If your GA4 property shows 12,437 sessions and 41 conversions this month, you actually know nothing useful about what is working. GA4 Looker Studio funnels turn that flat number into a five-step story you can act on before Monday coffee: how many people hit pricing, how many clicked checkout, and where the real drop-off happened. Add a weekly AI summary and you get a plain-English recap in your inbox instead of another dashboard you will forget to open.
TL;DR: Build GA4 Looker Studio funnels once (about 90 minutes), then use Make.com to send the last 7 days of funnel data to Claude every Monday. You get a five-sentence briefing that names the one number that changed last week, not another chart you will ignore.
What you’ll get from this setup
- A working 5-step GA4 funnel piped into Looker Studio with source/medium breakdown
- A weekly AI summary delivered every Monday at 7am your local time
- A copy-paste Claude prompt tuned for solopreneur traffic volumes (100–5,000 sessions/week)
- A sanity checklist so you catch broken tracking before you trust a single number
Why GA4 Looker Studio funnels beat the default GA4 reports
The default GA4 funnel exploration works, but it is trapped inside the GA4 UI. You cannot share it with a cofounder or a freelancer without granting property access. You cannot blend the numbers with Stripe or ConvertKit. And you cannot feed the raw table to an LLM on a schedule.
GA4 Looker Studio funnels solve all three problems in one dashboard. Looker Studio lets you publish a view-only link, blend GA4 with any connector Google supports, and expose the underlying data as a table an automation can read [source-needed]. That last capability is what makes the weekly AI summary possible — every other benefit is a bonus.
[test-claim] On a small SaaS I run, the GA4 default exploration reported a 68% drop between view_pricing and begin_checkout. The Looker Studio version, filtered to exclude bot traffic from two specific ASNs, showed the real drop was 54%. That single filter changed which experiment I ran the following week, and it moved the funnel two percentage points in ten days.
The 5 events your funnel actually needs
Skip the 40-event tracking plan you read in some enterprise blog. For a solo founder or a 1–5 person team, five events are enough to power useful GA4 Looker Studio funnels:
session_start— auto-collected by GA4, nothing to configureview_itemor a customview_pricing— fires when someone loads your pricing pagebegin_checkout— fires when someone starts the paid signup flowadd_payment_info— fires when card details are enteredpurchase— fires on successful payment (server-side is best)
If you use Stripe Checkout, the last three events can come from the official Stripe → GA4 integration or from a webhook you fire yourself [verify pricing]. If you use Lemon Squeezy, Paddle, or Gumroad, add them as custom events via Make.com, a Zapier scenario, or Google Tag Manager.
Step 1: Configure GA4 key events in 15 minutes
Before your GA4 Looker Studio funnels can show real data, GA4 needs to know which events matter. Open GA4 → Admin → Events. Find each of the five events above. If one is missing, fire it from your app code, or create it as a custom event from a matching page URL and event condition.
Mark each of the five as a “Key event” using the toggle in the events table [source-needed]. That toggle is what makes them selectable in funnel explorations and, more importantly, in Looker Studio’s funnel chart component.
Gotcha: GA4 sometimes takes 24–48 hours to backfill key events into reporting once you toggle them [source-needed]. If you build this on a Sunday for a Monday briefing, you will get an empty funnel. Set the toggles on a Friday and your first weekend of data will be clean.
Step 2: Build the GA4 Looker Studio funnels dashboard
Head to lookerstudio.google.com and create a blank report. Add the GA4 data source, pick the correct property (the display names look near-identical for many accounts, so double-check the measurement ID), and give it a minute to load the schema.
Now add a Funnel chart via Insert → Chart → Funnel. Configure it like this:
- Dimension: Event name
- Steps: Add the five events in order
- Metric: Unique users triggering the event — not raw event count, which double-counts anyone who reloads the pricing page
- Date range: Last 7 days, comparison to previous 7 days
- Filter: Exclude internal IPs, exclude known bot user agents
If you drive traffic from LinkedIn, Twitter, or a newsletter, add a source/medium dimension as a segment. Now the same GA4 Looker Studio funnels dashboard shows you which channel converts and which channel quietly wastes your paid ad budget.
Save the report and share a view-only link with yourself. You will need that link for the automation and for the weekly briefing email itself.
Step 3: Automate the weekly AI summaries with Make.com
Here is where GA4 Looker Studio funnels stop being another dashboard and start being a Monday-morning briefing you will actually read. The scenario I run in Make.com:
- Trigger: Scheduler module, every Monday at 6:30am your timezone
- Step 1: Google Analytics Data API module — pull the last 7 days of funnel event counts, grouped by event name and source/medium
- Step 2: A second Analytics module — pull the previous 7 days for a like-for-like comparison
- Step 3: Anthropic Claude module (or OpenAI if you prefer) — send both JSON blobs with the prompt below
- Step 4: Gmail, ConvertKit, or your ESP — email the summary to yourself and any cofounder who needs it
In each Analytics module, use the Data API v1 endpoint. Set the dimension to eventName and the metric to eventCount, add a dimension filter that keeps only the five funnel events, and set the date range to “yesterday minus 6 days” through “yesterday”. You will end up with a clean JSON payload under 2KB, which costs almost nothing to send to Claude.
