Open beta — completely free, no card required.
GA4 conversion diagnostics · Open beta

Your GA4 already shows the drop.
ConvRadar tells you the cause and the fix.

Connect GA4 in two minutes, ask what’s wrong with your funnel, and get a prioritized list of the leaks killing conversions and how to fix each one.

Reads from
Google Analytics 4 Connect your GA4 property in 2 minutes
Runs in
Claude Add as MCP connector in Claude.ai ChatGPT Add as MCP connector in ChatGPT Telegram Chat with ConvRadar from Telegram Web Try the live demo right here

The analysis runs on our servers — not the AI guessing.

Find my biggest leak

Connect GA4 in 2 minutes — your first diagnosis runs automatically.

Open beta — completely free, no card. GA4 set up by Shopify, Wix, or an AI builder and never checked? The first thing ConvRadar tells you is what your tracking captures — and what’s missing.

LIVEDemo ShopMessages: 0 / 5Off-topic: 0 / 3
Compare the last 7 days to the previous 7. What changed?
Pulling overview metrics
Mapping funnel

Last 7 days vs previous 7 — Demo Shop

MetricThis weekLast weekΔ
Sessions6,9147,512−8.0%
Conversions5178−34.6%
Revenue$6,724$9,810−31.5%
CR0.74%1.04%−28.8%

Where it broke: the drop is concentrated in add_to_cart → begin_checkout (70% → 56%). Traffic mix is steady — this is a checkout-step regression, not an acquisition problem.

Ask me to dig deeper — e.g. "why is checkout dropping?" or "is this device-specific?"

How it works

Sign up. Add to Claude, ChatGPT, Telegram, or this page. Talk to it.

Three steps, and the third one is the whole product. The only thing between you and your first verdict is one hour of data collection — kicked off automatically the moment you connect.

01

Sign up & connect GA4

Create an account, connect your GA4 property with one Google OAuth click. Read-only — disconnect anytime.

~2 minutes
02

Use it where you already work

Paste the ConvRadar connector URL into Claude (web, Desktop, Cowork) or ChatGPT, ping it from Telegram, or just talk to it on the webpage. The MCP tools install themselves — no extension, no install wizard.

One paste
~1 hour
Initial data collection & baseline
03

Talk to it

Ask Claude, ChatGPT, Telegram, or the webpage chat what's going on. ConvRadar pulls the right slice of GA4, runs the diagnostic, and turns the result into a verdict you can defend.

First answer in ~10 min of chat
Why the answers hold up

Diagnosis runs on our server. Not in the chat.

Simple lookups — sessions, CR, top channels — we hand back as-is. But when you ask why something dropped, the analysis runs as Python and SQL against our own Postgres, not as a guess from the LLM.

01

Your GA4, in our Postgres.

Ingested nightly into fact tables. No live API calls, no sampling, no thresholds.

02

Diagnostics run as code.

Funnel drops, anomalies, segment attribution — Python and SQL on our server, against a 300+ audit playbook. The verdict is computed before the model sees it.

03

The LLM writes the sentence.

It receives the finished finding and turns it into the answer. The reasoning happens before the writing.

Read-only OAuth. Every number traces back to the GA4 slice it came from.

The product, unedited

Eleven honest minutes of the product in use.

Not a sizzle reel. No cuts, no fast-forward — me opening ConvRadar and using it end to end, exactly the way you will.

Like a senior analyst, on chat

It's a senior analyst in your chat.

Same interface you already use to write emails, brainstorm copy, debug code. You ask in plain English, ConvRadar pulls the right slice of GA4, and the answer comes back as a verdict — not another dashboard.

  • No new app to learn. Lives inside Claude, ChatGPT, Telegram, or right on this page.
  • Asks in plain English. Answers with numbers, segments, and the next action.
  • On call 24/7 — free while the beta is open.
$150–300/hrtypical senior analyst
Freeopen beta — no card
300+audits behind the playbook
Claude · acme.com analytics · illustrative
ConvRadar connected
YT
You· 14:32
Revenue's down ~18% week-over-week and traffic's fine. What's actually going on?
Claude Senior Analyst · ConvRadar
Pulled the funnel and split by device. Mobile-Safari checkout collapsed after your Mar 14 deploy — step 3→4 dropped from 71% to 9% on iOS 17. Desktop is fine. Paid channels are clean.
convradar.funnel_audit(site=acme.com, window=14d, split=device)
Verdict
Roll back the Mar 14 checkout deploy for iOS users. Recovers ~$15.4k/wk, ~92% confidence.
Ask a follow-up…
The silence after the chart

Your data isn't the problem. The silence is.

You open the report. Sessions held, conversions slipped, revenue is down — and that's where the help ends. The chart shows you the dip and walks away. You spend the afternoon poking around segments, trying to figure out if it's mobile, the new landing page, the pricing test you shipped last week, or just a noisy Tuesday. By 6pm you have a guess, not an answer. You ship something. You hope.

