---
title: "The 10% Rule: A 60-Day ChatGPT Ads Test Funded by Google"
description: "Move 10% of your Google Ads budget to ChatGPT for 60 days. Use one qualified-lead scorecard and a kill rule agreed before launch to decide where the money goes next."
url: "https://groas.com/post/take-10-of-your-google-ads-budget-and-give-it-to-chatgpt-for"
image: "https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/cef47934-4da8-4ff9-9a37-c9dadb2fad2c.png"
published: "2026-11-01T06:14:48.597Z"
modified: "2026-11-01T06:14:48.656Z"
---

November 1, 2026 · 10 min read

# The 10% Rule: A 60-Day ChatGPT Ads Test Funded by Google

[Alexander PerelmanHead Of Product @ groas](https://groas.com/author/alexander-perelman)[LinkedIn](https://www.linkedin.com/in/alexander-433793253/)

![A ten-square chocolate bar with one weak corner square snapped off onto its own saucer beside a kitchen timer: a test funded by 10% of existing spend.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/cef47934-4da8-4ff9-9a37-c9dadb2fad2c.png)

In this article

1. [1. Day 0: Set the 10% cap and choose what gets cut](#1-day-0-set-the-10-cap-and-choose-what-gets-cut)
2. [2. Day 0: Name the Google group ChatGPT must beat](#2-day-0-name-the-google-group-chatgpt-must-beat)
3. [3. Days 1–7: Spend about six hours making leads comparable](#3-days-17-spend-about-six-hours-making-leads-comparable)
4. [4. Days 8–30: Hold the cap and change one thing at a time](#4-days-830-hold-the-cap-and-change-one-thing-at-a-time)
5. [5. Days 31–50: Compare qualified leads, not attractive dashboards](#5-days-3150-compare-qualified-leads-not-attractive-dashboards)
6. [6. Day 60: Apply the rule you signed on day 0](#6-day-60-apply-the-rule-you-signed-on-day-0)
7. [7. When the 60-day constraint lifts, give the money to the winner](#7-when-the-60-day-constraint-lifts-give-the-money-to-the-winner)

Move **10% of your current Google Ads budget to ChatGPT ads for 60 days**, and add no new money. If ChatGPT cannot beat your weakest viable Google ad group on cost per qualified lead, it does not get a budget increase on day 60.

I am skeptical of new ad placements by habit. I ran Google Ads for years for ecommerce, home services and a SaaS client that changed its offer every quarter. I watched the same cycle repeat: big promise, thin reporting, then someone has to explain why form fills went up while sales stayed flat. ChatGPT ads could be another turn of that wheel. I will test them, but I will not give them fresh budget on faith.

Google supplies the comparison, 10% caps the spend, and 60 days sets the clock. The rule for what happens afterward gets written before the first dollar moves. Here is the sequence I would run for a lead-gen account.

## 1. Day 0: Set the 10% cap and choose what gets cut

Say you spend $20k a month on Google. For the next 60 days, spend $18k on Google and $2k on ChatGPT. Total spend does not move. Roughly $4k goes through ChatGPT over two months; that is a reallocation, not an extra $4k bill.

Take the money from a weak Google ad group rather than shaving 10% off every campaign. Leave the group running at a reduced budget so you can compare both channels over the same dates. If one group cannot supply the full 10% while remaining a useful comparison, trim other weak groups too. **Do not cut a winner to manufacture a test.**

This is why I prefer a carve-out to new money. Add $2k and every ChatGPT lead looks incremental, even though nobody has checked what that $2k would have bought on Google. Move the $2k and ChatGPT has to earn its place against the work you displaced.

The channel is new enough that outside benchmarks will not settle the question for you. The [US test began Feb 9, 2026, and self-serve Ads Manager opened May 5, 2026](https://www.airops.com/blog/chatgpt-ads-performance). OpenAI says there are no cross-advertiser benchmarks yet, and delivery and spend fluctuate. I used to tell clients to wait for benchmarks before testing a new placement. I was wrong. Your own account gives you a more useful comparison than somebody else’s average CTR.

Before you allocate an hour to setup, check the audience. [ChatGPT ads show to Free and Go users, while Plus, Pro, Business and Enterprise remain ad-free](https://www.webfx.com/blog/ai/chatgpt-ads-vs-google-ads/). If your buyers chiefly use paid seats, skip the experiment. Do not spend 60 days proving you cannot reach them.

