I Built a Tool This Week (And What It Taught Me About the Future of This Work)
Last week I wrote about how I’m actually using AI in Google Ads. The workflows. The sparring partner concept. The speed gains.
This week I want to zoom out a bit.
Because something happened that made me rethink what my job actually is.
I built a tool.
Not a spreadsheet. Not a dashboard. A real, working tool that anyone can use. Upload your product feed, get a health score out of 100, see exactly what’s broken and how to fix it.
I shipped it in a day.
And that single day taught me more about where this industry is heading than the last six months of running ads.
The problem I kept seeing
Here’s what happens in almost every e-commerce account I audit.
The campaigns look fine. The bidding is reasonable. The structure makes sense.
But the product feed is a mess.
Missing descriptions. Titles that say “Blue Shirt XL” instead of something a human would actually search for. No brand names. No GTINs. Categories that don’t match what Google expects.
And the brand has no idea. Because nobody looks at the feed. The agency manages campaigns. The dev team manages the website. The feed sits in between, and nobody owns it.
I’ve had this conversation so many times. “Your campaigns are fine. Your feed is the problem.” And then I’d spend 30 minutes explaining what a product feed even is.
So I thought: what if I just built something that shows them?
What I actually built
The tool is simple. You upload your Merchant Center feed (XML, CSV, TSV, whatever), and it scans every product against 20+ validation rules.
Title length. Description quality. Missing images. GTIN coverage. Category mapping. Price formatting. All the things Google cares about but most brands never check.
It gives you a score out of 100 and a breakdown by category - so you can see that your “clothing” products score 82 but your “accessories” score 41 because half of them are missing descriptions.
The interesting part? I wired up AI to rewrite the worst-scoring products. Feed it your 10 worst titles and descriptions, and it rewrites them following Google’s best practices.
I’m not charging for the scan. The basic report is free. Because the value isn’t in telling someone their feed scores 63. The value is in what happens next.
Why I’m telling you this
In my first newsletter, I said I’d share what I’m seeing. Real stuff. Not frameworks for the sake of frameworks.
So here’s what I’m seeing.
The line between “running ads” and “building tools” is disappearing.
Two years ago, if I wanted to build something like this feed scanner, I’d need a developer. A designer. Weeks of back and forth. Budget. Project management.
This week, I described what I wanted, iterated on it, and shipped it. In one sitting.
I’m not a developer. I can read code, kind of. I understand how APIs work, mostly. But I didn’t write this thing line by line. I worked with AI to build it - describing what I needed, testing it, fixing issues, pushing it live.
And that changes everything about what a small agency can do.
The shift I didn’t expect
When I started eCom Ads, the value proposition was clear. We run your Google Ads. We’re good at it. We charge a monthly fee.
That’s still true. But it’s not the whole picture anymore.
What I realized this week is that the most valuable thing I can do for a prospect isn’t run their ads. It’s show them what’s actually going on.
A free tool that reveals their feed quality does more for trust-building than any case study or sales deck ever could. Because it’s not me telling them they have a problem. It’s their own data telling them.
And once they see the score, they either fix it themselves (great, I’ve helped someone) or they realize they need help (and now they trust the person who showed them the problem, not the person who just claimed they could fix it).
This is the same logic behind the free audit I mentioned last week. Give away the diagnosis. Let the treatment sell itself.
What product feeds taught me about Google Ads
Here’s something I didn’t fully appreciate until I built the scanner and started running real feeds through it.
The average feed quality is terrible.
I’m not talking about small brands with 50 products. I’m talking about brands doing millions in revenue with thousands of SKUs. And their feeds look like they were set up once in 2019 and never touched again.
Missing descriptions on 40% of products. Generic titles that say nothing about the product. No brand field. No color, no size, no material. Just the bare minimum to get the feed approved.
And these are the same brands wondering why their Shopping campaigns underperform.
Google’s algorithm can only work with what you give it. If your product title is “Sneaker White 42” and your competitor’s title is “Nike Air Max 90 Men’s Running Shoe - White/Black - Size 42,” who do you think shows up first?
