Your Best-Selling Product Isn’t Your Most Profitable? Use AI to Find Out What’s Making You Money

Your best-selling product gets most of the attention because it generates the most orders, sells the most units, or brings in the most revenue. It may also be the product your employees recommend first, the one customers recognize most easily, and the item you feature most often in your marketing.

That does not necessarily mean it is the product making the business the most money. Once you account for product cost, discounts, returns, fulfillment, labor, customer acquisition, and what customers purchase afterward, a quieter product can sometimes contribute more profit than the bestseller.

AI can help you put those pieces together and analyze the role each product actually plays in the business. The objective is not to demote your bestseller simply because another product has a higher margin; it is to understand the difference between what sells the most and what creates the most economic value.

Why Your Best-Selling Product May Not Be Your Most Profitable

Sales volume and profitability answer different questions. Sales volume tells you what customers buy most often, while profitability tells you how much value remains after the relevant costs of producing and selling that item are considered.

Imagine Product A generates $100,000 in revenue with a 20% gross margin, while Product B generates only $60,000 but carries a 55% gross margin. Before considering any other costs, Product A contributes about $20,000 in gross profit and Product B contributes about $33,000.

The product with lower revenue is contributing substantially more gross profit. That does not automatically make Product B the more important product, because Product A may attract new customers or lead to additional purchases, but the revenue leaderboard alone would hide that difference.

AI can help you analyze these relationships across dozens or hundreds of products. The analysis becomes especially useful when you combine basic margin information with customer behavior rather than looking at each product as an isolated transaction.

Start With Revenue, but Do Not Stop There

Begin with the numbers you probably already track. Pull units sold, revenue, average selling price, product cost, discounts, refunds, and gross margin for the products you want to compare.

Then add the less obvious costs where they materially affect profitability. These might include shipping subsidies, packaging, payment fees, marketplace commissions, fulfillment labor, installation time, customer support, warranty claims, or other costs that vary meaningfully by product.

You do not need a perfect cost-accounting system before you can begin. Even a first pass can help you see that two products generating similar revenue may require very different amounts of money and work to sell.

A useful AI prompt might look like this:

I am comparing the profitability of several products. For each product I have units sold, revenue, average selling price, product cost, discounts, refunds, shipping or fulfillment costs where available, and estimated labor requirements. Calculate the major profitability differences, show the assumptions you use, and identify where missing cost information could materially change the conclusion. Do not rank the products using an arbitrary AI score, and do not assume the highest-margin item is automatically the most strategically valuable.

That final instruction is important because profitability is only one part of the product’s role. You want AI to help organize the economics without turning a complicated business decision into a simplistic leaderboard.

Compare Revenue, Margin, and Workload Together

Two products can look equally successful on a sales report while having very different economics underneath them. The differences become easier to see when you place multiple dimensions side by side.

What you notice	What may be happening	What AI can analyze	What to investigate
High revenue but low gross margin	The bestseller may generate volume without much profit per sale	Revenue, cost of goods, discounting, margin	Whether volume creates enough total profit to justify the economics
Strong margin but low sales	A profitable product may be underexposed or serve a smaller audience	Margin, traffic, visibility, conversion	Whether more customers should be introduced to it
High sales with frequent discounts	Revenue may depend heavily on promotions	Full-price versus discounted sales, margin, repeat behavior	Whether customers will buy without the discount
High revenue with many returns	Reported sales may overstate economic value	Revenue, returns, restocking, support, refund cost	Why customers return the product
Moderate sales with repeat purchasing	The product may create longer-term customer value	Purchase history, repeat rate, lifetime revenue	Whether it is an important retention product
Strong sales with high labor requirements	Profit may be reduced by service or fulfillment effort	Labor hours, fulfillment, support, customization	Whether operational costs are fully reflected
Low-margin product frequently leads to other purchases	It may function as an entry product	First purchase, subsequent purchases, basket behavior	Whether the customer relationship makes the economics worthwhile

The table helps prevent one metric from controlling the conversation. A product can perform poorly on one dimension and still serve an important role in the overall business.