The same goes for a budget too small to run the test. The reported self-serve floor is $25 a day, about $750 a month. If your Google spend is under $7,500 a month, its 10% share falls short of that floor. Keep the money in Google rather than quietly breaking the cap.

## 2. Day 0: Name the Google group ChatGPT must beat

Pick one underperforming Google ad group as the opponent. It needs at least 30 qualified conversions in the previous 60 days and a reasonably stable cost per qualified lead. A group with three leads and one unusually good week is not a control; it is a story waiting to be told badly.

Record its spend, qualified leads and cost per qualified lead before the test. Then keep recording those figures during the 60 days, while the group runs on its reduced budget. Use the earlier period to understand what normal looks like, but make the final decision on **the same dates and the same definition of a qualified lead** for Google and ChatGPT. If the budget cut makes the Google group too thin to measure, choose a different group or do not run this version of the test.

Do not compare ChatGPT with your whole Google account. The strongest branded group can make any new placement look hopeless; an account average can hide an expensive lead source. The question is narrower: can ChatGPT do better with money currently assigned to one of your weaker groups?

Nor should [case studies reporting 15.3% lower CPA for WeightWatchers or 67% net-new customers for Dose](https://trendyinai.com/blog/chatgpt-ads-review) become your forecast. Those figures come without the spend, sample or dates you would need to use them as a budget rule. Write down your own baseline. That page is the argument you will return to on day 60.

![Google and ChatGPT ads compared using one shared qualified-lead scorecard](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/b8d8e93e-0963-4d42-92c9-695c1b9e6765.png)

## 3. Days 1–7: Spend about six hours making leads comparable

**Shared lead-quality tracking comes before creative.** Capture the [GCLID alongside email and phone on Google leads, stamp qualified_at and won_at in your CRM, and import the relevant offline conversions into Google daily](https://growthsparx.com/blog/offline-conversion-tracking-google-ads.html). Bid Google toward the qualified stage, such as an SQL or demo, rather than rewarding every raw form fill. Otherwise, Google gets better at finding people willing to submit forms, which is not necessarily the same job.

For ChatGPT, set up its pixel or Conversions API and use UTMs to identify its traffic in GA4 and your CRM. A GCLID is Google’s identifier; it is not a shared identifier for both channels. What you share is the CRM definition and scorecard: the same qualification criteria, the same status dates and one place to count qualified leads. If one platform reports form fills and the other reports SQLs, the next 53 days are theater.

Then rewrite one landing page for the way a chat user arrives. Google pages answer keywords. A ChatGPT visitor may arrive after a conversation and have follow-up questions. Cover price range, fit and what happens after submission. Keep the form the same as the Google page’s form so you do not introduce a second avoidable difference.

Someone must own the setup in the accounts, not in a slide deck. That means Ads Manager in your name, the pixel and UTMs connected to your reporting, conversation research, context hints, creative tests and a person watching bids while spend reporting [lags by seven to eight hours](https://outrankco.sg/services/chatgpt-ads-singapore/). Those are operating tasks. Calling them a strategy presentation does not do them.

## 4. Days 8–30: Hold the cap and change one thing at a time

ChatGPT targeting is not a Google keyword list. [Advertisers describe relevant conversations in natural language, and an inference engine matches ads in a relevance-weighted second-price auction](https://www.webfx.com/blog/ai/chatgpt-ads-vs-google-ads/). There are no keyword lists, demographics, Auction Insights or impression-share reports to work through. The hands-on levers are context hints, the feed, headline and image tests, and landing-page relevance. If you catch yourself planning to mine ChatGPT search terms, stop. That control is not there.

Set a $4 starting max bid and a daily cap that keeps ChatGPT within its 10% monthly share. Leave the bid alone for seven days at a time and change one context hint per week. A spend dashboard that runs behind real time is a poor place to make hourly decisions.

Watch **cost per qualified lead, not cost per click**. Reported click-through rates help explain why traffic may arrive slowly: [Adthena measured 0.91% in ChatGPT against 6.4% in Google, while other readings ranged from 0.68% to 1.06%](https://trendyinai.com/blog/chatgpt-ads-review). Reported costs vary too: [small tests put CPCs from $4.41 to about $13, Workshop Digital paid $3.14 across 739 clicks, and OpenAI recommends a $3 to $5 starting max bid](https://www.airops.com/blog/chatgpt-ads-performance). None of those numbers tells you whether your own clicks become qualified pipeline.