It’s not a bidding problem. It’s a data problem.
And most agencies aren’t looking at it. Because feed optimization isn’t sexy. It doesn’t show up in monthly reports. Nobody brags about fixing title structures on LinkedIn.
But in my experience, it’s one of the highest-leverage things you can do for Shopping performance. Boring and high-impact. My favorite combination.
The AI part (keeping it honest)
I said last week that AI is a sparring partner, not a replacement. I meant it. And building this tool reinforced it.
The AI-powered rewrites are genuinely good. Feed it a bad title and a list of issues, and it’ll write something that follows every best practice Google recommends. Proper structure. Front-loaded keywords. Right character length.
But here’s the thing.
It doesn’t know your brand voice. It doesn’t know that you call your products “kicks” not “sneakers.” It doesn’t know that your German market descriptions should sound different from your UK ones. It doesn’t know that one product line is being discontinued so you shouldn’t invest in optimizing those titles.
The AI writes a solid B+ title. Getting to A+ still requires a human who understands the business.
Same as last week. Same as it’ll probably be next week. The pattern keeps repeating: AI handles the 80% that used to take forever. The 20% that actually matters still needs you.
Why I keep coming back to tools
This is the reflection part. The thing I’ve been chewing on all week.
I’ve always loved building systems more than doing repetitive work. That’s probably the ADHD talking. The novelty of creation is more stimulating than the grind of execution.
But it’s more than that.
When you run someone’s ads, you’re doing work for one client. When you build a tool, you’re solving a problem for everyone who has it.
One scales linearly. The other doesn’t.
I’m not saying I’m going to stop managing ads. That’s still the core of what eCom Ads does. But I am saying that the tools we build around our services - the scanners, the audits, the diagnostic frameworks - those might end up being more valuable than the services themselves.
Not because they replace expertise. Because they make expertise accessible.
A brand in Dublin who’ll never hire us can still run their feed through our tool, see the issues, and fix them. And if they can’t fix them alone, they know who to call.
That’s the flywheel I want to build. Tools that create trust. Trust that creates conversations. Conversations that create clients. Clients that generate insights. Insights that create better tools.
What I learned this week (the honest version)
Some quick things, because I said I’d be transparent about what works and what doesn’t.
What worked:
Building a useful tool fast by working with AI instead of trying to code everything manually
Making the basic scan free with no email gate - just value, no friction
Adding AI rewrites as a premium layer instead of charging for the whole thing
Connecting directly to the Merchant Center API instead of relying on third-party tools
What I’m still figuring out:
How to price the AI rewrite feature (free sample of 10, then what?)
Whether to gate the PDF report behind email or keep it open
How to make this work for brands that don’t know how to export their own feed
Whether anyone actually cares about feed quality enough to use a tool like this
That last one is real. I might have built something nobody uses. That’s fine. The worst case is I learned a bunch about feed optimization patterns and AI-powered rewriting. The best case is it becomes a meaningful lead generation tool that helps brands and brings in clients.
I’ll know in a few weeks. I’ll tell you either way.
Where this connects
If you read the first edition, you know the promise. Real insights from real work. Not tips. Not theory.
This week, the real work was building something. Next week it might be a campaign restructure, or a client conversation that changed how I think about attribution, or an AI experiment that flopped.
I don’t know yet. That’s the point.
I’m not writing from a content calendar. I’m writing from whatever’s actually happening. And this week, what happened was that I built a tool in a day that would have taken a team weeks to build two years ago.
If that doesn’t make you think about what’s possible - whether you’re an agency, a freelancer, or a brand doing this in-house - I don’t know what will.
The tools are here. The question is what you build with them.
If you want to try the feed scanner yourself, it’s at tools.ecom-ads.com. Upload your feed. See what comes back. Let me know if the AI rewrites are helpful or terrible.
And if you’re building things with AI in your own work - whether it’s marketing tools, internal processes, or something completely different - I’d genuinely like to hear about it. Reply to this email. I read everything.
- Raoul