I had finally won the first grant for the Finn Wardman World Explorer Fund, the charity we run in memory of my son. Google Ad Grants. Ten thousand dollars a month in free Google Search ads, every month, for as long as we keep the account healthy. I was really excited, because it sounded like a lot of money, and I figured I could use AI to write the ads, pick the keywords and have solid traffic going within a week.
Thirty days of real numbers later, the traffic is telling a different story. The ads have spent $10.89 of the ten thousand. They have sent us zero applicants. Over the same four weeks, half of our grant applicants came from ChatGPT, young people asking an AI how to pay for a trip they cannot afford and being pointed at us.
How we got the grant in the first place
I wrote about the application back in June, the day I found out Google gives registered nonprofits this money at all. The short version for anyone starting now: you need registered charity status (we are a US 501(c)(3), which took two years of flailing to get), a Google for Nonprofits account, and a website that passes Google's website policy.
That last part is where we failed the first time. Google rejected the activation in July. Their reason was that the site did not load quickly, did not have clear navigation, and did not have substantial, up to date content and calls to action. I looked at the site in my browser and it looked fine. The problem was that Google's reviewer was not seeing what I was seeing. Our site host was serving old cached copies of half the pages to crawlers. One still had a "Donations coming soon, we are not yet a registered charity" pop-up, sitting right above the line saying we were a registered 501(c)(3). Another had a dead Donate link and an old address.
We fixed every page at the source, forced them to re-render, checked them the way a crawler sees them rather than the way I see them, and reapplied. The account went live on the third of August. That was a solid day or two of work just getting the site ready. I have a web background, but AI was doing most of the heavy lifting on that.
What a machine reads on your website is not always what you read on it.
Five weeks of nothing
The ads did not show once for the first five weeks. Zero impressions. I had AI write the campaigns, keywords and ads, the account said everything was eligible, and nothing happened.
The status line on both campaigns read "Eligible (Limited): your targeted goal is missing a primary conversion action." We chased it for days. I was sure the signup event had never been imported. The change history said it had. Then I was sure it was a stray account goal. We removed it and nothing changed. Two confident wrong answers, both written down in the log as wrong, which is the only good thing I can say about them.
The real cause was a loop. The campaigns were set to Maximise Conversions, which means Google bids to get you signups. Our conversion was imported from Google Analytics, and an imported conversion only counts as set up once it has fired at least once from an ad click. No clicks meant no conversion. No conversion meant nothing for the bidding to aim at. Nothing to aim at meant no ads, which meant no clicks. Round and round, for five weeks, with ten thousand dollars a month sitting there unspent.
The fix was two changes. We switched both campaigns to Maximise Clicks with the two dollar cap Google allows on grant accounts, and they started serving the same afternoon. Then we added a native Google Ads conversion tag on the application page, which Google can see directly without waiting for a click, so the account can move back to conversion bidding once it has enough data.
What the ads actually did
Here are the last thirty days, read off the live account:
| Campaign | Impressions | Clicks | CTR | Cost | |---|---:|---:|---:|---:| | Find Your Dream | 33 | 2 | 6.06% | $7.25 | | Vipassana Reset | 39 | 2 | 5.13% | $3.64 | | Finn's Story | 0 | 0 | | $0 | | Total | 72 | 4 | 5.56% | $10.89 |
The click-through rate is fine. Google wants grant accounts above five percent and we are there. Volume is the problem. Seventy two people saw an ad in a month.
The search terms report explains some of it. The people seeing our ads were searching for school sports grants, youth football funding, and Vipassana centres in Alberta and Ontario. Parents looking for money for their kid's team. Meditators looking for a centre near home. None of them are an eighteen to twenty six year old with a plan to go somewhere. We added a dozen negative keywords to stop those searches and paused a keyword with a Quality Score of one out of ten.
The people we fund do not search Google for "travel grant for young adults." There is almost no volume on the terms that describe what we do. Fourteen of our first twenty three keywords were flagged low search volume and never ran at all. Of course, every charity will have its own search terms and keywords that only it will be familiar with. It's not a perfect science, there is a lot of experimenting and checking, so don't be discouraged if your results look like mine. I am not, and I plan on making Google Ads, through this grant, a solid source of traffic. It's just not there yet.
Where the applicants actually came from
Google Analytics tracks every grant application as an event, and it records where the person came from. Over the same four weeks there were six real applications, plus one test we ran ourselves. The sources:
| Source | Applications | |---|---:| | ChatGPT | 3 | | Direct | 1 | | Instagram | 1 | | Google organic search | 1 | | Google Ads | 0 |
Half of the real applicants came from ChatGPT. The ChatGPT visitors also stayed, over a minute and a half on average, and one in six of them went on to apply. The paid ad clicks averaged zero seconds, which mostly means bots and Google's own ad reviewers.
Think about that, 1 in 6 are converting. That's showing real demand, so I am definitely going to lean into that channel.
Nobody asks an AI "grant for travel." They ask something closer to "I'm 22, I want to spend three months learning to surf in Portugal, I can't afford it, is there any way to get funded?" The AI does the searching, reads a few sites, and answers. If your site is one it can read and trust, you are in the answer.
What we changed for AI search
None of this is expensive. Most of it is an afternoon of work with Claude doing the typing. If you need help with this, it's what I do, and I have a permanent 50% off retainer for nonprofits. And chances are good that if you have a simple question, you can reach out to me and I'll answer it at no cost.
We added an llms.txt file, a plain text summary at the root of the site that tells AI tools what the fund is, who it funds, how much, and where to apply.
We checked robots.txt and made sure it explicitly allows the AI crawlers: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot and PerplexityBot. A surprising number of sites block them by default without knowing.
