(how we hit a $1.2M run rate) Week 9 of 10 - The $9,452 autopsy. Every dead ad, every number, every lesson I paid full price for.

Day 126. $125,000 ARR on the product. Over $100K/month across the whole business. 374 days left.

I've been promising this issue for a month. Here it is: every ad dollar I've spent on this challenge, laid out on the table, cause of death included.

Total product ad spend to date: $9,452.

Most founders will never show you this, because ad spend is where the smart-guy costume comes off. You're about to see me pay $456 for a single lead, pay $750 for literally nothing, and then, one change later, get the same leads for $171. The lessons are all here, priced at full retail, so you can have them for free.

This is Week 9 of the roadmap: paid ads, last on purpose. Now you'll see exactly WHY they're last.

Four campaigns. Here's each one, what it spent, what it produced, and what killed it.

Campaign 1: "High-Ticket Verticals" (Meta). $4,560.85. 10 leads. $456 per lead. My first real campaign, launched back when I believed my customer was a lawyer. Every mistake at once: targeting four countries (a chunk of this money went to UK and Australia clicks I could never serve), budget pooled so the algorithm fed the worst performers, and ad copy that listed FEATURES to people who don't buy features. This campaign ran while I was still in the wrong lane, and the $456 is what the wrong lane costs, per lead, with receipts.

Campaign 2: Shopify App Store ads. $750. 3 installs. 0 customers. The purest tuition I ever paid. 21,000 competing apps, 2,700 new ones a month, and me showing up as option 14,001 with a features list. Killed with zero regrets. It's also the campaign that triggered the entire brand-promise awakening from the early weeks, so maybe it was the most valuable failure on this list.

Campaign 3: "Problems v2" Canada (Meta). $3,593.37. 21 leads. $171 per lead. Same product as Campaign 1. Same platform. Same me. Less than half the cost per lead. What changed? Everything Week 2 taught: one country instead of four, the receptionist lane instead of the software lane, and copy that led with the PROBLEM (the missed call, the money bleeding at 2am) instead of the feature list. The campaign is literally named "Problems v2" because the whole fix was leading with problems. Position first, then advertise. Advertise first and you're just paying to broadcast the wrong message.

Campaign 4: "Problems v2" United States (Meta). $548.47. 1 lead. $548 per lead. The humbling one. The EXACT ads that produced $171 leads in Canada produced one $548 lead in the US. Same creative, same copy, same product. Different market, triple the auction prices, different result. Geography is a variable, not a detail. Test one market at a time, and never assume a winner travels.

Blended: $9,452 for 32 Meta leads plus 3 Shopify installs. Some of those leads became real customers. Most didn't. Honest verdict: my ads have been tuition, roughly break-even at their very best, and the septic tank company that found me for free is still my favorite customer acquisition story.

Rule 1: Write your kill rules BEFORE you spend a dollar. Mine now: any ad set that burns $150 with zero leads gets paused, no negotiation, no "one more day." Any creative with frequency above 3.0 is fatigued and gets swapped. Decisions made in advance are cheap. Decisions made while watching your money burn are always, always too slow. I held dying ad sets for weeks because the LIFETIME average looked okay while the last seven days screamed. Judge the last 7 days. Lifetime numbers are where dead campaigns hide.

Rule 2: Ads amplify. They don't fix. The v1 campaign didn't fail because Meta is bad. It failed because it amplified a bad message. When the message got fixed (Week 2), the SAME platform delivered leads at a third of the cost. If your organic funnel doesn't convert, ads just help you lose money at scale.

Rule 3: Ads are a research lab wearing a sales costume. Here's the one genuinely great thing ads did for me: the audience segments revealed WHO responds. The collections angle, the situation-based customer profile, which pains pull clicks. $50 a day bought me market data faster than any survey. If you run early ads, run them to LEARN, and consider every lead a bonus.

Rule 4: The funnel behind the ad decides everything. A $171 lead that no-shows the demo is $171 of nothing. Ads only started making sense for me AFTER Week 8's machine existed: the confirmation sequences, the briefs, the follow-ups. Ads are the last channel you turn on because they're the first channel to punish an unfinished funnel.