Calculate Gross Profit, Not Just Gross Margin Percentage

Margin percentage is useful, but percentage alone can also mislead you. A product with a 70% margin that sells twice a month may contribute less total gross profit than a product with a 30% margin that sells hundreds of units.

Ask AI to calculate both gross margin percentage and total gross profit dollars. Then compare those figures with the amount of work, inventory, marketing, and customer support each product requires.

Suppose Product A produces $20 of gross profit per sale and sells 1,000 units, while Product B produces $80 of gross profit per sale and sells 150 units. Product A contributes $20,000 and Product B contributes $12,000, so the lower-margin product still contributes more total gross profit.

The business needs both views. Margin tells you the economics of an individual sale, while total contribution shows what the product produces at its actual sales volume.

Look at Discounts by Product

A product may appear successful because it sells frequently during promotions. Its profitability can look very different when you separate full-price sales from discounted sales.

Give AI the regular price, actual selling price, discount amount, units sold, and margin for each transaction or period. Ask it to compare how often each product sells at full price and how much profit remains after promotions.

You may find that the bestseller sells in large quantities but is almost always discounted. Another product may sell fewer units while maintaining its price and contributing more profit per transaction.

That distinction can change your promotional strategy. Instead of automatically discounting the product that already sells well, you may decide to test whether its demand is strong enough to support fewer promotions.

Include Returns and Refunds

Revenue reports can make a product look excellent before returns are taken into account. A product with high sales and an unusually high return rate may generate additional shipping, processing, restocking, support, or damaged-inventory costs.

Compare return rates and reasons across products. AI can group return comments and customer-service notes to help determine whether one product creates significantly more post-purchase expense than others.

A high return rate can also reveal a marketing problem. Customers may be purchasing because the product description creates an expectation the actual product does not meet.

That means profitability analysis can lead back to positioning and customer communication. Fixing the expectation gap may improve both margin and customer experience.

Account for Labor When Products Require Different Amounts of Work

Product profitability becomes more complicated when some items require substantial employee time. Customization, installation, preparation, consultation, packaging, assembly, or customer support can make one product much more expensive to deliver than another.

Estimate how much employee time each product requires and assign a reasonable labor cost where useful. The estimate does not need to be precise to the penny if your purpose is to understand the relative difference.

Suppose two services each generate $2,000 in revenue. One requires four employee hours and the other requires 15, so treating them as equally valuable because the sales figures match would miss a major operational difference.

AI can help calculate revenue and estimated contribution per labor hour. That can reveal products or services that generate impressive top-line revenue while consuming a disproportionate share of your available capacity.

Look at Fulfillment and Shipping Costs

Shipping can substantially change product economics, particularly when the business absorbs some or all of the cost. Heavy, fragile, oversized, refrigerated, or unusually packaged products can create expenses that are easy to overlook in a basic product-margin report.

Compare shipping revenue collected with actual shipping and packaging expense where possible. Ask AI to show which products generate the largest gap between what customers pay for delivery and what fulfillment actually costs.

The same issue can appear inside a physical business. One product may require special storage, handling, preparation, or equipment, and those operational requirements can affect the true economic contribution.

Not every overhead expense needs to be assigned perfectly to every unit for the analysis to be useful. Focus first on costs that materially differ among products and could change the business decision.

Analyze Product Profitability by Sales Channel

The same product can have very different economics depending on where it is sold. A sale through your own store, website, marketplace, distributor, delivery platform, or wholesale partner may produce different fees, discounts, fulfillment costs, and customer acquisition expenses.

Break product profitability down by channel when the data allows it. AI can compare the same item across sales environments and show where the economics materially change.

A bestselling product may be highly profitable through your website but far less attractive through a marketplace charging significant commissions. Another product may perform better through wholesale because the fulfillment process is simpler even though the selling price is lower.

This helps you move beyond asking “Which product is most profitable?” to asking “Which product-channel combinations are most economically useful to the business?”

Look at Customer Acquisition Cost

Some products require much more marketing investment to sell. If one product depends heavily on paid advertising while another is purchased through organic search, referrals, or repeat customers, revenue alone may hide an important difference.

Connect product sales with acquisition source where you reasonably can. Then compare advertising cost, lead-generation cost, or other acquisition expense with the profit generated by customers who first purchase each product.

