A practical, step-by-step guide for service businesses, local operators, and B2B companies

Every business is sitting on data it isn’t using. Sales records, customer transactions, supplier invoices, delivery logs, foot traffic patterns — all of it tells a story about what’s working, what isn’t, and what’s about to change. The barrier to reading that story used to be expensive consultants and enterprise software. Now it’s a conversation with an AI.

In Southeast Asia, where I live most of the year now, I inadvertently fell into the helping (primarily) hotels optimize their operations and, during a time of significant economic uncertainty and reduced travel, reduce their costs, using AI. This is the approach that we use.

This process is partly derived from our Nudgement work and applies in essence to any business; it outlines how to start using AI, any platform, any model, to make smarter, more informed decisions with the data you already have.

Step One: Feed Your Data to the AI

Start with what you have. Export your sales data, transaction records, or revenue reports into a spreadsheet. Upload it to the AI platform you’re using – Claude, ChatGPT, Gemini, it doesn’t matter. The model doesn’t need to be specialized. What matters is that the data is yours and that you’re asking the right questions. To hone in on the gaps and problems in your business it’s ideal to start with your top five and bottom five performing products or services.

Prompt the AI to identify your top five opportunities. These could be product lines, service categories, customer demographics, geographic zones, or time-based patterns, whatever is meaningful to your business. Then ask it to identify the bottom five. This alone is often revelatory. Most business owners have a gut sense of what’s working, but the data frequently tells a different story. Your highest-revenue product line might also be your lowest-margin one. Your slowest day of the week might actually be your most profitable per transaction.

Step Two: Ask for Operational Optimization

Here’s what many people don’t know or miss: AI platforms already understand business models. You don’t need to explain what a restaurant is, or how a logistics company works, or what the margin structure of a retail operation looks like. That knowledge is built in.
So once the AI has digested your performance data, ask the next question: given this data, how can operations be optimized? The AI will draw on its understanding of your business type and your specific numbers to surface recommendations you might not see from inside the operation. Maybe your staffing schedule doesn’t align with your revenue peaks. Maybe your highest-margin items aren’t getting prominent placement. Maybe your delivery radius is costing you more than it’s earning. These are the kinds of insights that emerge when you combine your data with a model that already understands the structural economics of your business.

Step Three: Automate the Optimization

Once you’ve identified what to optimize, the next move is to set up recurring processes. This is where AI shifts from a one-time insight tool to an ongoing operational advantage.

Build a recurring workflow, a simple one, where you feed updated data on a regular schedule and ask the AI to flag changes, anomalies, or emerging patterns. Review it with your team. Set up monitoring for the metrics that matter most.

For a service business, that might include job completion rates, average ticket value, customer return frequency, and cost per acquisition. For a product business, it might be inventory turn rates, supplier lead times, and per-unit margin trends.
The goal is not to replace your judgment. It’s to make sure you’re never surprised by something the data already knew.

What Can Be Optimized Using This Approach

The list is longer than most operators realize. Pricing structure and margin analysis. Staffing levels relative to demand curves. Inventory management and reorder timing. Marketing spend allocation across channels. Customer segmentation and targeting. Supplier negotiations informed by actual cost trends. Menu engineering for restaurants. Service bundling for professional services. Route optimization for delivery and field service. Scheduling efficiency for appointment-based businesses.

All of these are data problems. And data problems are exactly what AI is built to solve.

Step Four: Anticipate Changes in Your Environment

This is where the real leverage is. Optimization is about improving what you’re doing today. Anticipation is about preparing for what’s coming tomorrow.
Start with publicly available signal data. Government statistics, commodity prices, labour market reports, municipal planning documents, census data, commercial real estate listings: all of this is available and all of it affects your business. AI can digest it faster than you can.

Take fuel costs as a concrete example. If you run a delivery service, a catering company, a field service operation, or any business where vehicles are part of the cost structure, fluctuations in gas prices directly affect your margins. Ask the AI to analyze the current trajectory of fuel prices in your region, factor in publicly available forecasting data, and project what those costs are likely to look like over the next six months. Then apply that projection to your business model. If fuel costs are trending up twelve to fifteen percent, what does that do to your per-delivery economics? Does it trigger a pricing adjustment? A delivery radius change? A shift toward batching orders?

Now do the same exercise with labour costs. Minimum wage changes are usually announced months before they take effect. Cost of living increases in your area are trackable. The practical question isn’t just what you’re paying people today, it’s what you need to pay them so they can pay their bills and stay. Employee turnover is expensive. A proactive wage adjustment informed by real data is almost always cheaper than replacing someone.

Step Five: Build a Signal Digestion Cadence

Set up a process to ingest this environmental data on a regular basis. Weekly, you should be reviewing your own operational metrics; revenue, costs, margins, throughput. Monthly, you should be reviewing external signals: commodity prices, labour market shifts, local economic indicators. Quarterly, you should be doing a full strategic review: does the trajectory of your environment require a change in your business model, pricing, staffing, or service mix? Depending on the nature of your business, you may need to make these reviews more frequent.

AI can do the heavy lifting on all three cycles. Feed it the data, ask for the analysis, and use the output to make decisions before changes hit rather than after.

For Local and Service Businesses

If your business depends on a physical location and a local customer base, there are specific signals you should be monitoring with AI assistance.
Population and neighbourhood demographic changes affect your potential customer base directly. If the median age in your area is shifting, or household sizes are changing, or income levels are moving, your product mix and pricing should reflect that. Municipal data and census updates make this trackable.
Commercial vacancy rates and rent changes in your area are leading indicators. If storefronts around you are going empty, that’s a foot traffic signal. If commercial rents are climbing, that affects your own lease renewal and the viability of the businesses around you, the ones whose employees and customers are also your customers.

If you’re a B2B company, the financial health of your clients’ businesses matters as much as your own. Rising commercial rents, local business closures, or sector-specific downturns in your client base will eventually show up in your revenue. Better to see it coming.

For Businesses With an Online Component

These effects compound when you add delivery, e-commerce, or any digital service layer. Your online business is affected by every local signal (population and demographic changes, income shifts, competitor activity) plus a whole additional set of digital signals: traffic patterns, conversion rates, acquisition costs, platform algorithm changes, and marketplace fee structures.

You should be using AI to monitor your numbers independently of any provider or third party. If you’re on a delivery platform, don’t rely solely on their dashboard. Pull your own data. Analyze your own margins per order. Understand your own customer acquisition costs. Platforms are optimizing for their economics, not yours. The only way to know whether a channel is actually profitable is to run the numbers yourself, and AI makes that analysis trivially easy.
Apply the same framework: what are your most profitable product lines and what are the least? Where can you see production or fulfillment improvements? Where are the supply chain inefficiencies? What does the trend data say about where costs are headed?

None of this requires enterprise software, a data science team, or a six-figure consulting engagement. It requires your data, an AI platform, and the discipline to ask the right questions on a regular schedule. The businesses that will thrive over the next five years are the ones that treat AI not as a novelty but as operational infrastructure, a tool that reads the signals faster, processes the data more completely, and helps you make decisions before you’re forced to react.

Start with your numbers. Ask the AI what it sees. Then ask what’s coming.