So after all that, did I quit ads? Not exactly. I did something weirder.

I turned most of the Meta machine off, and I moved real money to the newest ad surface that exists: ads inside ChatGPT.

Think about why. Back in Week 3 I showed you that ChatGPT fetched my website 579 times in one month to answer real people's questions. My customers aren't just scrolling feeds anymore. They're ASKING. And an ad that appears at the moment someone is asking about your exact problem is a different animal from an ad that interrupts someone's doomscroll. Meta sells interruption. AI surfaces sit inside the question itself.

Is it working? Too early for honest numbers, and this newsletter only prints honest numbers. I've spent enough there to get real data, and you'll get the full report, cost per lead and all, the same way you got today's autopsy. If it flops, you'll see that too.

The rest of the freed-up budget went to outbound across social platforms, the Week 4 and Week 5 systems running at full volume. After nine weeks of building owned channels and outreach machines, renting attention finally became optional. That was the plan all along: ads last, and only where the customer is already asking.

Now the twist, because this week handed me the perfect counterpoint to $9,452 of paid education.

We've spent months polishing Preeya's presence in meetings. Her greeting. Her timing. How her name appears as a participant. We agonized over that stuff. And on customer calls this month, watching real people react, here's what I learned:

Nobody cares.

They're mildly impressed she's there, and it makes zero difference to their decision. You know what actually made their eyes light up? Three unglamorous things: the call was recorded, they could ask questions against a live transcript, and a clean summary showed up afterward. That's it. That's the product, as far as the customer is concerned. The details we obsessed over, the theater of it, the perfect participant name, might as well not exist.

I paid $9,452 for ads to teach me who my customer is. Then a handful of honest customer conversations taught me what they actually value, for free.

We're acting on it, too. That insight is now steering a mobile app we've started building, centered on exactly what customers said they wanted: recordings, transcripts you can interrogate, summaries that show up without asking. Not the features we thought were impressive. The ones they told us they'd pay for.

Talk to your customers. It's the highest-ROI marketing channel on this entire roadmap, and it's the only one with no ad account.

1. If you run ads: pull your last 7 days, not lifetime. Check frequency on every creative. Apply the $150 kill rule to anything at zero results. Do it today, not Friday.

2. If you don't run ads yet: write your kill rules NOW, in a doc, before the first dollar. You're writing them as a calm person on behalf of a future panicked one.

3. Everyone: book three calls with existing customers this week and ask three questions. What almost stopped you from buying? What actually made you buy? What do you actually use? Their answers are worth more than my entire $9,452, and yours are free.

Next Wednesday: the finale. Week 10, the Compounding Machine. How all nine channels feed each other, what I'd do differently starting from zero, and the plan from $1M to $10M. Bring coffee.

Day: 126 of 500

Product ARR: $125,000

Combined business: $100K+/month run rate

Total product ad spend: $9,452 (autopsied above)

Ad budget now: Most Meta campaigns off. Testing ChatGPT ads (report coming), with the rest moved to outbound across social platforms.

Team: Hired our SECOND salesperson. The layers from Week 8 are getting deeper.

New experiment: Built a sell sheet and started conversations with local US distributors. If software can be sold like a physical product through people who already have the relationships, that's a channel nobody in my space is using. Early days. Full report when there's data.

In the lab: A mobile app built around what customers actually valued: recordings, live transcripts, summaries.

Biggest realization this week: Paid ads taught me who my customers are for $9,452. Customer conversations taught me what they value for free. Do the free one first.

Hit reply with your worst ad number. Cost per lead, cost per install, money spent on a campaign that produced absolute silence. Mine's $548 a lead, and it's printed above forever. Let's make this a support group.

And if you know a founder about to launch their first ad campaign this week, send them this issue BEFORE they spend. This is the $9,452 heads-up I never got.

See you next Wednesday for the finale.

Dhiraj