Last weekend I gave myself a challenge: find a real problem, build something that solves it, and ship it before Monday. Here's the honest story of what I built, why, and what I learned.
The Problem I Kept Seeing
Small businesses waste 25–40% of their Google Ads budget every month. Not because they're careless, because Google's dashboard makes it nearly impossible to know what's actually working.
You log in and see clicks. Impressions. CTR. A bunch of numbers that look like progress but don't answer the one question that matters: is this making me money?
I looked at the existing tools trying to solve this. Every single one WordStream, Opteo, TrueClicks, Adalysis, is built for agencies and PPC professionals managing $25K+/month. They're powerful, complex, and start at $130/month.
Nobody built one for the dentist or plumber spending $500/month who just needs one clear answer in plain English.
That's the gap I decided to fill.
The Idea: A Waste Score
I wanted one number. Like a credit score, but for your Google Ads account.
Connect your account → get a score out of 100 → see exactly where the waste is → get 3 specific things to fix today.
No jargon. No agency. No 47-page report.
The tool would check:
Keywords spending money for 90+ days with zero conversions
Whether conversion tracking is set up correctly (most accounts have this broken)
Budget concentration, is all your money in one campaign with no backup?
Search terms triggering your ads that have nothing to do with your business
Then Claude AI turns the raw data into plain English fixes a non-marketer can actually act on.
Saturday: Building the Core
I started Saturday morning with a blank Next.js project and one goal: get a real Google Ads account connected and show the waste data by end of day.
The hardest part wasn't the code. It was the Google Ads API approval process, Google requires a manual review of your developer token before you can pull real account data. I submitted the application and built everything else while waiting.
By Saturday evening I had:
Google OAuth working, users could connect their ads account in one click
The audit engine pulling campaign, keyword, and conversion data
Claude AI turning that raw data into a plain-English Waste Score and fixes
A basic report page showing the results
The Claude integration was genuinely the most satisfying part to build. Feeding messy spreadsheet-like API data in and getting back "Pause these 6 keywords, they've cost you $340 this month with zero sales" felt like magic.
Sunday: Polish and Launch
Sunday was about making it feel real.
I added Stripe so the free tier shows your Waste Score and top fix, and the paid tier ($29/month) unlocks the full report, negative keyword CSV export, and PDF download.
Then I built the landing page. The most important element was a demo report card, showing a sample audit with a real dollar amount of wasted spend. People need to see what they're getting before they'll connect their account.
By Sunday evening, adsiftly.com was live on Vercel.
I submitted it to ProductHunt, created an IndieHackers page, and wrote this article.
The One Thing Nobody Warned Me About
The Google Ads developer token takes days to approve.
You apply through your Google Ads Manager account, submit a form describing what you're building, and then wait. Google reviews it manually. There's no way to speed it up.
During this waiting period your token only has "Test Account Access", you can query test accounts but not real ones.
I built the entire product not knowing if the core API access would even be approved. That's a real risk for a weekend project.
If you're building anything on the Google Ads API: apply for the developer token on day zero, before you write a single line of code.
What I'd Do Differently
Ship the landing page first, build second.
I built the product then launched. Should have been: landing page → collect waitlist emails → validate demand → build. I could have tested the idea in 4 hours instead of 48.
One paid tier, not three.
I launched with Free / $39 / $149. The moment I added a third pricing tier I started second-guessing which features belong where. For a weekend launch, one paid tier is enough.
Mock data is your best friend.
Don't wait for OAuth and API connections to test your UI. Build with realistic fake data first, wire up the real API last. I lost hours waiting for things to connect when I could have been building.
Where It Stands Now
AdSiftly is live. The waitlist is open. The developer token is pending.
Revenue: $0 (being honest).
Learnings: a lot.
The real launch happens when Google approves the developer token and real users can connect real accounts. Until then, everything is validation, and validation is the most important thing at this stage.
The best side projects aren't the ones with the most features. They're the ones that solve one boring, specific, painful problem for one clear group of people.
Wasted Google Ads spend is boring. Specific. Painful. And almost every small business running ads has it.
That's enough to build on.
Try AdSiftly free at adsiftly.com — no credit card, takes 60 seconds.
If you've ever run Google Ads and felt confused by the results, this is built for you.
Top comments (0)