My Google clicks are dropping. Is SEO dead?

No. Clicks are shrinking because Google and ChatGPT increasingly answer on the page, so people click less even when you rank well. That's not the same as being invisible. The earlier signal is whether AI is reading your pages at all: HadoSEO shows every one of those reads, which pages, and why.

What does it mean when an AI 'reads' my page?

One of three things: it's learning about you (gathering pages for its model), filing you for search (storing your page so it can show up in results), or answering someone's question about you right now (reading the page live to build an answer). HadoSEO tells you which, per visit.

Why isn't my Lovable or Base44 site showing up on Google?

Your app renders its content with JavaScript in the browser. Most search and AI crawlers request the page and read the raw HTML, which is an almost-empty shell with a and a script tag. They see no content, so there's nothing to index. HadoSEO serves those crawlers a fully rendered version of the same page instead.

Will Google penalize prerendering as cloaking?

No. Cloaking is showing bots different content than users. Prerendering serves bots the same content your users see, just rendered ahead of time. Google calls this a workaround and prefers sites that build pages on the server long-term. Content parity is the rule, and HadoSEO serves identical content to bots and humans.

Which setup should I use, DNS or Cloudflare Worker?

If you're on Lovable, Base44, Bolt or Replit, or can't change server code, use DNS: we render and log at the edge, from any registrar. If your domain is on Cloudflare (any framework), use the small Worker we provide, with no cutover and no downtime. Crawl analytics is the same on both.

My site is already server-rendered. Why would I need HadoSEO?

Rendering makes your pages readable. It doesn't tell you who's reading them. The Cloudflare Worker runs in passthrough mode: your site keeps serving everyone while every crawl is logged by page and by bot, with origin-captured snapshots so you can still see exactly what each bot read.

Do I need a developer?

For the DNS path: no. If you can edit a domain setting, you can do this. For the Cloudflare Worker: yes, a little. Someone deploys a small provided snippet on your Cloudflare account, and your app itself is untouched.

How do you know a bot is real?

We check each request against the IP ranges and reverse-DNS records the operator publishes: GPTBot, ClaudeBot, PerplexityBot, Googlebot, Google-Extended, Bingbot and more. Requests that use a bot's name from someone else's IP are flagged as unverified.

[ logged · verified · rendered when you need it ]

See every page AI crawlers read.

ClaudeBot
reads.
Then make sure it can.

See exactly when ChatGPT, Google and other bots read your site, and which pages they care about. Set up in minutes, with no code.

Run free scan Start free trial

verified botsno samplingDNS or Cloudflare Worker

access.log

[ your access logs, already parsed ]

timebotpathresult

15:52:10ChatGPT-User/pricing200answer
15:52:02Timpibot/product/meta-tags200
15:51:58ChatGPT-User/product/prerendering200answer
15:51:52ClaudeBot/security200training
15:51:45Googlebot/docs/api200index
15:51:40meta-externalagent/blog/llms-txt200training
15:51:32OAI-SearchBot/200index

built for

Lovable

Base44Bolt.newReplitFloot

Next.js

Nuxt

Astro

React

Angular

Trusted by 1,500+ sites

[ the shift ]

Rankings holding steady. Clicks falling. Is SEO still working?

Google and ChatGPT now answer questions right on the page, so people click less, even when you rank well. That doesn't mean your SEO stopped working. It means clicks stopped being the whole story.

AI reads are the new impressions: proof that AI is looking at your site, and which pages it cares about.

34.5%

fewer clicks for the #1 result when an AI Overview appears

Ahrefs · Apr 2025 →

68%

of US Google searches end without a click

SparkToro · 2026 →

8%

of people click a link when an AI summary shows vs. 15% without

Pew Research · Jul 2025 →

[ the old scorecard ]

rankings ✓→clicks ↓→traffic ↓

Breaks when AI answers first.

[ what hadoseo adds ]

AI reads ↑→likely cited ↑→visits from AI ↑

Shows your work landing before the clicks do.

[ one page · 90 days · weekly ]indexed to 100 at week 1

Google clicks -12%AI reads +140%visits from AI +50%

Same page. Clicks alone say "nothing's happening." AI reads say otherwise. Illustrative sample.

[ crawl analytics ]

Every request. By page, by bot.

Pick any page and see who read it. Open any bot to see what it read, and any version to see exactly what it was served.

app.hadoseo.com/acme-app.com/pages/pricing

acme-app.com▾

Pages ›

/pricing

Last 30 days ▾

CrawlsTotal crawls

760

Last crawlLast crawled

2h ago

CachedServed from cache

100%

532 verified104 declared124 unverified

Crawls by bot

GooglebotGPTBot

Last crawl by bot

Googlebotverified2h ago

GPTBotverified5h ago

ClaudeBotverified1d ago

PerplexityBotverified2d ago

"Slackbot"unverified6d ago

✓Every visit logged as it happens. Real bots checked against published IP ranges.

[ client-side apps ]

Your app works. Crawlers see a blank page.

Most AI-built apps render in the browser. Search and AI crawlers don't run that code. They see a blank page and move on.

view-source:acme-app.com · via HadoSEO

1+<html>

2+ <head>

3+ <title>Acme · know why customers leave</title>

4+ <meta name="description" content="…">

5+ <meta property="og:image" content="…">

6+ </head>

7+ <body>

8+ <h1>Churn analytics for SaaS</h1>

9+ <p>Acme watches every account…</p>

10+ <a href="/pricing">Pricing</a> <a href="/docs">Docs</a>

11+ <div id="root"></div>

12+ </body>

13+</html>

✓Full page, title and meta, served to every crawler.

Built with Base44, Lovable or React? See the four signs →

The fix is not a rebuild. It's one DNS record. See the DNS path ↓

01 · search console

search.google.com

Page indexing38 pages

/pricingIndexed

/features/alertsIndexed

/blog/churn-signalsIndexed

✓Pages leave the queue

02 · google

google.com

site:acme-app.com

acme-app.com › pricing

Pricing · Acme

acme-app.com › blog

5 churn signals to watch · Acme

✓Every page can be found

03 · slack

app.slack.com

You 10:42 AM launched: acme-app.com

acme-app.com

Acme · know why customers leave

Spot churn risk early.

✓Your title, your image

Run free scan How our prerendering service works →

[ the funnel ]

From crawl to customer, measured end to end.

[ HadoSEO does the work ]

12,840crawler reads

99% served

12,690served in full

46% of reads

[ proof it paid off ]

5,906answer fetches

14% likely cited

824likely cited

23% clicked

189visits from AI

Funnel, last 30 days (illustrative sample)
stage	count	conversion
crawler reads	12,840	—
served in full	12,690	99% served
answer fetches	5,906	46% of reads
likely cited	824	14% likely cited
visits from AI	189	23% clicked

Sample site, one month. Bar heights are compressed so small numbers stay visible.→ 189 people from AI answers this month

12,840

crawler reads

12,690

served in full · 99%

5,906

answer fetches · 46% of reads

824

likely cited · 14%

189

visits from AI · 23%

[ platform ]

One request path. The whole crawl surface.

[ ai answers ]

Which AI answers with your pages

Live fetches while an assistant writes a reply, and how many likely made it into the answer.

5,906answer fetches · 30d824likely cited14%likely cited189visits from AI

engine answer fetches likely citedfetched · likely cited

ChatGPT1,840 indexed2,410 · 386 16%2,410 · 386 16%

Perplexity610 indexed1,520 · 228 15%1,520 · 228 15%

Claude480 indexed1,090 · 131 12%1,090 · 131 12%

Gemini920 indexed640 · 51 8%640 · 51 8%

Illustrative sample · your dashboard counts real requests+ 4 more engines →+ 3 more engines →

[ ai answers ]

Which AI answers with your pages

Live fetches while an assistant writes a reply, and how many likely made it into the answer.

5,906answer fetches · 30d824likely cited14%likely cited189visits from AI

engine answer fetches likely citedfetched · likely cited

ChatGPT1,840 indexed2,410 · 386 16%2,410 · 386 16%

Perplexity610 indexed1,520 · 228 15%1,520 · 228 15%

Claude480 indexed1,090 · 131 12%1,090 · 131 12%

Gemini920 indexed640 · 51 8%640 · 51 8%

Illustrative sample · your dashboard counts real requests+ 4 more engines →+ 3 more engines →

[ prerendering ]

Prerendering

Fully rendered HTML for every bot. Automatic on DNS, optional on the Cloudflare Worker.

GPTBotrendered · 38ms

Googlebotcache hit · 12ms

per request: rendered, cached or origin

[ attribution ]

Real bot or impostor?

Checked against each company's published IP ranges and reverse DNS.

verified84.2%10,811
declared11.5%1,477
spoofed4.3%552 flagged

[ intent ]

Why AI is reading you

Learning about you, filing you for search, or answering a question right now.

answer fetch46%
search index38%
training16%

[ rank ]

Rank & AI visibility

Google position and AI answer traffic, side by side.

pos 4 ▲2● likely cited 41 ▲9

[ routing ]

Routing rulesDNS

Send old links to their new pages, properly. No code changes.

/old-pricing301 → /pricing

/docs/*proxy → docs host

12 rules active

[ agents ]

MCP server

Every chart on this page, readable by your agent. 19 tools.

get_crawl_summaryread

create_routing_rulewrite

See it work ↓

[ mcp server ]

new · beta

Ask your agent why the crawlers left. Let it fix the cause.

Connect Claude, Cursor or any MCP client with one API key. Your agent reads the same crawl data you see in the dashboard, finds what broke, and fixes redirects and caching when you say yes.

read

Crawl analytics

By page, by bot, raw logs, visits from AI

diagnose

What changed

Drops, 404s, spoofed bots, cache misses

fix · with approval

Live config

Redirects, cache rules, blocked paths, purge

claude mcp add --transport http hadoseo \
  https://api.hadoseo.com/functions/v1/mcp \
  --header "Authorization: Bearer hado_sk_…"

Settings → Connectors → Add custom connector
URL     https://api.hadoseo.com/functions/v1/mcp
Header  Authorization: Bearer hado_sk_…

// ~/.cursor/mcp.json
{ "mcpServers": { "hadoseo": {
  "url": "https://api.hadoseo.com/functions/v1/mcp",
  "headers": { "Authorization": "Bearer hado_sk_…" }
} } }

Transport  Streamable HTTP
URL        https://api.hadoseo.com/functions/v1/mcp
Header     Authorization: Bearer hado_sk_…

Replace hado_sk_… with a key from Settings → API Keys.

Get your API key Setup guide · 19 tools

every plan
read-only keys
one key per agent
revoke instantly

agent · hadoseo connected

● 19 tools

Why did GPTBot stop crawling us last week?

get_crawl_summary · 30dget_page_analytics · /docs/*

GPTBot reads fell 38% after Oct 20. Since the docs move, /docs/* returns 404, so it stopped coming back. Six moved pages need redirects:

/docs/setup301 → /guides/setup

/docs/api301 → /guides/api

+ 4 moreGPTBot · Googlebot · ClaudeBot hit these

Apply these 6 redirects?ApplyNot now

Yes, then refresh the cache.

create_routing_rule ×6 · writepurge_cache · write

✓ Six redirects are live and a fresh render is queued. I'll check GPTBot's next reads of /guides/* if you ask tomorrow.

Illustrative sample · blue = read, amber = write

[ setup ]

Two ways in. Same crawl data.

Pick by stack. Both paths log every request; the difference is where your traffic flows and whether we render for you.

Most client-side setups finish in about 5 minutes. If you can edit a domain setting, you can do this.

path a · dns record

your domain provider · DNS

DNS recordsyourdomain.com

TypeNameValueTTL

MX@mail.protection.example1h

A@137.66.32.95Auto

TXT@v=spf1 include:example …1h

✓Saved. Crawlers start getting full pages.

path b · cloudflare worker

dash.cloudflare.com

Workers Routesyourdomain.com

RouteWorkerStatus

yourdomain.com/*hadoseo-workerActive

Added to the zone once. No cutover, no downtime; visitors pass straight through.

✓Route active. Every request logged.

[ + ]compare the two paths

feature	DNS	Cloudflare Worker
best for	DNSClient-side apps: Lovable, Base44, React, Angular	Cloudflare WorkerServer-rendered sites, or client-side apps you run on Cloudflare
setup	DNSAn A record at your domain provider. No code, no redeploy. About 5 minutes.	Cloudflare WorkerA small Worker on your Cloudflare zone, with your API key. About 10 minutes.
you need	DNSAccess to your domain's DNS settings	Cloudflare WorkerYour domain on Cloudflare
traffic flows	DNSThrough HadoSEO's edge	Cloudflare WorkerThrough your own Cloudflare zone
crawl analytics	DNS✓ every request	Cloudflare Worker✓ every request
prerendering for bots	DNS✓ automatic	Cloudflare WorkerYour choice: prerendered pages, or analytics-only

[ what you'll know ]

Questions you can answer without a developer.

In the dashboard, or by asking your agent.

Is ChatGPT actually reading our site?

See every visit from ChatGPT, Claude, Perplexity and Google, by page.

ChatGPT · read /pricing · 2h ago

Did our new blog post get picked up?

See which AI read it, and when, within hours of publishing.

/blog/new-postChatGPT ✓ 3h after publish

Which pages does AI care about most?

