Build Your First Google Ads Dashboard in Claude

Build Your First Google Ads Dashboard in Claude (No Code Needed)

Turn three spreadsheet downloads into a dashboard that shows which ad spend brings in real customers, not just form fills. Six steps, about an hour, no BigQuery.

Level: Beginner  ·  Time: about 60 minutes  ·  You need: Claude, Google Ads, and a lead list

Jump to a step0 of 6 steps done
Your finished dashboard (this data is for demo purposes)
Click the tabs to view the figures

This guide is for marketers who run Google Ads for leads and have never built anything with Claude. If you’re still mapping out how ads, landing pages and follow-ups fit together, start with our full-funnel marketing strategy guide.

Before you start, have these ready

  • You need a paid Claude account to create projects and upload files, check your account plans through Anthropic’s help center.
  • Access to your Google Ads account.
  • Your leads from a CRM tool such as HubSpot or Zoho, or a Google Sheet.

Why Google Ads reports aren’t enough

Google Ads tells you how many people filled in your form. It can’t tell you whether those people were a good fit or ever paid you. That part lives in your CRM.

What Google Ads sees What your CRM sees Ad click Form fill MQL SQL Customer Your dashboard joins both views
Those 118 leads in your Google Ads report may look great until you check your CRM and find that only three of them ever paid you.

Words you’ll see

WordDESCRIPTIONS
CRMWhere you keep your leads and customers (HubSpot, Zoho, or a spreadsheet)
MQLA lead that looks like a good fit
SQLA lead your sales team agreed to follow up
ArtifactAn interactive page Claude builds for you in the chat
ProjectA Claude workspace that remembers your files and instructions

Step 1: Pick one question

5 minutes · No tools needed for this step

Most dashboards fail because they try to answer everything at once, and you end up with a wall of charts you never act on. So before you download anything, decide on the one question you most want answered, since that question will shape every file you pull in Step 2.

If you’re stuck, start with the one most marketers care about: which campaigns bring in leads that actually turn into customers?

Examples of QuestionsDownload these
Which campaigns bring good leads? Start hereCampaign report + lead list
Where am I wasting money?Search terms report + lead list
Am I doing better than last month?Campaign report covering two months
Which keywords bring leads that never buy?Keyword report + lead list with stages
Pick a row. The right column is your shopping list for Step 2.

Done when: Your question is written in a note. You’ll paste it into Claude in Step 4.


Step 2: Download your data

15 minutes · Google Ads + your CRM

You only need three files to get started:

  1. Your Google Ads keyword report
  2. Search terms report and
  3. Lead list

The one rule that matters is to pull all three for the same date range, such as the last 60 days, because if the dates don’t match, Claude ends up comparing spend from one period against leads from another and the numbers stop making sense.

File 1 and 2: Google Ads keywords and search terms

  1. In Google Ads, open Keywords.
  2. Set the date range in the top right.
  3. Check that Cost, Clicks, Impressions and Conversions columns are showing.
  4. Click the Download icon and choose .csv.
  5. Open Search terms and repeat steps 2 to 4.
Google Ads · Keywords
Keywords 1Search terms 5AdsAudiences
📅 1 Sep – 31 Oct 20262⬇ Download4
KeywordCampaignCostClicksImpr.Conv.
custom sofa klSofa – GenericRM 1,82081214,32046
built in kitchen cabinetCustom KitchenRM 1,4104666,98019
rumah kemasBrand SearchRM 6405121,94034
3Make sure the Cost, Clicks and Conv. columns are visible before you download.
Button positions shift between Google Ads updates. Look for the calendar and the download arrow.

File 3: Your lead list

Export from your CRM, or copy from your Google Sheet. Keep the green columns. Delete the red ones.

leads-sep-oct.csv
NameEmailPhoneDate createdStageCampaign / UTMLost reasonDeal (RM)
A. Tana.tan@…012-…02/10/2026SQLcustom_kitchen8,400
R. Kumarr.k@…017-…04/10/2026Leadsofa_genericBudget too low
N. Aziznur@…019-…05/10/2026Customerbrand_search3,200
Delete names, emails and phone numbers. Claude doesn’t need them, and it keeps your customers’ data private.

No keyword in your lead list? That’s common. Claude can still match leads by campaign name or landing page. Adding UTM tags to your ad links will make next month’s report sharper.

Done when: you have three CSV files, same date range, no personal details.


Step 3: Set up a Claude Project

10 minutes · Time to work on Claude

One thing Claude lacks out of the box is context about your business, because every new chat starts with a blank slate. A Project fixes that by keeping your files and instructions in one place, so Claude already knows your currency, your lead stages and what counts as a good lead the moment you open a new chat.

  1. In Claude, create a new Project called “Google Ads Dashboard”.
  2. Upload your three files to the Project’s files.
  3. Paste the instructions below into the Project’s instructions, then edit the brackets.
Claude · Project: Google Ads Dashboard

Files

CSVkeywords-sep-oct.csv48 KB
CSVsearch-terms-sep-oct.csv112 KB
CSVleads-sep-oct.csv21 KB

Instructions

You are my PPC analyst. I run Google Ads to get leads for [your business]. Currency: RM…
What a ready Project looks like: three files on one side, your instructions on the other.
Project instructions · copy and edit
You are my PPC analyst. I run Google Ads to get leads for [your business].
Currency: RM.
Our lead stages, in order: Lead > MQL > SQL > Customer.
A "good lead" is one that reaches SQL or Customer.
If something in the data is unclear or missing, ask me instead of guessing.

The last line matters most. It tells Claude to flag gaps instead of filling them in, which keeps your numbers honest.