Do not declare a winner in week two because the cheaper channel produced more forms. Keep the cap, fix obvious tracking problems and give the CRM time to tell you what those forms became.

## 5. Days 31–50: Compare qualified leads, not attractive dashboards

By day 31, you may have enough discomfort to quit early. Do not mistake discomfort for a result. [StubGroup reported a 0.21% conversion rate in ChatGPT versus 3.71% in Google; Workshop Digital spent $2,319 in ChatGPT for zero MQLs while Google Search produced two](https://www.airops.com/blog/chatgpt-ads-performance). Those tests are reasons to demand a lead-quality comparison, not reasons to substitute their outcomes for yours.

Run one weekly table for both channels: spend, qualified leads, cost per qualified lead and lead-to-SQL rate. Use the shared CRM definitions and matching date windows. Keep raw lead counts available for diagnosing the funnel, but do not let them decide the budget. Remove CTR from the decision table entirely. What the deck calls _upper-funnel efficiency_, I call paying for clicks that never become pipeline.

Account for reporting delays before reading a single bad day as a trend. [Google’s imported offline conversions report against the click date, and GCLIDs expire after 90 days](https://growthsparx.com/blog/offline-conversion-tracking-google-ads.html). ChatGPT spend can lag in its dashboard. Judge weeks, not afternoons, and keep importing qualified outcomes promptly.

Suppose the live Google comparison holds at $180 per qualified lead and ChatGPT sits at $600 after 15 qualified leads. You do not yet have the sample required to scale, and the gap is not a bid-tuning victory waiting to happen. Keep the agreed cap through day 60. **Same scorecard, same dates, no excuses for either side.**

## 6. Day 60: Apply the rule you signed on day 0

Write the rule before launch and get approval from whoever owns the Google budget. Mine has three outcomes:

1. **Increase ChatGPT’s share only if it beats the live Google comparison on cost per qualified lead and produces at least 20 qualified leads.** Winning on five leads is not enough; coming close on 20 is not a win.
2. **Hold at 10% for 30 more days** if ChatGPT misses the winning mark but stays within 50% of Google’s cost per qualified lead and its lead-to-SQL rate is improving. This buys more observation, not more budget.
3. **Kill the test and return the 10% to Google** if it meets neither condition. Do not approve a larger repeat to make the first result feel worthwhile.

That agreement prevents the inevitable day-60 rewrite: a channel misses the qualified-lead target, then somebody changes the presentation to emphasize clicks. A beta placement does not deserve a softer budget rule because its reporting is interesting.

For regulated advertisers, bring compliance into that approval before launch. An [independent audit of the first month, using 91 sock-puppet accounts and more than 3,600 ads, found no guardrail failures on violent or hateful content but did find that low-income zip-code accounts received ads more often and in greater numbers](https://www.brookings.edu/articles/understanding-the-beginning-of-advertising-on-chatgpt/). That does not make the test an automatic no. It makes guardrails and ownership part of the decision made before spending.

This is also where I separate hands-on management from consulting. A consultant can hand you the rule. A manager has to wire the tracking, run the capped bids, revise hints, reconcile the CRM and recommend killing a test they spent two months operating. I put both channels under one owner because separate owners have a habit of producing separate definitions of a lead.

[Importing offline SQLs into Google tripled qualified pipeline on the same budget in one B2B case I point clients to](https://groas.com/post/b2b-google-ads-pipeline-quality-optimization-offline-conversion-import-case-study). If your forms rarely become pipeline, start with [why B2B Google Ads leads do not convert](https://groas.com/post/b2b-google-ads-leads-dont-convert-pipeline-optimization-case-study) before buying a second channel. groas runs [Search Ads across Google, Bing and ChatGPT](https://groas.com/pricing) for a flat $1,499 a month, with bids, keywords and budgets worked around the clock and a landing page for every ad group. For narrower experiments, use these [small-budget ChatGPT ad tests with a question and kill rule for each](https://groas.com/post/7-chatgpt-ad-tests-you-can-run-on-a-small-budget-each-with-a). Hire the team willing to kill its own test at day 60. Anyone else is selling you a second retainer.

## 7. When the 60-day constraint lifts, give the money to the winner

If ChatGPT won, keep its 10% and test a 20% share for the next 30 days under the same qualified-lead rule. Reallocate from Google; do not add new money just because the first test worked. If ChatGPT earned a hold, leave it at 10% for one more 30-day window and require it to close the cost-per-qualified-lead gap. If it lost, return the money to Google.

