I Started Running TikTok Ads for My New AI Law Firm App. Here’s What Happened.

Google LLM Rundown Video

People assume TikTok ads are for junk products, impulse buys, or just people dancing around.

I didn’t want to believe that. Because I know that TikTok has one of the engaged audiences online right now. It’s fast traffic, and it feels like new-age TV with how long people spend on the app. And I wanted to test one thing.

How quickly could I test an idea for a new AI law SaaS product MVP.

Here’s the rundown of a new TikTok ads campaign I just started:

This just kicked off 6 days ago. I’d say that’s fast based on what I’m seeing now.

So I decided to test something that shouldn’t work on TikTok:
an AI tool for lawyers doing contract review.

Super specific. Extremely targeted. Like finding a needle in a haystack for people who might use this thing.

Not “AI for fun.”
Not “AI for marketing.”
Real contracts. Real consequences.

This is what I did, what worked, what didn’t, and what stood out to me almost immediately.


The real problem isn’t speed

Every AI conversation in legal tech right now is obsessed with speed.

Faster reviews.
Fewer hours.
More leverage.

That’s not the real problem.

The real problem is TRUST.

Do you trust AI? Or not?

Specifically:

“What happens when AI misses something… and I’m the one responsible for it?”

That’s the tension people are feeling in the field.
Not curiosity. Responsibility.

So instead of building “another AI contract tool,” I built something simpler:
a decision-support tool that helps lawyers sanity-check AI contract output before relying on it.

Not legal advice.
Not replacing judgment.
Just pressure-testing the AI’s work.

I vibe-coded the first version using ChatGPT and Lovable.

Here’s my Lovable join link. If you subscribe to Lovable too, I’ll earn a commission. That helps fund my next cup of coffee, next test, and more fun stuff here at AI for Law Firms!


This was not about perfection. More so spinning up a rapid MVP. Check out the tool I made here


How I set up the TikTok test

I wanted speed to insight, not a perfect funnel.

Because I’ve learned that you can have a great idea. You can imagine something cool others can use. But if you cannot get real people in the market to raise their hand and say, “Yes. I like this and I will use it.” Then even the best idea can flop.

You’ve got to get it in front of people ASAP.

I used TikTok Instant Forms

I didn’t send traffic to a website or even a landing page. There’s a LOT of friction in that. Plus you’ve got a lot of elements to test when you introduce a landing page in the funnel.

Instead, I used TikTok’s Instant Forms and then showed the tool on the thank-you page.

You can create an Instant form under Tools > Instant page. Here’s what mine looks like for this campaign:

Why?

  • Less friction – People can drop off at any point. The more steps you add, the more chance for drop off.
  • Faster feedback – Instant forms run right inside TikTok.
  • Lawyers don’t want to “explore a site”

Flow was simple:
Ad → Instant Form → Tool


I recorded five 1-minute videos

No script. Just cues.
No studio. Right in my office with a webcam.
No polish. I used a lot of my 1st takes.

I recorded them with Riverside and talked direct to the camera.

Here’s what my ads look like. I made the 9:16 (vertical aspect) but TikTok’s ad previewer in the backend shows them as wide view. Actual TikTok users see the 9:16 versions:

I also used Riverside’s caption feature. I like that because a lot of the time I have audio muted when I’m using TikTok. And videos with captions engage me. But videos without captions often confuse me without audio. Easy add here by just turning on captions.

I also used Picture-in-Picture with my webcam to save screen real estate. Another Riverside feature I just turned on.

Then I opened the tool I made. Demonstrated how it worked. And talked over that as I did it.

For the copy, I asked ChatGPT for a few hooks to get me started.

The goal wasn’t to explain features.
It was to call out a moment lawyers recognize instantly.

You know… present pain (not future imaginary would-be problems)

Here were the hooks.

Hooks that produced conversions

  • “Contract moment that keeps every lawyer up at night.”
  • “If you’re already using AI for contracts, you probably don’t need to learn AI.”
  • “Most lawyers don’t know this about AI.”