Done when: your Project shows three files and your edited instructions.


Step 4: Ask Claude to build your Google Ads dashboard

10 minutes · Get it done with a single prompt

Open a new chat inside your Project and paste this prompt. Swap in your question from Step 1 if it’s different.

Build prompt · copy and paste
Using the files in this project, build me an interactive Google Ads dashboard.

My main question: Which campaigns bring in leads that turn into customers?

Create 3 tabs:
1. Overview: spend, clicks, conversions and cost per conversion by campaign,
   compared with the previous period.
2. Waste Finder: search terms and keywords that spent money but brought
   no leads, or only leads that never reached SQL. Sort by money wasted.
3. Lead Quality: match my leads to campaigns. Show how many reached each
   stage (Lead, MQL, SQL, Customer) and the cost per SQL.

Use simple tables and bar charts. Add a date filter.
At the top of each tab, show the date range and anything you couldn't match.
Claude · new chat in Google Ads Dashboard
Using the files in this project, build me an interactive Google Ads dashboard…
I matched 212 of 258 leads to a campaign. 46 leads had no campaign or UTM, so I’ve listed them separately. Here’s your dashboard.
▦
Google Ads DashboardInteractive page · open
What to expect: Claude reports what it could and couldn’t match, then opens the dashboard beside the chat.

The first version won’t be perfect. That’s normal. Click through every tab, note what looks off, and fix it in Step 5.

Done when: you can open the dashboard and click through all three tabs.


Step 5: Make it tell you what to do

15 minutes · Follow-up prompts

Right now your dashboard tells you what happened, which is useful, but the real value comes when it starts telling you what to change. The trick is to ask Claude for one improvement at a time, so you can check that each change makes sense before you build on it.

Before
Search termCostConv.
sofa murah bawah rm100RM 4129
ikea sofaRM 3186
used sofa for saleRM 2900
After one follow-up
Next step: add 3 negative keywords to save about RM 1,020 a month.
Search termCostVerdict
sofa murah bawah rm100RM 4120 SQL
ikea sofaRM 3180 SQL
used sofa for saleRM 290No leads
One prompt makes the difference: The “after” view puts the most expensive problems at the top and spells out what to do about them.
What you wantPaste this
A to-do listAdd a Next Steps box at the top that lists the 5 biggest problems, ranked by money at stake.
Your own idea of wasteOnly count a search term as waste if it spent more than RM50 and brought no SQLs.
Fair comparisonsCompare every number with the previous period and show the change as a percentage.
Fewer mystery leadsList the leads you couldn’t match to a campaign, so I can check them.
Simpler chartsRemove anything I can’t act on. Keep only charts that help me decide where to spend.

Before you add any chart, ask yourself what you’d change because of it. If the answer is nothing, leave it out.

Want a second opinion on the landing pages your ads send people to? Our free site audit series shows what we check on real Malaysian brands.

Done when: your dashboard has a Next Steps box that names at least one change to make.


Step 6: Check the numbers before you trust them

10 minutes · Google Ads + Claude

AI can misread a column or skip a row. Tick each check below before you move any budget.

Build the checks in · copy
Add a note at the top of each tab showing the date range, any missing dates, and how many leads were unmatched.

Done when: all five boxes are ticked. Now you can act on what the dashboard says.

The whole guide on one screen

1
Pick one questionDecides your files
2
Download 3 filesSame dates, no personal data
3
Set up a ProjectFiles + instructions
4
Paste one promptClaude builds the tabs
5
Ask for next stepsOne change at a time
6
Check the numbersFive ticks before acting

When you’re ready to level up

Working from CSV files is the easiest way to start, but it does mean downloading fresh exports every time you want updated numbers. Once that starts to feel like a chore, these options let Claude pull the data in for you:

OptionWhat it gives youEffort
CRM connectorClaude reads your leads directly (HubSpot, Zoho and others)Low
Google Ads API via Google’s MCP serverLive Google Ads data, no exportsMedium
BigQueryA daily copy of your ad data that you keep long termHigh

For a deeper, technical build using all three, read Search Engine Land’s guide to building a Google Ads dashboard with Claude.

Start small

You don’t need the perfect dashboard on day one. Start with one question, three files and a single prompt, and once you trust what the numbers are telling you, come back and add the next question.

Wrapping Up

Google Ads will always tell you how many people filled in your form, but it was never built to tell you who actually became a customer. That’s the gap this dashboard closes, and with Claude, closing it takes an afternoon and three spreadsheet downloads instead of a developer and a data warehouse.

Once you’ve built it, you’ll probably find yourself opening it before every budget meeting, because it’s hard to go back to judging campaigns on form fills once you’ve seen which ones actually bring in paying customers. Keep it simple at first, and add new questions as they come up.

If your ads are sending people to landing pages that don’t convert, even the best dashboard can only show you the problem. That’s where a proper look at the page itself helps, and it’s exactly what we do in our free site audits.

FAQs

Do I need to know how to code?

No. You describe what you want in plain English, and Claude builds it.

Is it safe to upload my Google Ads data?

Remove names, emails and phone numbers first. Campaign and spend data is usually fine, but follow your company’s data policy.

Will the dashboard update by itself?

Not with CSV files. Upload new exports and ask Claude to refresh it. A connector can make this automatic.

Why don’t my numbers match Google Analytics?

Google Ads and GA4 count conversions differently, so small gaps are normal. Pick one source and stick with it.

Can I add Meta or Microsoft Ads later?

Yes. Upload their exports to the same Project and ask Claude to add a tab for each.

Demo figures are fictional. Screens are illustrations; Claude and Google Ads layouts change over time. Last checked 1 Nov 2026.

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *