---
title: "How to Build a Google Ads Dashboard with Claude that Ties Spend to Outcomes"
url: https://adtech.lol/articles/3358e2f8-a64a-4c3c-a66b-97e0e1c25a96
source_url: https://searchengineland.com/how-to-build-a-google-ads-dashboard-with-claude-that-ties-spend-to-outcomes-498456/
source_site: searchengineland
author: "Tania Brown"
published: 2023-10-12T00:00:00+00:00
tags: ["Google Ads", "PPC", "CRM", "Dashboard", "Analytics"]
publisher: AdTech Radar
---

# How to Build a Google Ads Dashboard with Claude that Ties Spend to Outcomes

> Create a comprehensive dashboard that connects Google Ads performance with CRM data to visualize the full funnel and improve PPC campaigns.

## Introduction
Google Ads tells you how much you spent on a keyword and how many conversions it drove, but not whether those leads were a good fit or what happened to them during the sales process. That information lives in your CRM.

When I took on PPC, I struggled to connect the two. I’d export our HubSpot contacts and check their landing pages and reported keywords, but the CRM lacked the Google Ads spend and performance data I needed to see the full funnel.

So I built a dashboard with Claude that matches the data for me. Now I can see what’s working, test changes, and measure their impact.

## What Google Ads Reporting Leaves Out
Google Ads reporting leaves a lot to be desired when you’re trying to see your full funnel. A couple of tools try to make the data more useful.

Google’s Gemini dashboards give you prompt-driven charts and AI summaries. Google announced them in May and has been rolling them out to accounts since September. They’re limited to Google Ads data, though, so you won’t see CRM data, profit data, or your other ad platforms in them. Google says Google Analytics support is coming soon.

CRM attribution, like HubSpot’s ads attribution, ties spend to deals without any code and shows you the ad group and keyword. What it can’t show you is which headline and description a contact came from.

But neither of these joins your ad data to lead quality at a granular level, and that’s the gap this dashboard aims to fill. Here’s what you need to build your own.

## Decide What You Want to Understand from Your Dashboard
Before you hook up any data sources, decide what you want to get from your PPC dashboard. Then check which data sources can give you that information.

For example, my dashboard is a Claude artifact, an interactive page Claude builds and updates for you, with a tab for each question I wanted answered.

### Account Overview: How is each campaign and ad group performing?
Spend, clicks, click-through rate, and conversions by campaign and ad group, compared with the previous period. It also shows which funnel stage each Google Ads conversion has reached (form fill, MQL, SQL, Opportunity, or Customer), allowing me to assess campaign effectiveness.

### Ad Groups and Assets: Which ads and headlines need work?
Each ad group’s ads, headlines, and descriptions, with Google’s ad strength and per-headline ratings. It points out when Google hasn’t rated enough headlines yet to show which ones to rewrite.

### Keyword Quality: Which keywords have landing page or relevance problems?
Every keyword’s Quality Score and its three parts (landing page experience, ad relevance, and expected click-through rate) next to its spend. A “Page brief” button appears for keywords whose landing page may need to be updated. The button copies a prompt for me to paste into a new session.

### Waste Finder: Where am I spending without getting qualified leads?
Search terms that spent money without converting, plus keywords whose leads never got past MQL and were disqualified. This helps me see which keywords I need to investigate further.

### HubSpot Leads: Which ad groups and keywords bring in leads that progress?
Every paid search lead, grouped by ad group, campaign, or keyword. Each lead shows how far it progressed (MQL, SQL, Opportunity, or Customer) or whether it was screened out by country or revenue. It also recovers paid leads HubSpot filed as direct or organic, by spotting visits that landed on an ad-only page.

### Ad Experiments: Did my changes work?
A log of changes (ad copy, landing pages, keywords, and bids), with before-and-after numbers worked out automatically from the change date so I can iterate and annotate.

### Priority Actions: What should I do next?
A ranked to-do list pulled from the other tabs. Each item has an “Ask Claude” button that copies a ready-made prompt for the right next step, such as a page brief or competitor ad research.

