We Had 7 Marketing Plans. All of Them Came From ChatGPT.

We were about to launch ShiftFocus into the OKR market. I pulled seven of my organic marketing executives into a room and told them to come prepared.

Bring your plan to get us to 100,000 highly targeted visitors a month. No fluff. Real strategies.

Everyone showed up with their plan. And every single plan looked the same.

Publish 30 blog posts a month. Create cornerstone content. Shoot 20 videos. Build OKR templates and give them away for free.

Write how-to guides. Post on social media five times a week. I sat there listening to seven people present, one after the other, and it hit me around the third presentation.

These aren't seven plans. This is one plan, copied from ChatGPT, presented seven different ways.

Nobody admitted it. But come on. Seven people don't independently arrive at "publish 30 posts and create cornerstone content" unless they all typed the same prompt into the same tool the night before.

We spent three hours going through every plan. Then lunchtime came. We ate. Came back. Went through the plans again. Discussed timelines.

Assigned responsibilities. Left that room feeling like we had a real strategy.

Week one, we published five pieces of content. Added images. Optimized titles. Felt productive.

Week two, we published the five pieces but didn't do much else. The team was already drifting back to client work that actually pays us.

Week three, I realized nobody had mentioned the 100K visitor goal in days. Including me.

By the end of the month, the goal was dead. Not because anyone quit.

Because no one was available. The team that was supposed to grow our SaaS was the same team servicing clients who pay us every month.

You can't borrow people from revenue-generating work and expect them to build something new in their spare time. That was the first lesson.

When the AI Told Us the Truth

After the plan fell apart, I did something embarrassing. I uploaded all seven of our team's plans into Claude and asked it one question.

Is this realistic?

The answer was brutal.

This plan looks great on paper. But 100K visitors per month is only possible if you execute this at a much bigger scale with a dedicated team, and even then it will take a very long time.

The companies that achieved this kind of traffic were either VC-backed with massive budgets, or they were early entrants in a niche where nobody else was competing.

They had years of head start and full-time content teams that didn't have other jobs.

Then it said something that changed how I think about startup goals entirely.

Instead of trying to attract 100,000 strangers through content, reach out to 100 people directly on LinkedIn.

• Get 20 conversations.

• Get 10 appointments.

• Get 5 beta users.

• Grow from there.

That was it.

• No 30-blog-post strategy.

• No video production calendar.

• No cornerstone content architecture.

Just talk to people who might actually use your product. Get feedback. Iterate.

Seven executives. Three hours of planning. And an AI gave us a better objective in about 30 seconds. Not because the AI is smarter than my team.

Because the AI didn't have ego. It didn't need to impress anyone in a meeting room. It just looked at our resources, our team size, and our runway, and told us what was actually possible. 

The Problem With Startup Goals

Here is what I learned the hard way.

Startups set goals like they are already big companies. They read about how Google does OKRs.

They download free templates from Lattice or Betterworks. They set quarterly objectives with 90-day timelines and monthly check-ins.

But those templates are built for companies with 200 people and dedicated operations teams. When you are five people or ten people wearing multiple hats, a 90-day OKR cycle is a death sentence. You need feedback now.

Not in six months when your SEO might start working. Not in three months when your content might start ranking. You need to know this week whether your approach is working.

If you reach out to 30 people this week and nobody responds, that's immediate feedback. Your messaging is wrong or your audience is wrong. Fix it next week.

If you publish 100 blog posts and wait six months for traffic, you have no idea what's working until you've already spent half your runway. By then your spirit is dead and your team has mentally moved on.

That's why channels like SEO are terrible as a primary goal for early-stage startups. It is an additional channel. A long-term bet. But you cannot build your entire KPI around something that gives you zero feedback for months.

When it is a startup, cross-dependencies are expected. One person handles marketing and customer support.

Another handles product and sales. That's fine.

But the KPIs you track should still be simple enough that anyone can update them in five minutes on a weekly basis.

If a key result takes an hour to calculate or needs data from three different tools, you are tracking the wrong thing.

The Client Who Got It Right From Day One

Let me contrast the GPT planning disaster with a client who did the opposite.

One of our agency clients wanted to start what would become the biggest capital firm in Texas. When they hired us, they had goals everywhere.

• Reach 10,000 followers on Twitter.

• 10,000 on Facebook.

• Raise money for real estate.

• Raise money for private equity.

• Invest in e-commerce software.

They were trying to do everything because they were good at raising money and figured every investment would 5x or 10x.

None of that mattered to us as the marketing team. We said forget all of that. We are going to give you one number.

20 qualified appointments per week. Each investor ready to put in at least $100,000. That is your only KPI.

• Not followers. 

• Not impressions. 

• Not website visits.

Appointments with people holding real checkbooks.

They loved it. And here is why it worked. Every week when we had our check-in call, the data was already updated before the call even started.

• How many appointments booked?

• How many showed up?

• How many had real investment capacity?

• How many were just tire-kickers?

All of it visible.

• No guessing.

• No "we're working on it."

• No stretch goals that let anyone hide behind ambition.

The team knew the number. If we hit 20, the week was green.

If we hit 15, we knew something was off in our outreach and we fixed it the next week. No waiting three months to see if our strategy worked. We got feedback every seven days.

That client raised multiple millions in the last two years.

• Without burning money on massive ad campaigns. 

• Without 10K Twitter followers. 

• Without any of the vanity metrics they originally wanted to track. 

• Just appointments.

• Just investors.

• Just money in the door.

They are still using the same KPI we set on day one.

What This Taught Me About OKRs

I've sat in too many planning meetings where the goals sounded impressive but had nothing behind them.

My own meeting with seven GPT-generated plans is proof that even smart people create fantasy strategies when they're optimizing for how the plan sounds instead of whether the plan can actually be done.

The Texas client worked because we did the opposite. We didn't ask "what's ambitious?" We asked "what can we measure this week that directly leads to revenue?" Then we tracked that one thing and made the dashboard brutally simple.

If your key result hasn't moved in two weeks, that's a problem. Not a "we'll revisit it next quarter" kind of problem. An escalation kind of problem. The founder or the CEO needs to see it.

A red dashboard where someone checked in and said "we're stuck" is ten times better than a dashboard where nobody checked in at all. Because at least the red one tells you where to look.

At some point the red turns green. Not because you lowered the bar. Because you kept showing up, kept looking at the real number, and kept fixing what was broken.

That's how goals actually get achieved. Not through stretch targets and quarterly planning theater.

Don't download a free OKR template and fill it in. Don't ask ChatGPT for a growth plan and present it to your team like you wrote it at 2 AM.

Look at your actual team, your actual bandwidth, your actual product, and ask one question.

What can we measure this week that  proves we are making progress?

If you can't answer that, your OKR is decoration. And no software is going to save you from a goal that was never real.