Blog

Your Netlify site loads but Google shows nothing. Usually a preview URL's noindex, an SPA fallback faking 200s, or an empty HTML shell. Find yours fast.

Your GitHub Pages site loads but Google shows nothing. Usually Jekyll eating _next/, the 404.html SPA trick, or base-path 404s. Find your cause fast.

Your Bubble app loads but Google shows nothing. Usually an empty HTML shell, default metadata, or a version-test URL. Find your cause in 60 seconds.

Your Figma site loads fine but Google shows nothing. Usually the noindex setting, homepage canonicals, or a Make app with no title tag. Find yours fast.

Cloudflare can block Googlebot at five layers: custom rules, bot modes, Under Attack, rate limits. Find the one acting in Security Events and fix it fast.

Your Lovable site loads but Google shows nothing. Usually a client-rendered shell plus a missing sitemap and canonicals. Find your cause in 60 seconds.

Your Bolt site loads but Google shows nothing. Usually one hidden setting — SEO Boost — plus a few publishing slips. Find your exact cause in 60 seconds.

Your Vercel deploy works but Google shows nothing. 8 causes from the preview noindex header to an empty HTML shell, each with a 30-second check and fix.

Your Framer site loads fine but Google shows nothing. Framer's defaults are clean; the cause is a page noindex, a domain split, or patience. Check yours.

Your Base44 app loads fine but Google shows nothing. We fetched live Base44 apps the way Googlebot does and got a 1.8 KB empty shell. Find yours fast.

Your Replit app loads but Google shows nothing. Usually an empty HTML shell, a private deployment, or a dead replit.app URL. Find your cause fast.

Your site is indexed on Bing but not Google. Rule out a real block with URL Inspection in two minutes, then fix the version fold, robots rule, or lag.

Search Console says noindex is in your X-Robots-Tag header but your code is clean. Check the header and hunt the source: previews, plugins, config, CDN.

Your SEO tool reports 403 or 429 errors but the site loads fine. Usually your firewall blocked the tool, not Google. Two checks settle it in minutes.

Discovered - currently not indexed means Google found your URL but has not crawled it. It is a crawl-priority issue, not quality. Here is how to fix it.

Search Console says Couldn't fetch but your Next.js sitemap loads fine in the browser. Read the Last read date, test like a crawler, fix the real cause.

Search Console says Success but Discovered pages shows 0. Usually the count lags the last read, or your sitemap serves zero URLs. Two checks settle it.

Search Console says URL is unknown to Google. Google has never seen that exact URL. Check the variant you inspected, your sitemap, and internal links.

One page at position 8.7: 6,193 impressions, 7 clicks in 22 days. Its 200 disclosed queries all asked a yes/no question the results page answers first.

Four times my own pages competed for one query. Google kept the page it had read and the one that fit the results. See which page won, then merge or wait.

Neither www nor non-www ranks better. Duplicate live versions, temporary redirects, and SSL errors do the damage. Test all four variants in 30 seconds.

An XML sitemap is the list of pages you want Google to crawl. A missing, invalid, or client-side rendered one hides your site. Check yours in 60 seconds.

No, Google has no 60-character limit. Titles are cut by pixel width, and titles over 60 characters get rewritten 76% of the time. The 30-second check.

A canonical tag tells Google which URL is the real one. Point every page at the homepage and Google drops the rest. Test yours with one curl command.

A soft 404 is a missing page that returns 200 instead of 404, so Google crawls junk and drops real pages. Test yours with one curl command and fix it.

Your Next.js route sends Cache-Control: no-store and Vercel shows MISS on every request. What puts it there, what it costs, and why it won't de-index you.

Your link preview shows no image on X, LinkedIn, or Slack. Usually og:image is missing, relative, a dead URL, or the wrong format. Find and fix it.

A JS file Google indexed without crawling, 349 days of hidden CSS, a Bingbot block that worked, and why noindex and Disallow cannot share a URL. All dated.

The real story of game-scout's first Google clicks: not the homepage, not a trick — one specific niche page and the long-tail search that found it.

Add the TrafficTriage MCP server to Claude or Cursor and ask your assistant why your site isn't on Google: the 8 checks and one verdict, in chat.

The managed route, step by step: connect Google Search Console to Claude or Cursor with the TrafficTriage MCP server - 6 checks, read-only, in chat.