Make.com charges by “operations” — the scenario uses roughly 6 operations per run, so about 24 operations per month [verify pricing]. That fits inside the free tier for most solo accounts and never shows up on the bill for paid ones.
Step 4: The exact GA4 Looker Studio funnels AI prompt
This is the Claude prompt I actually use. It is tuned for tiny data — 100 to 5,000 sessions per week — where percentages get noisy fast and you care about direction more than statistical significance:
You are a growth analyst for a solo SaaS founder. You will receive two JSON blobs: this week's funnel and last week's funnel. Each has five steps: session_start, view_pricing, begin_checkout, add_payment_info, purchase.
Write a 5-sentence Monday briefing:
1. The single biggest change week-over-week (up or down, absolute number and percentage).
2. Which funnel step is now the weakest (lowest step-to-step conversion).
3. One plausible cause based on the shape of the change.
4. One experiment I could run this week to test that cause.
5. One number to watch next Monday.
Do not use jargon. Do not hedge. If sample size is under 100 sessions, say "too small to conclude" for point 1 and skip points 3 and 4.
The “too small to conclude” line matters. Without it, any capable LLM will confidently narrate random noise as a trend and you will waste a week chasing a signal that was not there.
A good week’s output reads like this: “Sessions grew 18% (2,140 vs 1,815). Weakest step is view_pricing to begin_checkout at 4.1%. Likely cause: the new LinkedIn campaign brought colder traffic. Test a shorter pricing page above the fold. Watch begin_checkout count on Monday.” If your output rambles or refuses to name a number, tighten the prompt — do not accept vagueness.
Step 5: Sanity-check the numbers before you trust the briefing
Before you make product or spend decisions from your GA4 Looker Studio funnels, run one manual check the first Monday:
- Compare the
purchasecount in Looker Studio to Stripe’s dashboard for the same date range. - If they match within 5%, ship it. Move on.
- If they do not match, your event is not firing on 100% of purchases.
The most common cause of a mismatch: purchase fires on the client side and gets blocked by ad blockers, iOS Safari privacy, or Brave browser defaults. The fix is to fire it server-side from a Stripe webhook. See {{internal:ga4-server-side-tracking}} for the full server-side setup when the discrepancy is above 10%.
Bottom-line recommendation
Do this whole thing in one sitting. Ninety minutes for the GA4 Looker Studio funnels build, another thirty for the Make.com automation and the Claude prompt. Skip the dashboard beautification. Skip the extra dimensions you will never look at. Skip the tempting third comparison window.
The point is one email, five sentences, once a week. If your setup cannot produce that, you have overbuilt it. If it can, you have replaced roughly two hours of Monday-morning analytics rummaging with 90 seconds of reading.
For any solopreneur or small team under 10,000 weekly sessions, this stack replaces every paid analytics dashboard I have tested. Above that volume, you will want to layer product analytics tools like Heap or Mixpanel — see {{internal:heap-mixpanel-comparison}} for when the jump is actually worth the money.
FAQ
Do I need Google Tag Manager for this?
No. If the five events are already firing (from a Stripe integration, a Segment pipe, or your app code), the whole funnel builds in the GA4 UI without touching GTM. GTM only helps if you need to add or modify events on the front end.
How much does this stack cost per month?
GA4 and Looker Studio are free. Anthropic Claude API for one weekly summary is a few cents per month at most [verify pricing]. Make.com’s free tier covers the scenario for most solo accounts [verify pricing]. Realistic all-in: under $5/month.
Can I use ChatGPT or Gemini instead of Claude?
Yes. Any long-context model handles this workload fine. I use Claude because it holds structured output better when the input is two JSON blobs and the output must be exactly five numbered sentences. Swap freely if you already pay for another provider.
What if my traffic is under 100 sessions per week?
Build the funnel anyway, but skip the AI summary for the first month. Below 100 sessions, week-over-week changes are almost always noise. Use the raw funnel to catch broken tracking, not to make product decisions.
Can I add Stripe MRR data to the same dashboard?
Yes, via Looker Studio community connectors [source-needed]. See {{internal:stripe-looker-studio-connector}} for the exact steps and the two connectors worth paying for.
How often should I redesign the funnel?
Every quarter, or whenever you launch a new pricing page. The five events stay the same; the labels and comparison windows should update to match whatever you are testing.
What to do next (10 minutes)
- Open GA4 Admin → Events and verify the five events above exist. Mark any missing ones as key events.
- Create a blank Looker Studio report and connect your GA4 property (2 minutes).
- Book a 90-minute focus block on your calendar this week for the full GA4 Looker Studio funnels setup plus the Make.com automation.