To be clear, I didn’t build this in a day. It grew out of need. To build your own, start with one simple need: “What do I want to know about my PPC campaigns so I can improve them?” Keep it in mind as you build, since it’s going to shape where you pull data from and how it gets reported.

### Choose Your Data Sources
The sources below don’t capture the full picture, and those gaps decide what your dashboard can show. Here’s what each one gives you as of October. Choose the sources that provide the information you most want in your dashboard.

#### Manual UI Exports
The Google Ads interface is the only place where most advertisers can access Auction Insights. It also has a version history for each ad, showing previous versions, when each one changed, and how each performed. You can download this information as a CSV.

The interface keeps two years of change history, whereas the API keeps 30 days. And the API’s change events may not include every change you see in the interface. If you want this information, you have to commit to re-exporting it and handing the files to Claude whenever you update your dashboard.

#### BigQuery Data Transfer Service
My dashboard reads from BigQuery, where my Google Ads data lives, through Claude’s BigQuery connector. I have a daily transfer set up to collect relevant data about each of my campaigns. The beauty of BigQuery is that once you’ve transferred the data into it, you own it. If you rely only on the Google Ads interface, you’ll only have your hourly, daily, and weekly data for 37 months.

The BigQuery transfer doesn’t include everything, though:
- No Auction Insights.
- No change history.
- No per-headline performance stats, only the headline text.
- No live data, since it only updates when the transfer runs.

#### Google Ads Scripts to Sheets or BigQuery
Google Ads Scripts can fill the per-headline data that a BigQuery transfer doesn’t provide. A script can run a GAQL query for each headline’s stats every day and save the results to Sheets or BigQuery.

#### Google Ads API through Google’s MCP Server
If you’d rather have Claude pull data from Google Ads directly instead of from BigQuery, Google’s MCP server connects Claude to the Google Ads API. Google open-sourced it in October 2025. It’s read-only and has three tools: list_accessible_customers, search (which runs GAQL), and get_resource_metadata.

Through the API, you get live data and 30 days of change events. You also get full per-headline stats with the headline text, for dates from June 5, 2025, and Performance Max search terms through campaign_search_term_view, with data from June 6, 2025.

#### Using the API Instead of the BigQuery Transfer
You can build a dashboard using the API alone, without setting up a BigQuery transfer. This gives you live numbers instead of the previous day’s data, per-headline stats without a separate script, Performance Max search terms, and recent change events.

The API only reaches back as far as Google keeps data. Hourly, daily, and weekly reporting data is limited to 37 months, while monthly, quarterly, and annual data is available for 11 years. Change events only cover 30 days. Auction Insights isn’t available through the API either. Google limits it to allowlisted accounts and isn’t adding new ones.

#### Your CRM
Your CRM shows what happened to each lead after the form fill. HubSpot, Pipedrive, Attio, Close, and Zoho CRM all have connectors in Claude’s directory, and Salesforce has one in beta as of October.

Bring in the fields that tell you whether a lead was any good and what happened to it:
- Lead status or lifecycle stage.
- Disqualification reason.
- Company size or revenue band.
- Deal amount.
- The date each stage changed.
- The source fields that record where the lead came from.

To tie each lead back to the ad that brought it in, you need a way to match the two. Many CRMs record source details, such as the campaign UTM, the keyword, and the first landing page, which can help you identify the campaign, ad group, and keyword. If you store the GCLID on each contact, you can match a lead to the exact click instead.

If you haven’t pulled this data in before, you may need to explain how your company categorizes leads and deals so your final reports reflect the true state of your funnel.

#### Google Analytics 4
GA4 shows you what visitors do after the click. Once it’s linked to Google Ads, it also shows Google Ads clicks and cost. It uses a different attribution model, though, so its numbers won’t match what Google Ads reports.

## Start Prompting
Once you know the sources you want to use to power your dashboard, you can prompt Claude with specific details.