Keep the shared offline tracking either way. The channel decision ends; the ability to see which leads become pipeline should not.

Most new placements will not earn a permanent budget in their first 60 days. A capped test can make that useful information. The waste I object to is the retainer that runs for a year without a kill rule, shared tracking or anything in its report more persuasive than clicks. When the constraint lifts, fund the winner. Do not fund the explanation.

## Frequently Asked Questions

### Does the 10% ChatGPT ad test require adding new budget, or can it be funded by reducing existing Google Ads spend?

The 10% allocation should be reallocated from a weak Google ad group instead of requiring additional spending. This method ensures that ChatGPT must earn its placement against displaced spend rather than gaining budget based on its incremental results.

### What criteria should be used to select a Google ad group to compete against the ChatGPT ads?

The chosen Google ad group must have at least 30 qualified conversions in the previous 60 days and exhibit a reasonably stable cost per qualified lead. A group with only a few leads or one unusually strong week should not be used as a baseline for comparison.

### How can lead quality and performance be compared between ChatGPT and Google Ads?

A single scorecard and identical qualification criteria must be used for both platforms. Performance should be measured using metrics like cost per qualified lead and lead-to-SQL rate, rather than relying solely on click-through rates or raw form submissions.

### How should performance levers be manipulated when running ChatGPT ad campaigns?

ChatGPT targeting does not rely on keyword lists or demographic targeting. Instead, the hands-on levers include adjusting context hints, testing headlines and images, and ensuring landing-page relevance matches the conversational nature of the visitor.

### What are the three conditions for determining the budget's future after the 60-day test concludes?

Budget should increase only if ChatGPT beats Google on cost per qualified lead and achieves at least 20 qualified leads. It should hold at 10% if it is within 50% of Google's cost per qualified lead and improving its lead-to-SQL rate. Otherwise, the test must be killed.

### How does the budget scaling work for ChatGPT after a successful initial test period?

If ChatGPT wins the initial test, its 10% share should be maintained, and then a 20% share should be tested for an additional 30 days. Budget reallocation must come from Google Ads, not from adding new funds.

### What kind of tracking setup is necessary for a valid cross-platform ad comparison?

It is necessary to implement shared identifiers and tracking methods, such as importing offline conversions into Google Ads and using UTM parameters and CRMs to link traffic in Google Analytics 4 to ChatGPT leads.

## Paid Ads & SEO, Run For You

Businesses pay one flat monthly fee per service, agencies resell it under their own brand, and groas does the work either way.

[See Pricing](https://groas.com/pricing)

