If you still own sales at your MSP, you probably already have a 40-plus-hour work week before you open Claude, ChatGPT, or another AI tool and start building something.
So if you’re going to work another 10, 20, or 80 hours this month, ask a basic question: Where will those hours make the most money?
For an MSP owner who still carries the sales number, the answer is usually sales.
Prospect. Run discovery calls. Follow up on open opportunities. Ask for referrals. Review the pipeline. Create the marketing that supports those conversations.
Without revenue, all the AI tools in the world won’t matter.
The expensive part of free AI is your time
AI makes building feel cheap because the software cost can be close to zero. You can ask Claude to help you build an agent, connect a few tools, automate a prospecting task, or create a marketing workflow that would have required a developer just a few years ago.
That’s exactly why it can become a distraction.
An MSP owner can lose a Saturday getting something to work, then another evening fixing it, and another afternoon improving it because the first version isn’t quite good enough. None of those hours show up as a software expense, but they still cost you something.
If you spent 80 hours last month experimenting with AI sales and marketing tools, what did those 80 hours replace? If that same time could have produced two new MRR contracts, the project wasn’t free. It was extraordinarily expensive.
I’m not arguing that MSP owners shouldn’t learn AI. I’m arguing that owner time has a job. Until sales has been successfully handed to someone else, part of that job is producing revenue.
Being able to build it doesn’t make building it a good business decision
Technical founders are particularly vulnerable to this problem because building things is fun.
Cold calling isn’t always fun. The seventh follow-up isn’t particularly interesting either. A stale pipeline review doesn’t deliver the same dopamine hit as watching a new agent complete a task correctly for the first time.
But the boring work is often what gets the contract signed.
West McDonald of goWest.ai gave me a useful example when we talked about how quickly AI development economics are changing. His team evaluated three initiatives for an MSP partner. One was going to cost about $110,000 to build. The other two were closer to $5,000 and $8,000.
His recommendation on the expensive project was simple:
“Don’t even look at that first one. Let’s revisit it in six months.”
The reason was that the tools were moving quickly enough that the same problem could become dramatically cheaper to solve later.
That same logic applies to your sales and marketing efforts.
Before you spend a weekend building an AI prospecting agent, check whether a prospecting tool already does what you need. Before you connect five applications to create a marketing workflow, look at software that already handles the workflow. Before you build a custom sales assistant, check whether one already exists.
You do not get a prize for owning more code.
Compare the build against the sales activity it displaces
MSP owners tend to compare a software subscription against zero.
A $300 monthly tool looks expensive next to a homegrown workflow that uses software you already pay for. That’s the wrong comparison if the homegrown version requires 10 owner hours every month.
Instead, compare the subscription against what those 10 hours could produce elsewhere.
If you’re still responsible for new-logo sales, put a dollar value on your time before starting any sales or marketing AI project. Then ask:
- How many hours will the first version take?
- How many hours will I spend maintaining it every month?
- Does a tool already solve most of this problem?
- What selling activity will I stop doing while I build it?
- If the tool works perfectly, will it produce more revenue than I could have produced using those hours to sell?
That last question kills a lot of interesting projects.
It should.
Buy before you build, especially around sales
A real sales system contains much more than the clever feature that caught your attention.
Prospecting involves targeting, data collection, enrichment, and workflow. Sales includes qualification, follow-up, pipeline management, calling, email, meetings, and account history. Managers need visibility into what reps are doing and where deals are getting stuck. Someone also has to maintain all of it.
When you build one AI function yourself, you inherit the work around that function too.
Sometimes that still makes sense. There are narrow tasks where a simple AI workflow can save time without becoming another system you have to own.
But an MSP with an owner-led sales function should have a much higher bar for projects that consume nights and weekends.
West gave another piece of advice in the same conversation that applies here: “If you wanna know the best way to use AI in your business, start with the work.”
For an MSP owner who hasn’t offloaded sales yet, some of the most valuable work is painfully obvious.
Call the prospect. Run the discovery meeting. Follow up. Ask for the business.
If you want help building a sales process that keeps your time focused on revenue-producing work, schedule a call with Carrie Richardson.
Photo: Formatoriginal / Shutterstock
