“Seventy-six percent of purchase decisions are made in-store.” It’s one of the most repeated numbers in shopper marketing, and it’s also, by the industry’s own admission, a dangerous one. Shopper marketing consultancy Engage has pointed out that in-store decision rates actually swing wildly by category and mission, from less than 5% for planned purchases such as milk to more than 60% for high-impulse missions, with very few categories getting anywhere close to that widely repeated 76% figure at all. The uncomfortable truth is that most brands don’t actually know how their shoppers decide, because they’ve never tested it. They’ve assumed it.
from under 5% for a planned purchase like kids’ milk up to roughly 60% for the most impulse-driven missions, and by their own account, very few categories get anywhere close to that widely repeated 76% figure at all. The uncomfortable truth is that most brands don’t actually know how their shoppers decide, because they’ve never tested it. They’ve assumed it.
The uncomfortable truth is that most brands don’t actually know how their shoppers decide, because they’ve never tested it. They’ve assumed it.
That’s the real subject of this guide. Product placement, findability, and on-shelf availability are usually treated as three separate visual merchandising problems: where a product sits, whether shoppers can locate it, and whether it’s there when they look. In practice they’re the same discipline, predicting how a shopper will actually experience the shelf. We’ve spent more than 15 years testing that shopper experience directly, across more than 2 million observed shopping trips and 65-plus CPG categories, and one truth holds regardless of the problem a client wants to solve: The brands that test before they commit outperform the ones that assume.
What Actually Determines Whether a Shopper Finds Your Product
Findability, as a discipline, is older than retail’s use of it. Heather Lutze is credited with coining the term around 2000 through her Findability Group, and Peter Morville, widely recognized as a founding father of information architecture, popularized it for a broader audience. Both were writing about whether people could locate what they wanted on a website. The retail application is the same question asked of a shelf instead of a search bar: Can a shopper find the product they came for, quickly and accurately, based on how it’s packaged, arranged, and signed?
Four factors have the greatest influence on findability:
Shelf dynamics. Eye-level and midshelf positions are easier for shoppers to spot than placements on top or bottom shelves, simply because they fall inside the zone shoppers scan first as they move through the store.
Branding. People find products with strong, recognizable packaging (think of red Coca-Cola cans) faster than they do products with packaging that blends into the category. Change the branding, and you change the findability, whether or not you intend to.
Signage. Navigational signage that helps a shopper locate the right aisle or category measurably speeds up the search, particularly in large or unfamiliar stores.
Assortment changes. Adding or cutting flavors, scents, or sizes changes what shoppers expect to find where. A shopper who has bought the same product from the same shelf spot for two years will look right past it in a new location, at least the first few times.
It’s worth distinguishing findability from its cousin, discoverability. Findability is about locating something you came in looking for. Discoverability is about noticing something you didn’t know you wanted: the $1 endcap, the impulse item near checkout. Retailers need both, but they’re solved differently, and confusing the two leads to shelf strategies that are good at surprise and bad at service.
That distinction matters because the data doesn’t support the old assumption that browsing and discovery should be the priority. In a National Retail Federation Consumer View survey, 73% of shoppers said they were seeking a specific item when they shopped in-store, compared with 27% who were just browsing, and 58% rated the ability to find what they wanted quickly and easily as the top factor in deciding where to shop at all. (NRF’s report is from 2018; we haven’t found a more recent NRF figure that supersedes it, and the direction hasn’t reversed in the store designs we’ve tested since.)
We’ve tested this directly, and the relationship between findability and basket size is stronger than most retailers assume. Analyzing virtual shopping behavior from more than 150,000 shoppers, across dozens of categories and more than 300 store layouts, holding category and total shopping time constant, we found that every additional minute a shopper takes to make their first purchase in a category reduces their final basket size by roughly half a product. Help a shopper find what they came for faster, and they typically buy more, not less, because they spend the time they saved on deeper shopper engagement with the rest of the category instead of giving up. One adult beverage client saw this play out directly: Their retail partner assumed a confusing category layout would drive serendipitous discovery, but our data showed the opposite. Fewer than 15% of virtual shopping missions lasted over 10 minutes, and shoppers who found their first product faster were the ones who lingered and bought more, not less. That result gave the client the evidence to push their retail partner toward a clearer, easier-to-shop layout.
How to Test Shelf Placement Before You Touch a Real Shelf
Effective product placement testing comes down to three requirements, and virtual simulation is built to satisfy all three at once.
Represent the real environment. A shopper reacting to an isolated image of a shelf doesn’t behave the way a shopper reacting to that same shelf in context does, with the surrounding products, the signage, the store’s floor plans, and the path they walked to get there all shaping the decision. Testing needs to reproduce that full environment, not a cropped version of it.
