There is a particular feeling that comes with a technology quote sitting on your desk. You know it is a real number, with real cost. But you can’t tell whether it is the right one for your business.
Add to that the low hum of the last two years, the sense that you should probably be doing something with AI and that your competitors have already figured out what.
In the past, the advice I would give was to prioritize the essentials, find a partner who knows what they are doing, and keep learning. I still believe two of those three. The third one is broken. Keep learning still sounds right, but the underlying meaning and purpose has shifted.
Here is what changed, and how to decide what to buy now.
What did the old technology advice get right, and what does it miss?
The old advice was: figure out what is essential, spend there first, and ignore the rest.
That still holds. What it misses is that “essential” used to be a stable category. For twenty years, if you needed a system to do something specific to your business, you found a SaaS product off the shelf, or somebody had to build it. That was your option. You either paid to have software made, or you changed how you worked to fit software somebody else made.
That fork is still there. It just moved, and a lot of what sat on the expensive side of it three years ago now sits on the less expensive side.
The question is no longer “can I afford to solve this.” It is “which side of the line does my problem sit on now.” That is a harder question, and it is the one nobody is answering for small business owners, because the people best positioned to answer it make more money when you land on the expensive side.
I run an agency that builds custom software development services. I am telling you that the line moved against us. And it is still shifting.
What actually changed when AI arrived?
With AI, the pivot moved to a different set of questions and solutions.
Master Wall is a client of ours, an EIFS and stucco manufacturer, meaning exterior insulation and finish systems, the coatings and wall assemblies that go on commercial buildings. Their customers are contractors and architects who need to know which system to specify for a particular wall, in a particular climate, at a particular code requirement.
For years, answering that question meant one of two things. Either a person at Master Wall answered the phone and looked it up, or somebody built a decision tree: a branching form where the user picks a climate zone, then a substrate, then a finish, and the form walks them to an answer. That second option is real software. Somebody has to map every branch, build it, test it, and update it every time a product changes.
We built Master Wall an application that allows the customer to fill out some information and get the solutions they need. There is no decision tree. There is no branching form to maintain. The documentation is the system. When a document changes, the answer changes.
A whole category of problem that used to require mapping out every path in advance can now be handled by pointing a model at documents you already have. This creates systems that can be automated, updated as new information or data is entered, and the system can adapt.
What did not change? Anything where being wrong is expensive. Anything touching money, compliance, medical or legal decisions, or a commitment made on your behalf. Those decisions are still expensive and need to be treated as such. They still need a system that does the same thing every time, which still means somebody builds it, and it requires a human in the loop.
What four questions should I ask before I buy any technology?
Tracsoft has been doing this work since 1998, and every technology decision I have watched go badly failed one of these four. And in most cases, the owner or decision maker could have known before signing.
1. Does this remove work a person is doing today?
Not work you imagine someone might do. Work that is happening this week, that you could name the person doing it and roughly how long it takes them.
Columbus Water Works came to us with two websites, one public and one internal, that behaved like locked vaults. Any change, even small copy edits, required calling an outside developer. Their communications team was spending real hours every month working around their own tools. We built them a custom content management system, meaning the software that lets non-technical staff edit a website, broken into components their team could rearrange themselves. They now publish updates in minutes instead of days, and they save more than twenty hours a month.
That project was worth doing because we could point at the twenty hours before we started. If you cannot name the work and its cost, you are investing in a solution.
2. Can I tell within thirty days whether it worked?
Pick the number before you buy. Write it down. If the honest answer is that you will know in a year, the real answer is that you will never know, and it will quietly renew every year regardless.
3. Who maintains this after launch, and do they work here?
This question is often overlooked. It can turn a great idea and project into a nightmare for your business. What we have found is that the business may have “that person” on staff who can help manage the site or application, and then they leave. When that happens, those updates and dependencies generate debt that can hurt the business.
Ask the question to your team before you sign. If the answer is “us,” ask what happens if that person leaves. Do you have a plan?
