in SEO.

Mixed intent the page cannot resolve. Four guides, four listicles, two product pages in the top ten. The engine is unsure. These SERPs churn. Higher work, uncertain timing.

When a teardown is genuinely ambiguous, do not spend three weeks losing politely. Route the keyword to a 60-minute teardown with a senior SEO consultant who has run this workflow against hundreds of queries. An hour of senior judgement always costs less than a wrong-fight sprint. When single-query attacks are not enough to move the programme, the right next step is a wider revenue-driven SEO services engagement, where the teardown becomes one input into a roadmap rather than the whole game.

What to do next

A single teardown takes a day. The decision to run one takes a meeting. The skill to read the matrix correctly takes ten teardowns. Pick a query from your roadmap where the position is stuck between 8 and 18, run the workflow end to end this week, and ship the counter-brief by Friday. That one exercise will tell you more about why your content is or is not ranking than three months of generic competitor reports.

If you want a teardown run against one of your priority queries by a team that has done this hundreds of times, book a 30-minute conversation and we will scope it. Bring the query, the URL you would attack it with, and your current position. We will hand back the counter-brief.