That's the problem. Not the data. The silence after.

ConvRadar talks back. Connect GA4 once and just ask — in Claude, ChatGPT, Telegram, or the chat right on this page — what's happening. It knows your numbers, knows where to look, and knows what to do next. Like a senior analyst who picks up the phone, except it's completely free during open beta and it's already inside the chat you use anyway.

Not a wrapper. Not a chat with no data. A 10-year CRO playbook, on call.
Why we know what to do

10 years of audits, compressed into a product.

I've been the top-rated analytics & conversion freelancer on Upwork for five years running. Over the last decade I've run more than 300 audits for e-commerce and SaaS teams — the reviews are public on my Upwork profile. ConvRadar is that playbook, distilled. Every check we run is a check I've run by hand, a hundred times. When you ask — in Claude, ChatGPT, Telegram, or the chat right here — what's wrong, you're getting the same answer I'd give if you'd hired me. At 1% of the price, available the moment you ask.

300+
Audits delivered
5 yrs
Top-rated on Upwork
126
Verified hypotheses in the diagnostic library
Ivan Pika — Conversion & Analytics

Ivan Pika — host of the Conversion & Analytics podcast. Every check in ConvRadar is one I've run by hand, a hundred times.

All episodes RSS

What you can ask

We don't just show the problem. We tell you the fix.

Real questions you've already asked yourself this week. Each one comes back with a verdict, the evidence, and — most importantly — what to actually do next. Not a chart. Not a "directionally interesting" observation. The fix. Sample answers below are illustrative — yours are computed from your GA4.

What can I do today to grow revenue?
Returns the single highest-leverage change you can ship in the next 24 hours — ranked by expected $ impact × confidence × effort. Not "test more variants." A specific surface, a specific change, a specific number.
Fix · no developer needed Re-enable PayPal on mobile checkout. 31% of mobile abandoners reach the payment step but bounce — desktop conversion is 2.4× higher. Est. +$8.4k/mo, ~88% confidence.
Not the diagnosis. The next thing to ship.
What's my biggest problem right now?
One verdict, ranked across the whole site. The thing that, if you ignored everything else, you'd want to fix this week.
Fix · hand to engineering iOS Safari checkout broke Mar 14. Roll back the deploy for iOS users.
Where am I losing the most users?
Maps every funnel step, finds the single biggest drop, and tells you in dollars what fixing it would recover.
Fix · hand to engineering Address step on mobile. Replace the autocomplete widget — recovers ~$11.2k/mo.
Why did conversions drop?
Returns the segment, the date, and the most likely cause — ranked against your own history, not the internet average.
Fix · no developer needed The app-install upsell you added before cart on Apr 1 is dragging the whole funnel down. Remove it for returning users, or move it post-purchase.
I shipped a change — did it work?
Pre/post window with confidence interval. "Significant lift", "no detectable effect", or "you broke something" — it'll say which, and what to do about it.
Fix +3.1pp lift, p<0.05. Roll out to 100%, then keep watching mobile.
Run a full audit.
The whole 10-year playbook in one pass — funnel, traffic mix, products, devices, anomalies. Verdict and a fix per finding.
Fix ~15 ranked actions with $ impact, ordered by what to ship Monday.
Is my conversion rate actually good?
Graded against your own traffic mix — a site fed by YouTube or social converts differently than one fed by branded search. Verdict: healthy, leaking, or broken, with the number that’s normal for your shape.
Verdict 1.1% on creator traffic is median, not broken. Your real outlier is the cart step on mobile — fix that before touching the landing pages.
Is it my traffic or my site?
ConvRadar rules one out instead of blaming both. Funnel steps steady + junk sessions = traffic. Traffic mix steady + a step regression = site. You get the verdict either way — including “your page is fine.”
Verdict Your booking page is cleared. The 300k new sessions are zero-intent — pause the campaign and hand this verdict to whoever runs it.
What the right answer actually looks like

Five findings from the audits behind the playbook. Anonymized under client NDAs.

Every row is the output of a single question — asked in Claude, ChatGPT, Telegram, or right here in the page chat. Verdict, evidence, recommended action — paste it into a Slack thread and stand behind it.

Industry Problem we found Fix we shipped Recovered
DTC apparel Mobile checkout CVR 38% below desktop Leak localised to the address step on iOS Replaced address autocomplete widget, removed a blocking JS error in the ZIP validator. +$42,800/mo mobile CVR +51%
SaaS Sign-up to activation drop at the "invite teammate" step Anomaly started the day after a deploy; only affected free plan Made the invite step skippable; activation recovered within 48 hours. +27% activation ~$9,100 MRR added
Beauty e-com Paid search CVR collapsed on one campaign Intent mismatch — page promised "gift sets", campaign bid on "refill" Split the ad group, pointed refill queries at the correct PDP, paused 4 mismatched keywords. +$18,400/mo paid CAC −23%
Marketplace Top-10 SKU with above-average traffic, below-average CVR Product images missing; variant picker broken on mobile Re-uploaded photography, rebuilt the variant picker as a native select on mobile. +$11,600/mo SKU view→cart +64%
Subscription Organic traffic grew 22%, revenue flat New traffic landed on a blog post that didn't link to the offer Added an inline CTA block and an exit-intent offer; kept the editorial tone. +$6,900/mo blog→trial +3.1pp

Numbers rounded. These are from the founder’s hands-on client audits — the same playbook ConvRadar now runs automatically. Anonymized under NDA; ask and we’ll show how each was measured.

