Getting Started with AI for Small Businesses: Cut One Outsourcing Bill First

Many business owners ask me: when it comes to getting started with AI as a small business, where exactly should you begin? My answer has become pretty consistent: start by opening up your outsourcing bills.
A few recent U.S. surveys happen to offer a very concrete starting point. For small companies that have truly woven AI into their daily work, the first change usually isn't layoffs. It's paying less for outsourcing, and they're more eager than others to hire more people.
Today's Highlights: Getting Started with AI? Look at Your Outsourcing Bills First
At the end of September, Citizens released its Q4 Business Pulse survey, which polled 500 business decision-makers from September 1 to 17. Among the companies that "regularly use AI across multiple functions," 94% have already reduced or cut spending on at least one outside service. The three most frequently cut were:
Marketing services: 60%
Data analytics or reporting: 39%
Bookkeeping or accounting: 38%
Even more notable is the hiring trend. Among these heavy users, 44% expect to add full-time staff in the next three months, compared with just 22% of the overall sample and only 8% of companies with no AI plans at all. They're also more optimistic: 74% expect revenue growth next quarter, versus 54% overall. Across the board, the share of companies planning to increase tech (including AI) budgets in Q4 also rose from 23% last quarter to 30%.
An October 12 Forbes article put another survey alongside it. TD Bank commissioned Wakefield Research to survey 200 small business owners. Nearly all of them had used AI to some degree, but only a minority said they used it deeply and saw big benefits. Some 40% felt underprepared for AI adoption, and the top two sticking points were data privacy and security concerns (31%) and a lack of employee training (24%).
As for benefits, 57% of the owners surveyed said AI helped them save time, 55% said it helped with cash-flow decisions, and 52% said it helped revenue. Among those who haven't yet seen the results they hoped for, the areas they most want to improve are productivity (50%) and customer service (47%).
An analysis by tech outlet TechInformed points out a key detail: the survey didn't distinguish whether respondents used AI to replace outside vendors or simply brought part of the work back in-house. "Reduced spending" could mean ending a contract, or just sending out fewer projects.
Also keep in mind that these are correlations, not causation. Companies that use AI across multiple functions may have had more breathing room and been more willing to invest to begin with. And both surveys come from banks, which have their own interest in promoting lending and digital services. The numbers are worth considering, but they're no guarantee that "use AI and you'll grow."
Jesse's Take: Put the Outsourcing Savings Toward People Who Can Use AI
What strikes me most in these numbers isn't the 94%. It's the gap between 44% and 8%. Companies that use AI well aren't replacing people; they're redirecting money they used to spend outside toward developing their own team. For small businesses in Taiwan, that order of operations matters a lot.
At many small Taiwanese companies, work gets outsourced in the first place because "nobody has time to do it." AI won't create time out of thin air, but it can shrink a three-hour task into one hour, so your in-house team can actually take it on.
Tip 1: Start with an outsourced task that's repetitive and easy to check
A common scenario: a 20-person trading company pays a marketing agency a fixed monthly fee to write social media posts and put together a monthly report, but the product information is scattered across the sales team's LINE groups (LINE is Taiwan's go-to messaging app) and a few Excel files, and the person who knows the customers best is really the owner. Work like this follows a fixed format and is easy to check, which makes it the ideal first practice exercise. On the flip side, high-stakes work like tax filing and contracts, where mistakes are costly, shouldn't be rushed back in-house. I suggest running both in parallel for a month, comparing the AI drafts side by side with the agency's deliverables, and only talking about scaling back once the quality is about the same.
Tip 2: You don't have to cut your vendors; you can change their role
Bookkeepers and accountants aren't AI's enemies. A more practical approach is to have AI organize and categorize the receipts first, then hand them to a professional for review. What you pay for shifts from "do it for me" to "check it for me." The fee usually goes down, while the risk doesn't go up.
Tip 3: Set aside part of the savings as a training budget
In the TD survey, a quarter of owners said they were stuck because their employees hadn't been trained. Plenty of companies are in the same boat: they bought the tools, but the owner is the only one using them. When you save on an outsourcing bill, set aside at least part of it for your team to take courses and practice, and agree up front on which customer data must never be pasted into AI.
A First Step You Can Take Today
Open up your payment records from the past three months and list every outsourced item in a table. Mark each one with two things: "Does it recur every month?" and "Is the output easy to check?" The item that gets a "yes" on both is your first practice exercise for getting started with AI. Next month, use AI to produce a first draft, then ask your original outsourcing partner or a colleague to help you spot the mistakes.
Further Reading
Related posts on JCXO
Original sources



Comments