Measure what’s actually happening, not just what shoppers say. Attitudinal surveys tell you what shoppers think they did. Visual attention tools tell you what they actually looked at. We use 3M’s Visual Attention Service (VAS), built into our platform as Visual Attention Analysis (VAA), to predict which shelf elements draw a shopper’s eye before a single physical shelf is touched. 3M’s own validation research, comparing VAS predictions against eye-tracking data collected at MIT, York University, and 3M’s own labs, found the model predicts eye fixations at approximately 90% of the theoretical eye-tracking limit once known measurement biases (such as the tendency to stare at the center of a computer screen) are corrected for. That’s not a substitute for behavioral testing, but it’s a fast, inexpensive way to flag which layout concepts are worth testing further and which aren’t worth the fielding cost.
Have a framework for interpreting the results. Data without a test-and-control structure is just a collection of numbers. We combine behavioral shopping missions with attitudinal survey data using test-and-control cell designs, so a client isn’t just told a number went up; they’re also told why, and what to change next.
This methodology is backed by published shelf-layout research. Erin Feeney, president and chief product officer at InContext, laid out the case plainly in a recent post on why arrangement testing keeps getting skipped: “Most companies would never launch a media campaign without testing it. So why do we still launch shelf plans without knowing how the audience will respond?” She pointed to a body of academic work, including a widely cited 2008 study showing that optimized shelf layouts are more profitable than manually created planograms, and more recent research out of Rotterdam School of Management finding that planograms organized around how shoppers actually think about a category outperform layouts organized by internal hierarchy or price tier.
We’ve seen the arrangement side of this play out directly. A global spirits manufacturer wanted to convince a major national retailer to rearrange its spirits aisle. Rather than argue the case, they tested the current planogram against a proposed one. The test layout produced a category sales increase of more than 10%, better metrics for the client’s own brand family, more shoppers trading up to premium products, and notably, better findability scores across the board. The arrangement change wasn’t just prettier. It was measurably easier to shop.
The Shelf Space Mistakes That Are Quietly Costing You Money
The most common space planning mistakes aren’t dramatic. They’re quiet, incremental, and easy to justify in the moment, and they show up as often in category management meetings as they do on the shelf itself.
Relying on historic data alone. Past performance and sales velocity are real signals, but shopper behavior shifts continuously, and shelf strategies or product assortments built entirely on last year’s sales data will always be reacting to a market that’s already changed.
Rolling out changes without testing first. Feeney’s post calls out this mistake directly. When a supply disruption, a new competitor cut-in, a regulatory shift such as a recent state-level dye ban, or another unexpected event forces an unplanned shelf change, teams under time pressure often skip testing and revise the planogram off instinct or spreadsheet math instead of shopper behavior. The risks are specific and predictable: blocked visibility, disrupted adjacency flow, and lost sales that don’t show up until the next sales cycle, by which point the cause is difficult to trace back to the shelf decision that created it.
Treating a physical mock store as the only rigorous test. Mock stores produce real insight, but they’re slow to build, expensive to reconfigure, and geographically limited to whoever can travel to see them. For a team that needs to test three or four concepts before a retailer meeting next month, that timeline often means testing gets cut.
Working from partial data. A shelf strategy informed only by an attitudinal survey that reports what shoppers say they’d do misses what they actually do. The two frequently diverge, and the gap is exactly where bad shelf decisions hide.
We’ve watched testing catch mistakes before they became expensive. A client considering delisting a low-share SKU ran a virtual switching study of repeated category shops with the product removed from the shelf and found that despite its small market share, the item drove an outsize walk rate when it was missing: A meaningful share of shoppers left the store entirely rather than substitute. The retailer kept the SKU and identified other, less critical items as better delisting candidates instead. In a separate case, a manufacturer with two packaging redesigns that promised more than $2 million in annual materials savings tested both concepts using a shopping exercise, an attitudinal survey, timed findability testing, and Visual Attention Analysis. Both redesigns underperformed the current package on noticeability, pickup rate, and conversion. The client halted the redesign before production, saving more than $500,000 in redesign and production costs and avoiding a likely revenue hit that the projected materials savings wouldn’t have covered.
Forecasting On-Shelf Availability Instead of Reacting to It
On-shelf availability (OSA), whether a product is actually on the shelf when a shopper looks for it, is a persistent, expensive problem. According to IHL Group’s 2025 inventory distortion research, the combined global cost of out-of-stocks and overstocks now runs $1.73 trillion annually, equal to roughly 6.5% of global retail sales. IHL’s analysis also found that retailers using AI-driven inventory management were seeing 2.3 times higher sales growth 2.3 than competitors relying on traditional approaches.
Most retailers manage OSA with tools that are useful but fundamentally backward-looking:
-
- Prioritizing core items, a space allocation approach that gives top sellers more facings to keep them in stock longer, usually by taking facings from slower movers, which shifts the stockout risk rather than fixes it.
- Historic sales data, using past purchase rates and sales velocity to estimate restocking timing.
- Shelf audits, manual, in-store auditing of what’s actually on the shelf and in the backroom right now.
- Predictive software, which blends historic sales and current inventory signals to flag likely stockout risk.
Each of these tells you what already happened or what’s happening right now. None of them tell you how shoppers will actually behave when a specific product goes missing, which is the piece virtual testing adds. The switching study described above is the clearest example: Rather than waiting for a real stockout and measuring the damage after the fact, a virtual test can simulate the removal of a product from the shelf and measure the shopper reaction (walk rate, substitution pattern, basket impact) before it happens in a real store. That’s forward-looking data a shelf audit or a sales history report simply can’t produce, because it hasn’t happened yet.
