From flying blind to running the funnel like an engineer.

In six months, a health consulting business went from a sales lead in a Google Sheet to a self-updating CRM with AI lead scoring. Instagram went from a 10% revenue channel to 28%.

Client Prism Health
Industry Root-cause health consulting
Scope CRM, attribution, AI scoring, automation
Timeline Six months (Feb to Aug 2026)

Full disclosure.

Prism is our own consulting business. We built this system for ourselves first, because we needed it, and because the best way to prove infrastructure works is to run your own operation on it.

Everything below is real. The numbers, the failed experiments, the surprises we didn't see coming. This is what six months of running Prism on the same stack we sell looks like.

Before: a sales lead, a spreadsheet, and no visibility.

The entire sales operation lived in Google Sheets. Follow-ups were manual. Reminders were manual. Nobody knew which Instagram post drove a booking, which landing page actually closed, or which lead was worth the sales team's time.

Decisions were made on feel, because feel was all there was.

That's the situation most small businesses are in right now. Ad money goes out, leads come in, deals close (or don't), and the loop that connects them is invisible. You're running your business on vibes.

What we built.

A CRM built around the funnel instead of around contacts. Every booking arrives with the full context of where it came from. The exact Instagram post, the landing page variant, the ad, the referrer. Every touch after that gets logged in one place.

On top of that, an AI layer that reads each incoming lead against six months of historical data and scores it before the call happens: predicted show rate, predicted close rate, expected value. The sales lead sees the score, decides whether the call is worth taking, and gets a follow-up scheduled automatically. The spreadsheet is gone.

What changed.

Six months of running the funnel on real data instead of guesswork.

10% → 28%
Instagram's cut of total revenue, last year vs. last 60 days. Same audience, different funnel.
6% → 15%
Instagram close rate. A 2.5× improvement after five rounds of landing page iteration. Winner: version 3, a dedicated topic-led page.
→ 0 hrs
Weekly time spent on sales admin. Every follow-up and reminder gets scheduled the moment a lead is logged.

Four decisions we couldn't have made before.

Each one is a real choice the data surfaced. Most of them counter-intuitive.

1. Killed the Instagram DM funnel entirely.

Instagram DMs, the classic "comment PRISM and I'll send you a link" play, were showing a 36% show rate and 7% close rate. Meanwhile, the same audience going through the bio link was showing 57% and up.

The DM funnel felt like it was working because bookings were happening. The data said it was the worst channel we had. We turned it off.

2. Discovered landing pages don't always help.

On Instagram, adding a dedicated landing page above the calendar pushed close rate from 6% to 15% and expected value to $92 per booking. Huge win.

So we tried the same thing on Twitter, where traffic was already going straight to the calendar and closing well. Volume dropped, quality stayed the same. We removed the landing page and went back to the bare calendar.

Landing pages are not a universal upgrade. What matters is matching the funnel to the audience, and you can't do that without measuring both.

CRM analytics dashboard screenshot

3. Found the surprise signals in the data.

The AI scoring model surfaced something we would never have looked for: iCloud email addresses convert at meaningfully higher rates. Once you see it, it makes sense. But nobody sits down and thinks "let me check email domain as a lead quality signal." The system just noticed.

Same for content strategy. Analyzing the topics leads mentioned in intake forms revealed that weight loss, blood pressure, and hormones had the highest expected value. That's not what we would have guessed. The content calendar is now built around those three topics.

4. Turned ads from guesswork into a data problem.

You can't run Meta or Google ads without proper attribution. Without it, you're guessing at ROI, and you have no idea which ad is actually working.

The default Instagram play is to retarget people who engaged with your account. We didn't do that. Instead we uploaded our real client list (every person who had closed) and used it to seed a 1% lookalike audience. Then a second layer on top: everyone who had shown up to a call and been rated highly by the sales lead, another 1% lookalike tuned tighter.

Show rate on the most recent round hit 79%. Historically, ads pulled 40 to 45%. That's low enough that scaling spend just meant more bad leads clogging the calendar, which is why we hadn't. Feeding the algorithm real quality data changed that.

The point.

None of these decisions required a bigger budget, a growth consultant, or a new marketing playbook. They required the data being in one place, connected end-to-end, and readable by something that could pattern-match across it.

That's what we build. The same infrastructure that runs Prism (the CRM, the attribution, the AI scoring, the automated follow-ups) goes into every business we work with. The specifics change. The system doesn't.

See what your business looks like rebuilt.

Book a demo to see the stack in action, or start with a free audit that shows what's fixable and what AI could be doing for you in week one.

Book a demo Get a free audit