Be cautious with attribution because customer journeys are rarely perfectly measurable. AI can help organize the available evidence, but it should not assign every sale to one marketing interaction simply because a platform claims credit.

The purpose is to understand whether one product requires significantly more acquisition expense than another. Even approximate differences can be useful when they are large enough to affect the economics.

Find Out Whether the Bestseller Is an Entry Product

A low-margin bestseller may still be one of the most valuable products in the business if it consistently creates new customer relationships. The first purchase may be less profitable because its role is to introduce people to the company.

Look at what customers do after buying the bestseller. Do they return, purchase higher-margin items, subscribe, book another service, or refer other customers?

AI can analyze first-purchase products and subsequent customer behavior. Ask it to compare the 30-day, 90-day, six-month, or annual value of customers based on what they purchased first.

You may discover that people who begin with the bestseller become unusually valuable customers over time. In that case, evaluating the product only by first-transaction margin would miss a significant part of its economic role.

Compare First Purchase With Lifetime Customer Value

Suppose Product A generates $15 of profit on the initial purchase while Product B generates $45. Looking only at the transaction makes Product B appear substantially stronger.

Now suppose customers who begin with Product A purchase four more times during the year while customers who begin with Product B rarely return. The customer-level economics may reverse the conclusion.

Give AI purchase history grouped by first product and ask it to compare repeat rate, subsequent revenue, gross profit, average time to second purchase, and other useful customer-value measures. Keep enough historical time in the analysis so newer customers are not unfairly compared with customers who have had much longer to repurchase.

This is particularly important for businesses with replenishment, subscriptions, follow-on services, accessories, or natural purchase sequences. The most valuable product can be the one that starts a profitable relationship rather than the one that produces the largest margin on Day 1.

Look at What Customers Buy With the Bestseller

Basket analysis can reveal another source of product value. The bestseller may have a modest margin by itself but frequently bring high-margin products into the same transaction.

Ask AI to identify products that commonly appear together and calculate the average basket value and estimated gross profit when the bestseller is present. Then compare that with transactions containing other products.

If the bestseller regularly drives attachment purchases, its economic contribution extends beyond its own margin. Removing or de-emphasizing it simply because the product has a lower percentage margin could reduce the value of the entire basket.

The opposite can also happen. A high-volume product may dominate transactions while rarely leading to additional purchases, which makes its individual profitability more important.

Identify High-Margin Products Customers Rarely See

Profitability analysis can reveal missed marketing opportunities. A product with strong margin and high customer satisfaction may receive surprisingly little website visibility, merchandising space, employee attention, or marketing support.

Compare profitability with exposure. Look at homepage placement, category ranking, email mentions, advertising, in-store display, employee recommendations, and other ways customers encounter the product.

AI can help identify products that appear economically attractive but underrepresented in marketing. Those products can become candidates for a controlled visibility test rather than an immediate large-scale promotion.

If additional exposure improves sales while the margin remains strong, the business may have found a relatively straightforward growth opportunity. If sales remain weak, the issue may be limited demand rather than insufficient marketing.

Look for Products With Strong Profit but Weak Positioning

A profitable product may be difficult for customers to understand. It could have a vague name, weak images, technical descriptions, or a value proposition that employees explain well in person but the website does not communicate effectively.

Compare the product’s financial performance with customer questions, reviews, website behavior, and sales conversations. AI can help determine whether customers who encounter the product understand why they would choose it.

A high-margin product does not automatically deserve more promotion if the offer itself remains confusing. Improving the positioning first may make the eventual marketing investment much more productive.

This is where product profitability and marketing strategy begin to overlap. Financial data tells you an opportunity may exist, while customer behavior helps you understand what would need to change for more people to buy.

Analyze the Products Customers Reorder

Repeat purchases can materially improve the economics of a product. A product with an average initial margin may become extremely valuable when customers replenish it every month or several times per year.

Calculate repeat rate and average purchase interval by product. AI can help identify which products are most likely to be purchased again and how quickly those repurchases typically occur.

Then compare acquisition effort with repeat behavior. A product that costs more to acquire initially may still have attractive economics if customers continue buying without requiring the same marketing expense each time.

This can also help you improve retention communication. If customers typically reorder a particular item every 60 days, the business can use that information to create useful reminders around the natural purchase cycle.

Look at Products That Generate Referrals

Some products or services create customers who are particularly likely to recommend the business. That contribution can be difficult to see in a standard profitability report.

If you track referrals, connect the referring customer with their purchase history. AI can help identify whether certain products appear disproportionately among customers who later refer others.

Do not assign all referral revenue to the original product as though the relationship were perfectly causal. Treat it as another signal that the product may create a particularly strong customer experience.

A product with moderate direct profit but strong repeat and referral behavior can be more valuable than its immediate transaction suggests. Customer behavior after the sale often reveals economic value that the initial invoice does not capture.

Examine Products That Create Customer-Service Burden

Some products generate a large number of calls, emails, troubleshooting requests, returns, replacements, or complaints. That support burden consumes labor and can reduce the profitability of an item that looks strong in the sales report.

Connect support tickets or customer-service interactions with products when possible. Ask AI to identify which products generate the most support relative to the number of units sold.

Then analyze the reasons. The issue might involve unclear instructions, product quality, setup difficulty, shipping damage, incorrect customer expectations, or another recurring problem.

Improving the product experience may increase profitability without changing the price. Reducing avoidable support activity can create economic value that never appears in a revenue-growth report.

Avoid Automatically Pushing the Highest-Margin Product

Once businesses discover that one item carries a much higher margin, there can be a temptation to promote it aggressively. That can create problems if the product serves a narrower need or is not the best option for many customers.

Use customer fit alongside profitability. The most profitable sale is not particularly valuable if customers feel they were pushed toward the wrong product and later return it, complain, or stop trusting the business.

AI can help identify which customer situations are most associated with different products. Employees can then use profitability information as one business consideration without allowing margin to replace good customer guidance.

The strongest product strategy aligns customer value and business value. Products that work well for both sides are usually more sustainable than products optimized around margin alone.

Do Not Assume Low-Margin Products Are Bad

Low-margin products can serve legitimate strategic roles. They may attract first-time customers, create store traffic, complete a product assortment, support a more profitable service, or give customers a reason to return frequently.

Define the role before judging the product. If a low-margin product generates profitable cross-sells or introduces many of your strongest long-term customers, the economics need to include those effects.

AI can help map those relationships across transaction history. The analysis should make the product’s role visible rather than simply labeling it a weak performer based on one financial metric.

You may still decide the margin needs improvement. The difference is that you will make that decision with a clearer understanding of what else could change if the product changes.

Watch for Products That Look Profitable Because Costs Are Missing

A profitability report is only as useful as the cost information behind it. One product can appear exceptionally profitable because labor, fulfillment, returns, packaging, or another major expense has never been assigned to it.

Ask AI to flag products whose cost structure appears incomplete. You can also ask which missing cost categories would be most likely to change the relative profitability of the products.

This is particularly important for service businesses where the primary cost is employee time. A service with very little direct material cost may look extraordinarily profitable until the hours required to deliver it are included.

Use the analysis as a decision aid rather than pretending approximate costing is exact. If the conclusion changes dramatically depending on one uncertain assumption, that uncertainty should remain visible.

Compare Profitability Over Time

Product economics can change. Supplier prices rise, shipping costs move, discounting becomes more frequent, employee time increases, and customer behavior changes.

Compare the same product across several periods rather than relying only on the most recent month. AI can identify products whose revenue remains strong while margins gradually deteriorate.

That pattern is easy to miss because the bestseller can continue producing impressive sales numbers. The business may not notice that each sale contributes less than it did a year earlier.

Historical analysis also helps distinguish temporary problems from structural ones. A one-month margin decline caused by a short-term supplier issue deserves a different response from a two-year downward trend.

Model What Happens if You Sell More of a Product

More sales do not always create proportionally more profit. Capacity limits, staffing, shipping thresholds, supplier pricing, overtime, and support requirements can change as volume grows.

Ask AI to model several scenarios using your actual assumptions. You could estimate what happens if sales increase 10%, 25%, or 50% and show the additional revenue, gross profit, inventory requirement, labor, and other variable costs.

Using the cost and capacity information below, model what would happen if monthly sales of this product increased by 10%, 25%, and 50%. Show revenue, estimated gross profit, additional inventory requirements, labor, fulfillment costs, and any capacity thresholds I have provided. Keep the assumptions visible and do not assume that costs remain constant if the information indicates they would change at higher volume.

This helps you evaluate whether a profitable-looking product is also scalable. An item can have strong current economics while becoming operationally difficult at much higher volume.

Model What Happens if You Sell Less of the Bestseller

You can also test the opposite scenario. If the bestseller is low-margin, ask what could happen if the business sold slightly less of it and redirected some attention toward higher-contribution products.

Include basket effects and customer acquisition in the scenario. If fewer bestseller sales also mean fewer new customers and fewer related purchases, the supposed margin improvement could disappear.

AI is useful for organizing these tradeoffs because several variables can change simultaneously. The output should be treated as a scenario based on assumptions rather than a prediction of what will definitely happen.

This exercise is particularly useful before making major merchandising or marketing shifts. It helps the business see what the bestseller may be supporting beyond its own profit line.

Give Products Different Strategic Roles

A useful product portfolio does not require every item to accomplish the same thing. Some products may attract customers, some produce margin, some create repeat purchasing, and others increase the size of the overall transaction.

You can define a small number of strategic roles based on actual behavior. The labels might include entry product, profit contributor, repeat-purchase driver, complementary product, premium option, or another category that makes sense for your business.

AI can help you identify patterns that support those roles, but the business should determine whether the categories are strategically meaningful. Avoid assigning elaborate labels simply because the model can create them.

Once the roles are clear, product decisions become more thoughtful. You can evaluate whether an item is succeeding at the job it is supposed to perform instead of expecting every product to win the same sales contest.

Use Customer Feedback Alongside the Profitability Data

A spreadsheet can tell you which product contributes more profit, but it cannot fully explain why customers choose it. Reviews, customer interviews, sales notes, returns, and service conversations provide another layer of information.

Give AI feedback grouped by product and ask it to identify recurring reasons for purchase, satisfaction, complaints, and unmet expectations. Then compare those themes with the financial performance.

A high-margin product with enthusiastic customers may deserve more visibility. A highly profitable product with recurring complaints deserves a very different response.

Combining customer and financial information helps protect the business from optimizing only for short-term economics. Profitable products become even more valuable when customers also feel that the purchase was worthwhile.

What I Would Do This Week

I would take the bestseller and four or five other meaningful products and build a simple profitability comparison. I would include revenue, units sold, selling price, product cost, discounts, returns, gross profit, estimated fulfillment or labor cost, and any other expense that differs materially between the products.

Then I would give the information to AI and ask it to calculate where the economics actually differ. I would specifically ask it to identify where a conclusion depends on missing costs or assumptions so a rough estimate does not accidentally become a false certainty.

Next, I would connect the products with customer behavior. I would compare first purchases, repeat purchases, basket combinations, customer acquisition sources, returns, support needs, and referrals where that information exists.

I would then identify one product whose role deserves a closer look. It might be a bestseller with weaker economics than expected, a high-margin product receiving very little marketing support, or an entry product that appears unprofitable until later customer purchases are included.

Finally, I would test one change rather than overhauling the entire product mix. I might reduce unnecessary discounting on the bestseller, increase visibility for a profitable underexposed product, improve positioning, bundle complementary items, or investigate a support cost that is quietly eroding margin.

Your best-selling product may still be one of the most important products in the business even if it is not the most profitable on an individual sale. AI can help you see the larger economic picture so you understand which products generate revenue, which create profit, which build customer relationships, and which deserve a different role in your marketing strategy.

If you want to connect marketing performance more directly with meaningful business outcomes, How Do You Know if AI Is Helping Your Marketing? provides a broader measurement framework. If the analysis shows that a profitable product is difficult for customers to understand or is not being positioned clearly, How to Use AI to Improve Your Marketing Offer can help you examine how the offer is presented.