The Playbook That Took Us From $17M to $350M Wasn't Sophisticated. That's the Uncomfortable Part.

Before Massively Useful, I spent years on the growth side at Xometry. We went from roughly $17 million in revenue to north of $350 million, with a NASDAQ IPO somewhere in the middle.

People assume there was a secret in there. A channel nobody else knew about. A tool with a five-figure monthly license. Some kind of unfair advantage that only unlocks above a certain revenue line.

There wasn't.

What we had was visibility. We knew which dollars turned into customers and which ones didn't, and we knew it fast enough to do something about it. That was the whole advantage. The team, the software, the budget — all of it existed to serve that one capability.

Here's the uncomfortable part. Almost none of it required scale.

I've now spent enough time inside contracting businesses to say this plainly: most of what mattered most at $350 million would work, essentially unchanged, at $2 million. It doesn't get run because nobody hands a contractor the playbook, and because the playbook looks too boring to be worth anything.

So here it is.

1. Every lead gets a source and an outcome

The enterprise word for this is attribution, and at Xometry we spent real money building it. Multi-touch models, data warehouse, the works.

But here's what I learned taking it apart later: about 80% of the decision-making value came from the crudest possible version. Two fields. Where did this lead come from, and did it turn into money.

That's it. That's the version you can run.

Right now, ask yourself where your last twenty leads came from and how many of them became jobs. If the honest answer is "Google's working, I think," you're not running a marketing budget. You're making donations.

What to do Monday: one field in your CRM — or one column in a spreadsheet — captured on every inbound lead. Source in, outcome out. Review it monthly.

What it usually reveals: one channel producing 40% of your lead volume and 12% of your revenue. It's almost never the channel you'd guess.

2. Cost per lead is a vanity number. Cost per booked job is the real one.

Every agency in the trades quotes cost per lead, because cost per lead is the number that makes them look good.

Run the arithmetic yourself. Two channels, a hundred leads each:

• Channel A: $40 per lead. Books 8% of the time. You spent $4,000 and got 8 jobs. $500 per booked job.

• Channel B: $180 per lead. Books 55% of the time. You spent $18,000 and got 55 jobs. $327 per booked job.

Channel A is the "cheap" one. Channel A is 53% more expensive.

This is the single most common expensive mistake I see in home services, and it's not a sophistication problem. It's a division problem. Cheap leads are usually the expensive ones, because cheap leads are usually low-intent leads, and low-intent leads eat your CSR's day without ever becoming revenue.

What to do Monday: divide spend by booked jobs, not by leads, for every channel you pay for. Then decide again.

3. Break the black box into five stages

Most contractors have two numbers: leads and jobs. Everything in between is a black box, which means when the ratio gets worse, you have no idea why.

At scale, that's illegal. Every stage has a conversion rate and a name on it. Yours doesn't need to be complicated — five stages covers it:

1. Lead received

2. Lead contacted

3. Appointment set

4. Estimate delivered

5. Job won

Two shops can both run a 22% lead-to-job rate and have completely unrelated problems. One is losing at stage 2 because calls go to voicemail after 4pm. The other is losing at stage 5 because they're priced 15% over market with no story for why. Same headline number. Opposite fixes. And if all you track is the headline number, you'll spend a year fixing the wrong one.

Worth knowing: this isn't a small-business failing. Researchers behind a well-known Harvard Business Review audit checked 2,241 U.S. companies — funded companies, with marketing departments — on how fast they responded to an inbound lead. Twenty-three percent never responded at all. Among those that did respond within 30 days, the average response time was 42 hours.

That study is from 2011, so don't take my word for its shelf life. Run the test yourself: fill out the contact form on three of your competitors' websites this week and time the callbacks. Then have someone do the same to you.

What to do Monday: five numbers on a whiteboard. Update them weekly. Find the one that's bleeding.

4. A test needs a threshold and a kill date before it starts

We never turned on a channel at Xometry without writing down, in advance, what success looked like and the date we'd shut it off if we didn't hit it.

Not because we were unusually disciplined people. Because without it, every channel survives. Six months in, nobody can prove it's failing, everyone remembers the two good jobs it produced, and the spend rolls forward on sunk cost and hope.

I'll be direct about this one: I've violated my own rule. I've put real money into paid acquisition for this business that didn't convert. The failure wasn't the ads. The failure was that I hadn't written the exit before I wrote the check, so I stayed in longer than the data deserved.

What to do Monday: before the next dollar goes out, write two sentences and put them where you'll see them. "This works if it produces X booked jobs at under $Y per job by [date]. If it doesn't, we stop." Then actually stop.

5. Price is the fastest lever in the business, and almost nobody pulls it

Big companies staff entire teams around pricing. Not because pricing is intellectually difficult — because it's the highest-leverage number they own.

Take a $2M contractor at a 12% net margin. That's $240,000 in profit.

Raise prices 5%. Hold volume flat. Revenue goes to $2.1M, your costs don't move, and profit goes to $340,000.

That's a 42% increase in profit from a decision. Not a campaign. Not a hire. Not a new tool. A decision made on a Tuesday.

You won't hold volume perfectly flat, and I'm not going to pretend otherwise — you'll lose some price-shoppers, and the real question is whether the margin gain outruns the volume loss. But you can only answer that question if you have stage 5 from the section above, because that's the data that tells you whether you're actually losing on price or losing on something you've been blaming on price.

No marketing dollar you will ever spend has an ROI like this one.

What doesn't transfer

I want to be honest about the limits, because plenty of "enterprise lessons for small business" content isn't.

Some of it genuinely requires scale. Brand campaigns you can't measure for eighteen months. Statistically valid A/B testing — you need volume to get significance, and a contractor running 80 leads a month doesn't have it, which means most "we tested it" claims at your size are noise dressed as insight. Specialist headcount. Long-horizon category creation.

Skip all of that. It isn't yours yet, and behaving like it is will cost you money.

The visibility layer is different. The visibility layer is just arithmetic and the discipline to look at it, and it doesn't care how big you are.

Where to start

Not all five. One.

Start with #1, because the other four depend on it. You cannot compute cost per booked job, diagnose your funnel, judge a test, or make a defensible pricing decision if you can't say where your leads came from and what happened to them.

One field. Every lead. Thirty days.

Then come back and tell me what you found — I suspect it won't be what you expected.

Danny Chang is the founder of Massively Useful AI, where he builds revenue systems for home service contractors. He previously helped scale Xometry from $17M to $350M+ through its NASDAQ IPO. If you want a second set of eyes on your numbers, the free 45-minute Revenue Audit is exactly the exercise above, run on your business.

Source: James B. Oldroyd, Kristina McElheran, and David Elkington, "The Short Life of Online Sales Leads," Harvard Business Review, March 2011.