Last month a roofing company owner in Gastonia asked me the question I hear every single week. "Greyson, do I need an agency, a consultant, or should I just have my nephew figure it out?" He was not being cheap. He genuinely could not tell what each one sells, because all three use the same words on their websites.
So here is the plain version. A consultant sells judgment by the hour. An agency sells an ongoing engine on retainer. DIY costs your weekends. Pick the wrong one and you will spend good money solving the wrong problem.
What each one actually sells
A consultant sells hours and judgment. You bring a problem, they diagnose it, they design the fix, and they either build it or stand over the build. What you are really buying is pattern recognition. They have seen your exact mess, the quotes that take three days, the phones nobody answers after five, the follow-up that never happens, at forty other companies. That experience is the product. Good consultants end engagements. Their job is to make themselves unnecessary on that problem.
An agency sells a marketing engine on retainer. Content, ads, campaigns, social, email. You pay every month and something shows up every month. Agencies are built for a different job than consultants: not fixing one broken workflow, but producing a steady stream of attention and leads. The retainer never really ends, because the engine needs fuel forever.
DIY costs your time instead of your money. YouTube tutorials, forums, a Saturday with a workflow builder. You learn the most this way, and for simple tasks it genuinely works. The price is your evenings, and the support contract is you, forever, at 9 PM when it breaks.
None of these is a scam. They are just different tools. The mistake is hiring a hammer to do a screwdriver's job.
Which problem fits which model
Match the model to the shape of your problem.
One broken workflow: hire the consultant. Your quotes take three days and you are losing jobs to faster competitors. Your phones go to voicemail after hours and those callers book with someone else. Your follow-up is "I'll get to it," and you never get to it. These are single, bounded, expensive problems. A consultant finds the costliest one, fixes it, and leaves. Weeks, not years.
You need leads every week: hire the agency. Your pipeline is empty and you need a machine that produces attention on schedule. Content, ads, nurture sequences, the whole apparatus. That is what agencies are for, and a good one will show you cost per lead without flinching.
Simple task, tight budget, and you like learning: DIY. Connecting a web form to a spreadsheet. A basic email auto-reply. A reminder text for appointments. If it breaks, the cost is annoyance, not lost customers. Go learn. You will understand your business better for it.
The gray zone most 20-person companies live in: you have outgrown DIY, but you do not need a full agency engine. One or two workflows are bleeding money, and you want them fixed by someone who has done it before. That is the consultant's home turf, and it is where most of the bad buying decisions happen, because this is also where the smoothest talkers sell.
Red flags in each lane
The consultant red flags. The big one: they only sell strategy decks. If the deliverable is a 40-slide PDF and the phrase "implementation roadmap" appears more than twice, you are buying a document, not a fix. Ask who builds it, who maintains it, and what happens when it breaks in month three. If those answers are vague, walk away. Second flag: they cannot name a price for phase one. A consultant who has done this before knows what the first phase costs. Third: they sell you their platform instead of solving your problem. If every answer is a subscription to their tool, you are the product.
The agency red flags. Watch for AI sold as a line item on a marketing retainer with nobody owning the result. "AI-powered lead gen" means nothing if no one can tell you what happens when a lead comes in at 11 PM. Ask for the mechanism, not the adjective. Second flag: the retainer grows but the reporting stays fuzzy. If month six looks exactly like month one on the invoice but the results are "brand awareness," you are funding their payroll. Third: they cannot explain what they would stop doing. An agency that does everything does nothing well.
The DIY red flags. These are red flags in yourself, which makes them harder to see. The big one: the automation works, so you stop checking it, and six months later you discover it has been quietly misfiring. Nobody noticed because noticing was nobody's job. Second: you built something only you understand, and now you cannot take a vacation. You did not automate a workflow, you built yourself a second job. Third: the "free" tool now costs $400 a month across seven subscriptions you forgot about. DIY has a way of becoming the most expensive option while feeling like the cheapest.
What good looks like under the hood
This section is for the technical reader, the IT lead or the dev who got handed this decision. Here is how you evaluate any of the three lanes like an engineer.
Ask where your data lives. Whatever gets built, your customer data, your conversation history, your pricing logic, should live in systems you control. If the consultant's answer is "in our platform," you are renting your own operations back from them.
Ask what happens when the model changes. AI providers ship new models constantly and deprecate old ones. A well-built system swaps the model like changing a battery. A badly built one needs surgery. Ask directly: "If our provider doubles prices tomorrow, what breaks?" The good answer is "nothing, we reroute." The bad answer is a long silence.
Ask who approves what. Every automation needs a human gate on the decisions that matter: money moving, promises made to customers, anything irreversible. If the design has no approval step, it is not automation, it is abdication.
Ask about cost per execution. Subscriptions hide the real number. Find out what one run costs, one call answered, one quote drafted, one lead followed up, and what happens to that number at 10x volume. Surprises here are how $200-a-month tools become $2,000-a-month tools.
Ask for the exit plan. "If we part ways in a year, what do we keep?" Documentation, credentials, data exports, the whole thing. A vendor confident in their work answers this happily. A vendor building a trap changes the subject.
This checklist works on consultants, agencies, and your nephew's DIY build alike. Print it out. Use it in every sales call.
The Monday morning action
Here is what to do this week. Write down the one workflow costing you the most money. Not three. One. Time it for five business days: how long quotes take, how many calls go unanswered, how many leads get no follow-up. Put a dollar figure on it, even a rough one.
Then match the shape. One broken, expensive workflow: talk to a consultant, and run them through the checklist above. Empty pipeline: talk to an agency, and demand cost-per-lead reporting in writing. Small, simple, low-risk: block out a Saturday and learn it yourself.
And if you want someone to find that costliest workflow with you before you spend a dollar on any lane, that is what the diagnostic is for. Thirty minutes, no pitch. We will tell you which lane fits, even if the answer is "do it yourself."
Questions we hear
What does an AI consultant do? An AI consultant finds the work costing you the most, designs a way to automate it, and either builds it or oversees the build. You are paying for judgment earned from seeing your problem dozens of times, not for software licenses.
How is an AI consultant different from an agency? A consultant sells hours and judgment aimed at one specific operational problem. An agency sells an ongoing marketing engine, content, ads, and campaigns, on a monthly retainer. Different problems, different models.
Should I DIY my business automation? DIY makes sense for simple, low-risk tasks when you have more time than budget and you want to learn. It stops making sense the moment a breakdown costs you customers, because then your weekends carry the support contract.
What should an AI consultant cost? It varies widely by scope, which is why you should ask for a fixed price on a defined first phase rather than an open hourly meter. [CONFIRM: Krastor's Blueprint pricing] A reputable consultant can tell you what phase one costs before you sign anything.