4. What happens to my data?
Where does your data live, who else can see it, and can you get it back out if you stop paying. Data security and compliance is the barrier small and medium-sized business leaders name most often, cited by 49 percent in Upwork’s 2026 State of AI in SMBs report. Read the contract and know who actually owns that data. We have seen many clients come to us in distress over this very issue.
Where does AI actually pay off for a business my size?
Three places, in my experience. And these are not the sexy, exotic options. It’s actually the boring, mundane parts of your business.
Answering questions you have already answered in writing. If your expertise exists in documents, manuals, specification sheets, past proposals, a model can make it searchable in plain language. Master Wall’s rebuild, which included a chatbot alongside a redesigned site, a custom submittal tool, a distributor map and material calculators, produced a 455 percent increase in home page views, a 118.9 percent increase in average engagement time, and a 33.9 percent increase in user engagement in the first ninety days.
Those numbers are not just the chatbot, the software, or the website alone. The full suite of tools based on the existing knowledge base made a technology solution beneficial.
Drafting anything you currently stare at. Proposals, job descriptions, the fourth version of an email you have written three times. The gain is real and modest, which matches the research. In Upwork’s survey, 74 percent of leaders said AI improved productivity, but for most of them the gain stayed under 25 percent. That is a good outcome. It is not a transformation, and a vendor promising one is describing something that has not happened at scale yet.
Being found. This is the one most owners have not thought about, because it is not about how you work. It is about how customers reach you, and it changed underneath everyone. More on that below.
Where does AI not pay off?
Most pilots do not become anything. The Upwork numbers show the gap. Around 41 percent of leaders were actively piloting AI for decision support and 34 percent for workflow automation, but when you look at what is actually scaled past the pilot, the figures drop to roughly 24 to 27 percent across data analytics, content generation and inventory management. Two things pile up in that gap: nobody defined what success looked like, and nobody owned it after the person who was excited about it got busy.
Owners are quietly skeptical, and they are not wrong to be. In the 2026 Small Business AI Outlook Report, a survey of 1,009 people at US companies with 2 to 250 employees published in January 2026, 57 percent of small businesses now invest in AI. In the same survey, 39 percent questioned whether their business actually needs the level of AI investment it is making, and 48 percent had at least some concern about over-implementing automation.
More than half are buying, and roughly four in ten are not sure why. If that is you, you are not behind. You are paying attention. It’s ok to be skeptical about AI and your business.
The gains are uneven inside your own building. The same report found managers saving 7.2 hours a week with AI while individual contributors saved 3.4. If you buy a tool for your whole team and measure the result on your own experience of it, you will overestimate what it did for everyone else.
And some of it should not be sold to you at all. If a pitch cannot tell you what work disappears and who maintains it, it is not a product, it is a subscription.
What order should I do this in?
Before the AI question, there is a question about foundations. Major on the technology that is going to get you the best return, not the shiny new thing.
We looked at more than 450 Google Business Profiles across the Chattahoochee Valley. The average profile scored just under 60 out of 100 on our scale, and the weakest categories were home services and professional services. Local service businesses have a lot of opportunity with a well optimized Google Business Profile. The common failure points are easy to fix: the wrong business category, service areas left blank, no photos, no posts, a description that was never written.
None of those businesses have an AI problem. They have low hanging fruit and they are not doing the minimum marketing for their business.
Fix what is free, then fix what is broken, then automate what is repetitive, then build what is custom. In practice that means claiming and completing your Google Business Profile, making sure your website is fast and measurable, making sure your name, address and phone number match everywhere they appear, and only then asking what a tool could take off your plate.
Our digital checkup exists to answer exactly that question in order, if you would rather not do the sorting yourself.
What about marketing? Has that changed too?
More than the operations side has shifted because of AI. The promise of the internet we have lived with for more than twenty years is gone.
When someone asks ChatGPT, Gemini or Perplexity for a plumber in Columbus, the model does not hand back a ranked list of ten links. It turns that one question into several searches, pulls the results together, and writes an answer, drawing on Google Business listings, Yelp, directories and whatever else it treats as a source of truth. This generated response will list the few businesses it deems worthy, and it will not be a list of ten blue links.
