For the last few years, the AI conversation in distribution has mostly been about possibility.
What can AI do? Which workflows should we test? Where should we run a pilot? How accurate does it need to be before we trust it?
Those are still important questions, but the conversation is beginning to change. Some of the largest distributors are now discussing AI in a very different place: their earnings calls.
US Foods recently said its AI initiatives are generating more than $75 million in annual adjusted EBITDA benefits across sales, supply chain, and enterprise functions, with that figure expected to exceed $110 million in 2027. More importantly, management said the largest contribution is coming from incremental sales and margin, not simply labor savings.
QXO is connecting AI investments directly to operational excellence and margin expansion and says one initiative has the potential to effectively double sales-force productivity in one of its business units. McKesson is similarly describing AI as part of a broader set of technology investments contributing to operating performance.
These are not startups presenting hypothetical ROI models. They are large distribution businesses beginning to talk about AI in the language CEOs and CFOs care about.
That may be the clearest sign yet that distribution is entering the next phase of AI adoption.
The Question Is Moving From Capability to Economics
In the early stages of any technology shift, capability gets most of the attention. Companies want to know whether something is technically possible before they worry about how much value it creates.
AI followed that pattern. Could it understand an email? Could it identify a product from an unstructured request? Could it summarize an account? Could it answer a customer question or recommend what a salesperson should do next?
As those capabilities become more established, the standard changes.
Executives increasingly want to know what happened to revenue, margin, productivity, working capital, customer retention, or operating expense after the technology was deployed.
That is a healthier conversation.
A distributor does not create value because it has more AI tools than its competitors. It creates value when those tools change the economics of the business.
The most interesting part of the US Foods example is not the size of the number. It is where the value is coming from. Management said the largest benefit is incremental sales and margin, with productivity contributing a smaller share. Its AI applications are helping salespeople identify additional products customers may buy, surface opportunities inside existing accounts, and spend less time on administrative work.
That is a much broader definition of AI ROI than simply reducing headcount.
The Best ROI May Come From Making Existing People More Productive
For many distributors, the first financial benefit from AI may not appear as a line item called "AI savings."
It may show up as revenue growing faster than headcount.
A sales representative who can manage more accounts without sacrificing service creates operating leverage. A customer service team that can absorb significantly more order volume without adding people creates operating leverage. A purchasing team that spends less time on repetitive decisions can focus on higher-value work.
US Foods offers an interesting example. As its digital and AI capabilities improve sales productivity, the company has reduced its planned rate of sales-force hiring from mid-single-digit annual growth to low-single-digit growth. It is not announcing widespread reductions in its existing sales team. It is saying that better productivity means the company may not need to add people at the same pace as the business grows.
That distinction matters.
The AI conversation in distribution has often been framed around job replacement. The more economically interesting question may be how much additional business an existing organization can support.
For a distributor planning to grow 20%, the question becomes whether operating costs also need to grow 20%.
Increasingly, the answer may be no.
Revenue May Be a Bigger AI Opportunity Than Cost Reduction
There is another reason the P&L conversation is important.
Most early AI business cases naturally focus on efficiency because efficiency is relatively easy to measure. If a task takes ten minutes today and two minutes tomorrow, the productivity gain is straightforward.
Revenue impact can be harder to isolate, but potentially much more valuable.
A sales rep may be able to identify that a customer stopped buying a particular category. AI might surface products that similar accounts purchase but this customer does not. Faster quoting may improve win rates. Better visibility into customer conversations may identify demand that never became an order.
This is already showing up in the way distributors describe their AI investments.
US Foods says the largest share of its current AI benefit comes from incremental sales and margin. QXO is investing across customer management, purchasing, pricing, inventory, quoting, and ecommerce while explicitly tying AI to productivity and margin expansion.
That should change how distributors evaluate potential use cases.
The question should not only be, "How much labor can we remove from this process?"
It should also be, "What can the business do better because this process works differently?"
Sometimes that means lower cost. Sometimes it means more capacity. Sometimes it means faster service, higher order accuracy, more customer coverage, or more revenue from the customers you already have.
The strongest AI investments may produce several of those outcomes at once.
AI ROI Should Be Tied to the Workflow
One reason AI ROI can become difficult to measure is that companies sometimes try to calculate it at too high a level.
They invest in AI broadly, then ask whether the overall business has become more productive.
A better approach is to attach the economics to a specific workflow.
If AI is being used on the order desk, measure what happens to manual touches per order, order accuracy, call answer rates, processing capacity, credits, and incremental staffing requirements.
If AI is helping sales, look at account coverage, quote response times, products per customer, sales per rep, and the amount of administrative work being removed.
If it is being used in purchasing, look at inventory availability, purchasing productivity, stockouts, working capital, and exception volume.
The metric should follow the problem being solved.
This is also why AI projects need clear business owners. Someone has to be accountable not just for deploying the technology, but for determining whether the underlying business metric actually improved.
Otherwise, organizations can end up measuring adoption instead of impact.
Ten thousand AI interactions may sound impressive. It means very little if nobody can explain what changed financially because those interactions happened.
The P&L Will Force Better AI Decisions
There is a positive side to this shift.
Once AI initiatives are expected to produce measurable operating results, companies become more disciplined about what they deploy.
The most interesting demo may not be the best investment. A relatively unglamorous workflow that occurs thousands of times every week may create substantially more value.
Order entry is a good example. There is nothing particularly exciting about eliminating repetitive manual entry from a routine transaction. But if that workflow happens tens of thousands of times a month, improving it can affect capacity, accuracy, customer response time, credits, and the economics of growth.
The same principle applies throughout distribution.
AI does not need to look impressive.
It needs to improve something that matters.
That is also why we may see fewer conversations about generic AI strategy and more conversations about specific operating metrics. The CFO does not need to know how sophisticated the model is. The CFO needs to understand what changed on the business side because the model was deployed.
We Are Entering the Accountability Phase
AI is not mature across distribution yet. Many companies are still experimenting, data and integration remain significant obstacles, and plenty of use cases will fail to generate meaningful returns.
But the expectations are changing.
When large distributors begin publicly attaching dollar values, productivity targets, hiring decisions, margin improvement, and revenue growth to AI initiatives, the rest of the industry eventually has to answer similar questions.
Not necessarily, "Are we spending enough on AI?"
But:
Where is it creating capacity?
Where is it increasing revenue?
Where is it improving margin?
Where is it allowing the company to grow differently?
Those are much more useful questions.
Final Thought
For the last several years, distributors could reasonably treat AI as an experiment.
That period is ending.
The technology is moving into everyday sales, customer service, purchasing, order processing, supply chain, and administrative workflows. As that happens, expectations will rise with it.
The companies that create the most value will not be the ones with the longest list of AI initiatives. They will be the ones that can draw a clear line between a change in the workflow and a change in the economics of the business.
We are beginning to see that happen now.
AI is no longer just showing up in distributor technology roadmaps. It is starting to show up on the P&L.
← Back to all articles