Connect an SEO MCP server to ChatGPT on the free plan: Developer mode, the New Plugin form, and the OAuth scope that breaks it when Claude works fine.

18 new lists, 17 stuck at Discovered - currently not indexed. Request Indexing did nothing; the link graph did. Every date, and the part I cannot prove.

Nine orphan lists, a link score that graded the wrong thing, and clicks up 1,055 to 2,147 in a month - the honest split of proven vs. too new to tell.

A real morning on a 67,000-impression site: the 7am Search Console report, the page that ranks and earns nothing, and how I pick what to write next.

Most SEO MCP servers are built for agencies with data plans. Here's which ones fit an indie hacker at zero, and the one job each of them does best.

74 green IndexNow runs, HTTP 200 each, and 105 of 161 sitemap URLs never announced to Bing. How the gap hid, the audit that found it, the one-call fix.

One site, one month: 45 auto-built lists earned 406 Google clicks, 34 articles earned 21. What supporting articles actually do, and how to check yours.

Our Search Console indexed count fell from 560 to 524. How we diffed impressions and inspected 436 URLs to name the losses, and why none needed a fix.

There is no bulk index-status API. URL Inspection takes one URL per call, 2,000 a day. What works instead, and the hourly sweep we built around it.

Search Console said crawled, currently not indexed on a page earning 4,993 impressions. It was wrong. How to tell a false verdict from a real eviction.

We edited two pages on the same day. One went quiet for nine days and came back stronger; one never recovered. Search Console daily data tells them apart.

We split 2,000 of one site's queries by what the answer demands. List-shaped queries clicked at 3.41%, yes/no questions at 0.13% at the same rank.

Our 25 'X vs Y' pages earned 4,732 impressions and 15 clicks in 90 days. Forums own comparison results; here is the two-minute SERP check to run first.

One page, ten internal links, and one click between all ten linking pages. Why links from pages Google never serves pass nothing, and how to audit yours.

Average position is computed correctly and still misleads: 74 to 94% of our pages' impressions sat on unnamed tail queries. Run this split before any fix.

Seven weeks of one-bet-per-cycle prioritization on a real site: numeric targets, HIT or MISS grades, and the diagnosis our own next grade disproved.

We built six hub pages, logged a prediction, and graded it. The hubs ranked at position 5 to 9.5 and returned 1 click. What the links did instead.

I tracked 165 pages on my own site for 21 days. One nav edit flagged 161 of them for two words. One alert was a page that lost 85% of its text.

I scored grouping by ranking page against four LLM runs on three sites. The free method hit f1 93.1% where gpt-4o managed 89.4% for five cents.

We configured 214 related links on game-scout.app and 22 never reached a page. What generated internal links actually render, and what moves when they do.

On game-scout.app, 107 curated lists earned 5,466 clicks in 90 days while 9,864 indexable catalogue pages earned 17. What that actually cost.

One tool returned 32,987 monthly for a head term and two long-tails, and 0 for the phrase our own Search Console measures most. What we use instead.

Search Console counted 2,619 clicks on game-scout.app in 28 days; GA4 logged 1,565 organic sessions. We split both by country and found the border.

Score, internal links, health checks and last-edit date all failed to separate 25 falling game-scout pages from the risers. One query pull did it.

Grouped by query, our 90-day Search Console rows held 9.5% of impressions and none of the 12 clicks. Grouped by page, the same window gave 102.9%.

On 10 game-scout list pages, 'best' queries sat at position 6.9 with 4.0% CTR; bare queries at 9.5 with 1.7%. Four live SERP pairs show what causes it.

Six guides at positions 4 to 6.6 drew 105 impressions in 28 days; a sibling at 6.6 drew 1,637. The impressions column names the ceiling, not position.

Our 28-day compare read impressions down 46%, position up 16 places. The daily series held two events: a harmless tail loss, then a real collapse.

My own page monitor listed 165 pages while the sitemap held 231. Its sitemap reader had failed every hour for 33 days and never logged an error.

Our device tab said mobile lost 651 clicks and desktop 40. Split by page, four of the five biggest losers lost impressions on both devices, not one.

Bing reported a better position than Google on all 28 pages both engines showed us, but its page report stops at position 10. How to compare them fairly.