Customers Buy Projects, Not Products

A customer walks into a home improvement store to build a deck. They do not want boards. They want a finished deck before the long weekend. Boards, joist hangers, screws, stain, a saw blade and a weekend of work are one decision to them.

Retail plans that decision as forty separate ones.

Every item has its own forecast, its own buyer and its own replenishment rule. That made sense. For most of retail history the only reliable signal was the till, and the till records items. It cannot record a project.

What the till cannot see

When a project basket is complete, the sales data looks healthy. Every item sold and every forecast was roughly right.

When one item in the project is missing, something different happens. The customer rarely buys the other thirty nine items and comes back later for the missing one. Often they leave and buy the whole project somewhere that has all of it. The sales record shows a quiet week for boards and screws. It never shows the deck.

The numbers on this have been stable for over twenty years. The largest study of retail stock outs, covering 71,000 consumers across 29 countries, found an average out of stock rate of 8.3 percent. When shoppers met one, 31 percent went to another store and 9 percent bought nothing at all. Four in ten sales left the building. A further 15 percent were postponed. The study put the average loss at 4 percent of sales.

That was for single items in grocery. In a project basket, the missing item takes the other thirty nine with it.

Demand that walks away leaves no trace in your own data. The next forecast is built on the quiet week, so the range gets a little thinner, and the next project walks too. This is not a failure of the people planning. It is the limit of the signal they were given.

Why item thinking breaks first in home improvement

Every category has some of this. Home improvement has the most, for three reasons.

Purchases are projects. A bathroom refresh, a fence, a garden bed, a run of storm preparation. The value to the customer sits in finishing the job, not in any single item. Harvard's Joint Center for Housing Studies expects US homeowners to spend 518 billion dollars on improvements and repairs in 2026. Almost none of that is spent one item at a time.

Projects are local and timed. What a project needs changes by region, by climate and by the weeks before a holiday or a storm. A national item forecast smooths out exactly the variation that matters.

The cheapest item often decides the sale. The industry's own research body, HIRI, found that the most cited reason for not buying was that the retailer did not have the exact product wanted, and that 60 percent of home improvement shoppers rely on a one stop shop. A missing connector or the wrong fastener can cost the full basket. HIRI also found that shoppers tend to buy as the project progresses rather than all at once, so a single job can mean three or four trips, and each trip is another chance for the basket to walk. Item level planning ranks items by their own revenue, so it puts the least attention on the item that carries the most risk.

What changed

For a long time there was no other signal. Now there is.

Customers announce projects before they buy them. HIRI found that between 75 and 91 percent of homeowners consulted an information source before their most recent purchase, and a third cross shopped retailers first. The largest home improvement retailer says the same thing in its own annual report: customers research online and check inventory before they come to the store.

They search for how to build a raised bed, compare deck stains, save fence designs, read reviews of tile saws and fill carts across the whole market. That public trail shows the project, the items it needs, and where and when it is forming. It also shows demand for things a retailer does not stock yet, which its own sales history can never show.

The unit of demand can finally match the unit of decision.

Planning by the job

Retailers who plan by the project ask different questions.

Which projects are forming in which regions over the next planning cycle, and how fast?

For each project, which items complete it, and where is our range missing one?

Which low value items protect high value baskets, and are they planned with that weight?

Where are customers searching for a project we only half support?

None of this replaces item planning. It sits above it, the way a building plan sits above a list of materials. The items still need forecasts. The forecasts just start from the job the customer is trying to finish.

The quiet week

The most expensive weeks in home improvement are rarely the ones with bad sales. They are the quiet ones, where the data says nothing went wrong because the customers who could not finish their project were never counted.

The scale is not small. IHL Group estimates that out of stocks cost retailers 1.2 trillion dollars a year worldwide, and that US and Canadian retailers lost 349 billion dollars of sales to inventory distortion in 2022. Those figures count the item that was missing. They do not count the project that left with it.

So the question for a merchant changes. Not what sold last year, but which jobs customers tried to start with us, and how many they could finish.

Shelves record what sold. Baskets across the market record what people were trying to do.

How we look at it

Orbix reads public demand signals across the market, groups them into the projects customers are planning, and shows where a range is missing the item that completes the job, region by region.

If you want to see which projects are forming in your categories, tell us the category closest to your next buy and we will walk you through the demand signal.