## Structured data

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://graph.groas.com/entity/groas","name":"groas","url":"https://groas.com/","description":"groas is a done-for-you marketing service, not software: elite people working with proprietary technology systems run search ads, Meta ads, SEO and AI search for businesses, and white label for agencies, with a dedicated senior strategist owning every account.","logo":{"@type":"ImageObject","url":"https://groas.com/icon-512.png","width":512,"height":512},"sameAs":["https://www.linkedin.com/company/groas/"]},{"@type":"WebSite","@id":"https://groas.com/#website","name":"groas","url":"https://groas.com/","inLanguage":"en-US","publisher":{"@id":"https://graph.groas.com/entity/groas"}},{"@type":"WebPage","@id":"https://groas.com/post/take-10-of-your-google-ads-budget-and-give-it-to-chatgpt-for#webpage","url":"https://groas.com/post/take-10-of-your-google-ads-budget-and-give-it-to-chatgpt-for","name":"The 10% Rule: A 60-Day ChatGPT Ads Test Funded by Google","description":"Move 10% of your Google Ads budget to ChatGPT for 60 days. Use one qualified-lead scorecard and a kill rule agreed before launch to decide where the money goes next.","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/cef47934-4da8-4ff9-9a37-c9dadb2fad2c.png"},"breadcrumb":{"@id":"https://groas.com/post/take-10-of-your-google-ads-budget-and-give-it-to-chatgpt-for#breadcrumb"},"dateModified":"2026-11-01T06:14:48.656Z","mainEntity":{"@id":"https://groas.com/post/take-10-of-your-google-ads-budget-and-give-it-to-chatgpt-for#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/take-10-of-your-google-ads-budget-and-give-it-to-chatgpt-for#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://groas.com/"},{"@type":"ListItem","position":2,"name":"Blog","item":"https://groas.com/blog"},{"@type":"ListItem","position":3,"name":"The 10% Rule: A 60-Day ChatGPT Ads Test Funded by Google","item":"https://groas.com/post/take-10-of-your-google-ads-budget-and-give-it-to-chatgpt-for"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/take-10-of-your-google-ads-budget-and-give-it-to-chatgpt-for#article","url":"https://groas.com/post/take-10-of-your-google-ads-budget-and-give-it-to-chatgpt-for","mainEntityOfPage":{"@id":"https://groas.com/post/take-10-of-your-google-ads-budget-and-give-it-to-chatgpt-for#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"The 10% Rule: A 60-Day ChatGPT Ads Test Funded by Google","description":"Move 10% of your Google Ads budget to ChatGPT for 60 days. Use one qualified-lead scorecard and a kill rule agreed before launch to decide where the money goes next.","datePublished":"2026-11-01T06:14:48.597Z","dateModified":"2026-11-01T06:14:48.656Z","image":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/cef47934-4da8-4ff9-9a37-c9dadb2fad2c.png"},"author":{"@id":"https://groas.com/author/alexander-perelman#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"keywords":"google, chatgpt, ads, qualified, rule, test, budget, days","wordCount":2084,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/alexander-perelman#person","name":"Alexander Perelman","jobTitle":"Head Of Product @ groas","description":"My background includes Goldman Sachs and Stanford Computer Science.","image":"https://groas.com/media/blog/4f17cac3a81acc224cc1b2cfab8f94bbc295a3089d13e8421125771ef8d4064f.jpg","worksFor":{"@id":"https://graph.groas.com/entity/groas"},"sameAs":["https://www.linkedin.com/in/alexander-433793253/"]},{"@type":"FAQPage","@id":"https://groas.com/post/take-10-of-your-google-ads-budget-and-give-it-to-chatgpt-for#faq","mainEntity":[{"@type":"Question","name":"Does the 10% ChatGPT ad test require adding new budget, or can it be funded by reducing existing Google Ads spend?","acceptedAnswer":{"@type":"Answer","text":"The 10% allocation should be reallocated from a weak Google ad group instead of requiring additional spending. This method ensures that ChatGPT must earn its placement against displaced spend rather than gaining budget based on its incremental results."}},{"@type":"Question","name":"What criteria should be used to select a Google ad group to compete against the ChatGPT ads?","acceptedAnswer":{"@type":"Answer","text":"The chosen Google ad group must have at least 30 qualified conversions in the previous 60 days and exhibit a reasonably stable cost per qualified lead. A group with only a few leads or one unusually strong week should not be used as a baseline for comparison."}},{"@type":"Question","name":"How can lead quality and performance be compared between ChatGPT and Google Ads?","acceptedAnswer":{"@type":"Answer","text":"A single scorecard and identical qualification criteria must be used for both platforms. Performance should be measured using metrics like cost per qualified lead and lead-to-SQL rate, rather than relying solely on click-through rates or raw form submissions."}},{"@type":"Question","name":"How should performance levers be manipulated when running ChatGPT ad campaigns?","acceptedAnswer":{"@type":"Answer","text":"ChatGPT targeting does not rely on keyword lists or demographic targeting. Instead, the hands-on levers include adjusting context hints, testing headlines and images, and ensuring landing-page relevance matches the conversational nature of the visitor."}},{"@type":"Question","name":"What are the three conditions for determining the budget's future after the 60-day test concludes?","acceptedAnswer":{"@type":"Answer","text":"Budget should increase only if ChatGPT beats Google on cost per qualified lead and achieves at least 20 qualified leads. It should hold at 10% if it is within 50% of Google's cost per qualified lead and improving its lead-to-SQL rate. Otherwise, the test must be killed."}},{"@type":"Question","name":"How does the budget scaling work for ChatGPT after a successful initial test period?","acceptedAnswer":{"@type":"Answer","text":"If ChatGPT wins the initial test, its 10% share should be maintained, and then a 20% share should be tested for an additional 30 days. Budget reallocation must come from Google Ads, not from adding new funds."}},{"@type":"Question","name":"What kind of tracking setup is necessary for a valid cross-platform ad comparison?","acceptedAnswer":{"@type":"Answer","text":"It is necessary to implement shared identifiers and tracking methods, such as importing offline conversions into Google Ads and using UTM parameters and CRMs to link traffic in Google Analytics 4 to ChatGPT leads."}}]}]}
```