Ranked by how often AI reads them to answer questions.

Which AI answers send us visitors?

Pages AI likely cited, and the people who clicked through.

41 likely cited · 12 visitors

Is our traffic drop an SEO problem?

Compare Google clicks with AI reads on the same page.

Google clicks → flatAI reads ↑ rising

Is someone scraping us?

Flagged when a scraper pretends to be a real AI.

"ClaudeBot" · not the real one

Founders

Built your app on Lovable or Base44? Make it readable to Google and ChatGPT, and see the proof. No developer needed.

Get found →

Marketers

Show your boss SEO is working even when clicks dip. See which pages AI reads, fetches to answer questions and sends visitors from.

See the proof →

Agencies

Give clients proof of AI visibility, page by page, for every site you manage.

For agencies →

[ from the community ]

Founders shipped. Then they got found.

“After a few edits guided by Kento, we were able to re-index our website within an hour! Excited to see the SEO results rolling in.”

Samed

Owner · Talya Roofing

“My Lovable website only indexed the home page, not my blogs. HadoSEO solved it instantly. Really happy with the price and service.”

Justin

Agency Owner · Nieuwblik

“My clients love Lovable but their sites never ranked on Google. Hado SEO fixes that issue easily. Now I can use it for real projects.”

Tijs

Agency Owner · Lerai

“A professional, well-thought tool that brings real SEO improvements to Lovable projects. Efficient setup and amazing results. Highly recommended.”

Jimmy

Lovable Expert · Upwork

“Hado SEO took our Lovable site from invisible to discoverable fast. Easy setup, clear fixes, great social previews, and amazing support.”

Nyah

CEO · unlock.luxe

★

4.4 on Trustpilot

Read the 10 reviews →

[ questions ]

The questions skeptics ask first.

[ guides ]

Learn how AI and Google read your site.

All guides →

seo guideLovable SEO Guide 2026: Fix Indexing in MinutesStep-by-step fixes so Google and AI can read your Lovable app, with the options tested and compared.Read the guide →

All guides →

[ latest ]

Fresh from the blog.

All posts →

Technical SEO

Oct 27, 2026 · 8 min read

40% of "Googlebot" Traffic Is Fake: What We Found in 1.3M Requests

We checked 1.33M requests claiming to be Googlebot across ~400 sites. 40.5% didn't come from Google. What fakes do, why it matters, and how to verify.

read the article →

Technical Help

Oct 26, 2026 · 10 min read

Crawled or Discovered – Currently Not Indexed: What It Means and How to Fix It

Crawled – currently not indexed vs Discovered – currently not indexed in Search Console: what each means, the usual causes, and a fix list for both.

read the article →

SEO Guide

Oct 26, 2026 · 11 min read

LLM SEO: How to Get Cited by ChatGPT, Claude and Perplexity

LLM SEO in 7 steps: let the right AI crawlers in, make pages readable without JavaScript, write answer-first content, and measure what AI actually fetches.

read the article →

All posts →

[ free scan ]

See every page AI crawlers read.

Rankings holding steady. Clicks falling. Is SEO still working?

Every request. By page, by bot.

Your app works. Crawlers see a blank page.

From crawl to customer, measured end to end.

One request path. The whole crawl surface.

Which AI answers with your pages

Which AI answers with your pages

Prerendering

Real bot or impostor?

Why AI is reading you

Rank & AI visibility

Routing rulesDNS

MCP server

Ask your agent why the crawlers left. Let it fix the cause.

Two ways in. Same crawl data.

Questions you can answer without a developer.

Is ChatGPT actually reading our site?

Did our new blog post get picked up?

Which pages does AI care about most?

Which AI answers send us visitors?

Is our traffic drop an SEO problem?

Is someone scraping us?

Founders

Marketers

Agencies

Founders shipped. Then they got found.

The questions skeptics ask first.

My Google clicks are dropping. Is SEO dead?

What does it mean when an AI 'reads' my page?

Why isn't my Lovable or Base44 site showing up on Google?

Will Google penalize prerendering as cloaking?

Which setup should I use, DNS or Cloudflare Worker?

My site is already server-rendered. Why would I need HadoSEO?

Do I need a developer?

How do you know a bot is real?

Learn how AI and Google read your site.

Fresh from the blog.

40% of "Googlebot" Traffic Is Fake: What We Found in 1.3M Requests

Crawled or Discovered – Currently Not Indexed: What It Means and How to Fix It

LLM SEO: How to Get Cited by ChatGPT, Claude and Perplexity

How much of your site can AI actually see?