We added FAQ schema, structured question and answer markup, to the home page, the grant page and the application pages. It is invisible to a person and very easy for a machine to read.
We wrote a long page answering the questions applicants actually ask, with eleven real examples of what we have funded and a table of what we will not fund. The "will not fund" part matters as much as the rest. When an AI knows exactly who you are not for, it gets more confident recommending you to the people you are for. You can see ours live: What types of applications we fund and don't fund. Scroll to the bottom for the twenty question FAQ, which is also marked up as FAQ schema for the machines.
We set up Bing Webmaster Tools and submitted the sitemap. ChatGPT's search leans on Bing's index, and a lot of nonprofits have never opened Bing at all. We had not either.
And we fixed the application form. It already asked every applicant "How did you hear about us?" The form handler behind it was quietly dropping the answer. Now every application writes the answer, the source, the landing page and the referrer into the sheet, so next month I will not need to match timestamps across time zones to work out who came from where.
The machines are reading your site, and in our case they were sending better applicants than the ads.
Why demand comes before donors
Every charity's first question is money. The fund gives grants of five to ten thousand dollars, and to give money away you first have to take money in. I have spent a lot of mornings on grants that fund the grant-maker, and the temptation with ten thousand dollars of free ads was to point them all at a donate button.
I think that is the wrong order, at least for us.
Donors give to a problem they can see being solved. A brand new fund with a donate button and no stories is asking strangers to trust a plan. A fund with a steady queue of young people applying, a handful of funded trips, and those young people coming back with photos and a story is showing them something already working.
So our order is demand first. Build the pipeline of applicants. Fund the best of them. Get the stories. Then go out and find donors, and use AI to find them in the same way it is now finding us applicants: by being the clear, readable, trusted answer when someone asks where to give to young people with a plan and no way to pay for it.
The ad grant still has a job in that. It keeps the fund visible, it builds history on the account, and once conversion bidding kicks back in it may well start sending applicants too. But I am not building the plan around it anymore.
Donations will not be easy. Nothing about a small charity is. But I think it becomes sustainable once the demand is visible.
FAQ: Google Ad Grants and AI search for small charities
What is Google Ad Grants? A Google programme that gives eligible registered nonprofits up to $10,000 a month in free Google Search ads. It is text ads on Google Search only, pointing at your own website.
Who qualifies for Google Ad Grants? A registered charity in a supported country (in the US, a 501(c)(3)), with a Google for Nonprofits account and a website that meets Google's website policy. Government bodies, hospitals, schools and universities are excluded and have their own programmes.
Why was our Google Ad Grants application rejected? The most common reason is the website. Google wants it to load fast, have clear navigation, substantial and current content, and clear calls to action. Check your site the way a crawler sees it, not in your own browser. Cached old versions of pages were what sank our first application.
Why is my Google Ad Grants account not spending? Most often it is the loop we hit. The campaigns are on Maximise Conversions, the conversion has never fired from an ad click, so Google has nothing to bid toward and does not show your ads. Switch to Maximise Clicks with the $2 cap to get serving, add a native Google Ads conversion tag, and switch back once you have roughly fifteen conversions in thirty days. The second common cause is keywords with low search volume, which never serve at all.
What does "Eligible (Limited): your targeted goal is missing a primary conversion action" mean? In our case, that the primary conversion action existed and was set correctly, but had never recorded a conversion through Ads. A Google Analytics import cannot verify itself until a real ad click converts. A native Google Ads tag verifies once Google sees it on the page.
What is the 5% click-through rule? Grant accounts need to keep account-wide click-through rate at or above five percent. Two consecutive months under it and the account can be suspended. Pause keywords under that rate once they have enough impressions to judge, and use negative keywords to block the searches that are not for you.
Do grant accounts need a conversion every month? Newer accounts are expected to record at least one conversion a month and to use conversion-based bidding. Pick the easiest honest conversion you have, like an email signup, not a donation.
Should I turn on AI Max or broad match to get more traffic? Google will email you suggesting it. We did not. Broad match on a grant account spends your click-through rate on searches you did not ask for, and that is how accounts get suspended.
Is Google Ad Grants worth it? Yes, but not the way I expected. It costs nothing, it keeps your charity visible, and it teaches you a lot about how people search. It is not, on its own, a pipeline. Treat it as one channel and measure it against the others.
How do I know if ChatGPT is sending traffic to my charity's website? In Google Analytics, look at traffic acquisition by session source. ChatGPT shows up as chatgpt.com. Then set applying or signing up as a key event and look at key events by source. That is how we found out half our applicants came from it.
How do I get my charity recommended by ChatGPT? Make it easy for an AI to read and trust you. Allow the AI crawlers in robots.txt. Add an llms.txt file. Add FAQ schema to your key pages. Write plain answers to the questions your people actually ask, including who you do not help. Get indexed in Bing. Keep your facts identical on every page, because contradictions (we had two different age ranges and two addresses) make an AI less sure of you.
What is llms.txt? A plain text file at the root of your site, like robots.txt, that summarises who you are and links to your key pages in a form AI tools can read quickly. Ours says what we fund, the age range, the grant sizes and how to apply.
Should a new charity chase donors or users first? Every charity is different. For a grant-maker like us, applicants first. Their stories are what make donors say yes.
Can AI help find donors too? That is our next piece of work, and I think it runs the same way: be the clearest, most trustworthy answer when someone asks an AI where to give. I will write it up once we have done it rather than before.
If you run a small charity and want help setting any of this up, I have done all of it on our own fund first. I work with youth nonprofits at a permanent fifty percent off, and this is the same setup I would build for you, the marketing end of the nonprofit money stack. And if you are eighteen to twenty six with a plan and no way to pay for it, you know where to apply.
Day 180 of 365.