Hooks that didn’t convert

  • “Everybody’s talking about speed with AI for law firms, but what’s the trade-off?”

That difference mattered more than I expected.

It’s early but here’s some takeaways for the ad creatives so far:

  • Ultra Specific Audience: Calling out the specific audience right in the hook works. I do not want to talk to everybody. TikTok is a MASSIVE platform. So I didn’t want to risk engaging the wrong people. I’m disqualifying the wrong people right from jump street.
  • Present Pain vs Imaginary Problems: Bringing up the problems with trusting AI, right to the surface works. This is the present pain strategy I talked about before. Never ever talk about future problems/pains. You’ll waste every ad dollar trying to convince someone of an imaginary problem that could happen in the future. Instead, do your research. Dig in and figure out what people are already dealing with right now (today).
  • Getting to The Point: Weaker “Everybody’s talking about…” hooks failed. Probably because they break the rule I just mentioned about present pain. It’s watered down now that I’m analyzing the early results. Worth testing. But I won’t want to repeat this in future ad creatives I make.

Budget and setup

  • $30 per day
  • Cold audience
  • New ad account
  • No warming
  • No fancy targeting

This wasn’t about scale.
It was about validation.


The early data

With under $20 in spend, here’s what came back:

  • 2,646 impressions
  • 25 clicks
  • $0.74 CPC
  • $6.95 CPM
  • 12 Instant Form submissions
  • $1.53 per lead

For legal + AI + cold traffic, that’s not normal.

Especially this early.

And I’ve got early validation from a dozen people who raised their hands. For the price I spend on one hamburger at the local brewery.


What stood out immediately

1. Speed didn’t convert. Consequences did.

The “speed” angle sounded smart.

It didn’t move people.

The ads that talked about responsibility, risk, and that moment when something goes wrong consistently did better.

“Contract moment that keeps every lawyer up at night” outperformed everything else.

TikTok doesn’t reward clever.
It rewards engagement. What sticks people, what do they keep watching, and what do they tap on.


2. TikTok didn’t care that this was serious B2B

There’s this belief that TikTok can’t handle serious topics.

That’s not what I saw.

TikTok cares about:

  • a clear moment
  • a human face
  • something that’s real

3. Instant Forms are underrated for early SaaS validation

A lot of founders overbuild before they know if anyone cares.

Instant Forms let me:

  • capture intent fast
  • show value immediately
  • validate the idea

It felt more like starting a conversation than running ads.

Similar to if I had 100 in-market neighbors next door, and I could holler over the fence, “Hey, I’ve got this new thing I just built for you. What do you think? Yea or Ney?”


What didn’t work (and why I’m glad)

The “speed vs trade-off” angle didn’t convert at all.

My read:

  • It’s abstract
  • It sounds like thought leadership
  • It doesn’t hit a specific moment of fear or responsibility

TikTok punishes vague.
It rewards specificity.


What I’m testing next

This is early, but here’s what I’m moving into next:

  • Breaking ads down into very specific contract failure moments
  • Testing “missed clause” vs “hallucinated clause” framing
  • Segmenting by how lawyers already use AI
  • Following Instant Forms with email + deeper tool usage

I’ll keep documenting this as it unfolds if you want – likely based on feedback you send in


What’s next?

I know a lot of AI SaaS founders are stuck in building mode. I 100% believe you should be building something great. And it’s easy to stick your head under a basket. But validation with the market should be a priority.

Waiting for SEO.
Waiting for partnerships.
Waiting for the “right time.”

You don’t need that to start learning.

You need:

  • one real problem
  • one sharp moment
  • one fast feedback loop

That’s is this test was.

If you want me to break down:

  • the Instant Form setup
  • the hooks in more detail
  • how I’d scale this
  • or how I’d adapt this to another serious B2B niche

Just reach out.

This is early.