This is where our Arrangement AI model earns its keep. Built on more than 2 million observed shopping trips and requiring a minimum of 1 million observed trips to train validly, Arrangement AI correlates at .80 or better with full virtual study results, which means a category team facing a potential availability disruption, such as a single-supplier shortage, a regulatory pull, or a shipping delay, can get a fast, directionally reliable read on which substitution or reallocation strategy will do the least damage. Speed matters here specifically because OSA problems are time-sensitive by nature: The value of knowing how shoppers will react drops fast once the shelf is actually empty.
What Testing Costs Compared with the Cost of a Wrong Decision
The financial case for testing shelf decisions before committing to them isn’t complicated, but it’s worth stating plainly, because the alternative (skipping the test to save time or budget) usually costs more than the test would have.
It catches expensive mistakes before they’re built. The packaging redesign case above is the clearest example: $500,000 in avoided production costs from a test that cost a fraction of that to run. In a similar case, a manufacturer planning to order 25,000 units of a promotional display tested “good, better, best” design concepts, expecting the most elaborate option to win. It didn’t. The simplest design performed best, and choosing it over the expensive option saved $40 per display and $1 million in total, money that would have been spent on a worse-performing display if the team had trusted its assumption instead of testing it.
It de-risks item launches. One manufacturer facing 17 possible in-store executions for a new product used Visual Attention Analysis to eliminate the weakest concepts, then ran custom virtual research on the strongest six before customizing recommendations by retailer and using virtual walkthroughs for sell-in. The result was 100% retailer execution of the recommendation, and when the launch was validated 36 weeks later, actual results landed within 0.1 share points of the test’s prediction. The manufacturer saved hundreds of thousands of dollars and months of physical mock-store setup time it would otherwise have spent testing executions that never had a real chance.
It’s faster, not just cheaper, than the alternative. A virtual shelf test can be fielded, analyzed, and turned into a retailer-ready recommendation in a fraction of the time a physical mock-store study requires, because there’s no lease, no build-out, and no travel coordination standing between a concept and a shopper’s reaction to it.
The pattern across every example in this piece is the same: The cost of testing is small and known in advance. The cost of not testing is large and discovered after the fact, in a delisted SKU that shouldn’t have been cut, a redesign that shouldn’t have shipped, or a display order that shouldn’t have gone with the expensive option.
Frequently Asked Questions
What is findability in retail merchandising? Findability is a measure of how easily and accurately shoppers can locate a specific product on a shelf or in a store, based on factors such as shelf position, packaging and branding, signage, and assortment structure. It’s distinct from discoverability, which measures a shopper’s ability to notice products they weren’t specifically looking for.
How does shelf placement affect sales? Shelf placement affects what shoppers notice, compare, and ultimately choose to buy. Eye-level and midshelf positions are easier to spot than placements on top or bottom shelves, and clear, well-signed layouts help shoppers find products faster, which our research shows correlates directly with larger basket sizes rather than smaller ones.
What is on-shelf availability (OSA) and why does it matter? On-shelf availability measures whether a product is actually present and accessible on the shelf when a shopper looks for it. It matters because out-of-stocks and their companion problem, overstocks, cost the global retail industry an estimated $1.73 trillion a year, according to IHL Group’s 2025 research, equal to roughly 6.5% of global retail sales.
How is virtual shelf testing different from a shelf audit? A shelf audit is a backward-looking, point-in-time check of what’s currently on a physical shelf. Virtual shelf testing is forward-looking: It simulates how shoppers will react to a shelf scenario, such as a product going out of stock, before that scenario happens in a real store, giving retailers and brands time to plan a response instead of just measuring the damage afterward.
What’s the fastest way to test whether a shelf change will work before rolling it out? Build the proposed change alongside the current layout in a 3D virtual environment, run both versions past real shoppers in a test-and-control study, and measure both what they did (behavioral data) and what they thought (attitudinal survey data). This is faster and less expensive than building a physical mock store, and it produces a specific, defensible answer instead of a guess.
Does making a product easier to find reduce impulse purchases? No. In our analysis of more than 150,000 shopper missions, shoppers who found their first product faster didn’t buy less; they bought more, because the time they saved searching for their first item was time they spent shopping the rest of the category instead of leaving. Every additional minute spent finding that first product cost roughly half a product in the final basket size.
Partner with InContext
Every mistake cited in this guide (the delisted SKU that shouldn’t have been cut, the redesign that shouldn’t have shipped, the aisle nobody could shop quickly) had the same root cause: a decision made on assumption instead of a test. Our ShopperMX platform and Arrangement AI model exist to close that gap, for placement, findability, and availability planning alike. And once a layout is approved, that same environment carries forward into a digital planogram, with planogram compliance checked in the field through augmented reality tools such as our SMX GO app, so what gets tested is what actually gets built. (For more on that side of the workflow, see our guide to retail store planograms.) Contact us to see what that looks like against your own category.