FAQ

Questions, answered before you ask.

Do I need a Claude or ChatGPT account?
Only if you want to use it from there. ConvRadar runs as an MCP connector inside Claude.ai / Desktop / Cowork and ChatGPT (any plan, including free). If you'd rather skip both, ConvRadar also lives in Telegram and right here on the webpage — same playbook, no external account.
NewWill this just be more AI noise?
No. The reason ConvRadar gives a useful answer where ChatGPT or a generic GA4 plugin doesn't is that it isn't a wrapper. It's a 10-year CRO playbook — same checks I run by hand for paying clients — wired into Claude as structured tools. When you ask "why did mobile drop", the connector pulls the right slice of GA4, runs the diagnostic against a hypothesis library of 300+ audits, and only then does Claude turn the result into a sentence. The reasoning happens before the writing.
NewWill this give me something I can actually defend in front of my boss or client?
That's the point. ConvRadar's answer always has the same shape: verdict, evidence, recommended action. You can paste it straight into a Slack thread, a status update, or a client doc and stand behind it. If someone pushes back, the evidence is right there — which segment, which date, which step in the funnel. The whole format is built for the conversation that happens after the audit, not just the audit itself.
What about my data — is it safe?
GA4 access is read-only via Google OAuth. We never write to your property, never see your raw user-level events outside the metrics GA4 already aggregates, and you can revoke access in one click. Numbers used to answer a question stay in the audit log so you can see exactly what was pulled. Disconnect GA4 or delete your account, and the GA4 copies we ingested are deleted with it.
NewMy GA4 was set up by an app or an AI builder and I’ve never checked it. Will this work?
Yes — and you’ll know exactly what shape your tracking is in before you trust a single number. ConvRadar reads what your property actually tracks and builds the funnel from what’s there — standard GA4 events or your custom equivalents. Ask “what does my tracking capture?” and you get the read-out: which events exist, how far back the history goes, and what’s missing for a full picture. If you ask about something your GA4 doesn’t track, the answer is “your tracking doesn’t capture this yet” — never a guess. A guided tracking-setup tool is next on the roadmap; today you get a plain-words read on what your setup can and can’t answer.
Does this work for SaaS, or only e-commerce?
Both — plus lead-gen, marketplaces, and subscription apps. The funnel logic is adaptive: for an e-com store it's view → cart → checkout → purchase; for a SaaS it's signup → activation → upgrade; for lead-gen it's visit → form view → submit. ConvRadar reads what your GA4 actually tracks and builds the funnel that makes sense for you, not a template.
How long until I get a useful answer?
Ten minutes to your first answer in chat — plus about 1 hour on top, on day one, while ConvRadar does the initial GA4 sync. Two minutes to sign up and connect GA4, one minute to paste the connector URL into Claude, ChatGPT, or Telegram (or just open the chat on this page), the sync runs in the background, and the first answer to whatever you ask comes back faster than the meeting you'd have called to discuss it.
Who is it for?
Honestly: e-commerce, SaaS, lead-gen, and anything in between — as long as it has a funnel and runs on GA4. Stores, subscription apps, marketplaces, B2B lead-gen sites, content businesses with paid CTAs, agencies running audits across many clients (one GA4 property per workspace today — a multi-client setup is in the works; ping us via Contacts to join the pilot). The funnel logic adapts to what your GA4 actually tracks, so a Shopify store and a SaaS dashboard both get the audit that makes sense for them, not a template.
How much does it cost?
Free during open beta. Sign up with your email — no card, no checkout, no time limit while the beta is open. We're collecting users and feedback before launch. A senior analyst is $150–300 an hour; ConvRadar is free right now. When the beta ends, pricing is announced by email in advance — nothing can auto-bill (there’s no card on file), and you can export or delete your data at any point.
Can I cancel anytime?
There’s nothing to cancel during the beta — no card on file, nothing can auto-bill. Disconnect GA4 or delete your account in one click, and your ingested data is deleted with it. When paid plans launch, you’ll get email notice first; cancelling then is one click too — no questions, no retention call.
Stop guessing

Connect GA4. Wait one hour. Get the fix.

Two minutes to sign up and connect GA4, one to paste the connector URL into Claude, ChatGPT, or Telegram (or just chat with it right here), an hour for the initial sync — then ask anything. Senior-analyst answers, completely free during open beta, no card required.

Get free access →
Get free access