Tell Claude:
- What you want to achieve with your dashboard.
- What connectors you want it to use to pull data.
- Which metrics you want to monitor.
- What time periods you want to be able to analyze.
- Which campaign or ad group characteristics you want to see at a glance, and which you want a more granular look into.
- How you want the information laid out, such as tables, line or bar charts, or pie charts.

Start with a few of your must-haves and, once you’re happy, add any additional elements you think may help you better run your campaigns.

For example, you might want to integrate Microsoft Ads and paid social campaigns on additional tabs with an overview tab that gives you the big picture across all your paid campaigns.

## Add Features That Make It an Analysis Tool
A report tells you what happened. An analysis tool helps you decide what to do next.

Once you’re able to pull accurate data and create a report, work with Claude to update it so you can:
- **Tie every view to a decision:** Before you add a new chart or table, know what you’d change based on it. Keyword quality next to spend tells you which landing pages to fix first. A list of lead outcomes by ad group tells you where to shift budget.
- **Join spend to lead outcomes at the level you manage:** If you make changes at the ad group or keyword level, that’s where your CRM data needs to land. Campaign-level totals hide the one keyword that’s sending you unqualified leads.
- **Define waste on your own terms:** Decide what counts as wasted spend for your business. Is it search terms that never converted, leads that were all disqualified, or both? Set a minimum spend before something counts and a bar for when total waste is big enough to act on. Spend you can’t judge yet, like leads nobody has worked, should show up separately and stay out of the waste total.
- **Compare against something:** A number on its own doesn’t tell you much. My Account Overview compares each period with the one before it, and my Ad Experiments tab provides before-and-after numbers from the date of each change.
- **Record what Google Ads doesn’t:** Google Ads doesn’t record edits to your landing pages. If you want to know whether a change worked, log it with what you expected to happen and how long you’ll wait before judging it. Keep the log somewhere outside the dashboard, like a Google Sheet or a file in Drive, since an artifact’s own storage can be wiped.
- **Point to the next step:** Rank what the dashboard finds by the money at stake so the biggest problems come first. In my dashboard, each priority action has an Ask Claude button that copies a ready-made prompt for the next step, like a page brief or competitor ad research. I also have different skills to research competitor landing pages and help analyze keyword viability.

## Fail-safes That Make the Numbers Trustworthy
AI reporting is only useful if you can trust the numbers. Before acting on your dashboard’s recommendations, build in checks to catch incomplete or inaccurate data.

My dashboard uses the first three checks below, and the last two are worth adding. Ask Claude to build in whichever ones you need.

- **Totals built from the rows underneath them:** Campaign totals are calculated from ad group rows instead of being pulled separately, so a campaign total can’t disagree with the ad groups below it.
- **Freshness banners:** These show the newest day of data and when the BigQuery transfer last ran, and flag transfers that are running late.
- **Warnings on incomplete date ranges:** If a date range extends beyond the newest complete day of data, the dashboard flags it. Otherwise, a partial day could look like a decline in performance.
- **Source notes on each page:** These explain where the data comes from and what’s missing, so you know the limits of what you’re seeing.
- **Missing-date checks:** These flag gaps in the selected date range that could otherwise look like a drop in performance.

It’s also worth periodically spot-checking your dashboard’s numbers against Google Ads reports.

My dashboard started with one question: Which Google Ads campaigns and keywords were bringing in leads that actually progressed through the sales funnel? Connecting ad performance with CRM outcomes helped me answer it, and each new question shaped what I built next.

You don’t need to build everything at once. Start with the question that matters most to your campaigns, make sure you can trust the data, and expand the dashboard as you find new opportunities to improve performance.

Source: [https://searchengineland.com/how-to-build-a-google-ads-dashboard-with-claude-that-ties-spend-to-outcomes-498456/](https://searchengineland.com/how-to-build-a-google-ads-dashboard-with-claude-that-ties-spend-to-outcomes-498456/)