If your information is inconsistent across those sources, the model does not pick the most accurate version. It gets confused, and a confused model will state something wrong with total confidence.
This is why the foundations above are not just housekeeping. Filling in your Google Business Profile properly is the same work that gets you into an AI answer, which means it pays twice: once in the map pack and once in the answer. Two birds, one stone.
We wrote about how AI search affects local business visibility in more detail, about using AI for small business if you want the tool-level version, and about lead generation for small businesses if the question underneath all of this is where the next customer comes from.
What does this actually cost?
A custom website build with us typically begins around $5,500. If you would rather see where your own project lands before you talk to anyone, our website cost calculator gives you a bracket in about two minutes with no sales call attached. A straightforward site with a handful of pages and no e-commerce usually comes back between $4,500 and $5,100.
Ongoing marketing is a monthly plan rather than a project. Local Visibility is $1,900 a month, the Grow Plan is $2,900, and the Strategic Plan, which is where custom AI and automation work lives, is $4,500. Each one is scoped in credits, so you can see what the money actually buys instead of guessing at it.
If you are not ready for any of that, a Digital Checkup is $149, and it credits toward your first project or plan.
Two things those numbers will not tell you. The largest variable in a build is not the technology, it is how much of your current process has to be untangled first. And the lowest quote is often the most expensive one you will ever accept, because a build nobody on your team can update turns into question three all over again.
One rule holds regardless of budget: never buy the platform before you have written down the work it is supposed to remove and the return you expect. See question one.
Where should I start this week?
Four things, and you can do all of them yourself. Set aside about an hour.
- Open your Google Business Profile and fill in every empty field. Category first, then services, then service areas, then photos. This is free, it is the highest return item on the list, and most of your competitors have not done it. Fill it out properly!
- Search for your own business in ChatGPT, Gemini and Perplexity. Make sure to use a query that you think a customer would use and it shouldn’t have your brand as part of the query. See whether you come up, and if you do, look at which sources the answer cites. Go make sure you are correct and complete in those specific places.
- Write down the three most repetitive tasks in your week and how long each takes. Not a wish list. A time sheet.
- Take that list to whoever you trust on technology and ask which of the three could be removed, and who would maintain the thing that removes it.
If you want a second opinion on where you stand, that is what our digital checkup is for, and if you would rather learn to do this work in house, our AI workshop for small businesses runs here in Columbus.
Frequently Asked Questions
How much should a small business spend on technology?
There is no single figure, because the biggest variable is not the technology, it is how much of your existing process has to be untangled first. A more useful rule than a percentage of revenue: never approve a purchase until you have written down the specific work it removes and who will maintain it afterward. If neither answer is clear, the budget question is premature.
Does my small business actually need AI?
Not necessarily, and being unsure puts you in good company. In the 2026 Small Business AI Outlook Report, 57 percent of small businesses invest in AI while 39 percent question whether their business needs the level of investment it is making. AI pays off reliably in three places: making written expertise searchable, drafting documents you already write repeatedly, and being found in AI search results. Outside those, ask what work disappears before you buy.
What should a small business fix before investing in AI?
The free foundations, in this order: claim and fully complete your Google Business Profile, make sure your website is fast and has working analytics, and make sure your business name, address and phone number match exactly everywhere they appear online. In our review of more than 450 Google Business Profiles across the Chattahoochee Valley, the average profile scored just under 60 out of 100, with most failures being fields left blank. That is free to fix and it improves both map results and AI answers.
Do I still need custom software now that AI exists?
Sometimes, and less often than three years ago. A large category of problems that used to require mapping out every path in advance, such as product selection tools and branching decision forms, can now be handled by pointing an AI model at documentation you already have. Custom software is still the right answer where being wrong is expensive: anything touching money, compliance, or a commitment made on your behalf needs a system that behaves identically every time.
How do I know if a technology investment worked?
Define the measure before you buy and pick one you can read within thirty days. Name the work being removed, name the person doing it today, and estimate the hours. When Columbus Water Works replaced a website that required an outside developer for every edit, the measure was hours returned to their communications team, and the answer was more than twenty a month. If your honest answer is that you will know in a year, you will never know.
