What Is In-Store Advertising ROI, and How Do You Prove It?

Walk into a busy store on a Saturday. A screen at the entrance is running a seasonal offer, a shelf-edge label is showing a price that changed overnight, and a phone in someone’s pocket buzzes with a discount for the aisle they are standing in. All three are in-store advertising. All three are asking for the same thing: one more decision in your favour before the shopper reaches the till.

The question a retail marketer actually gets asked, though, is not whether any of that works. It is what the programme returned. This page answers that: how to calculate the number, which costs belong in the denominator, what a believable lift test looks like, and which of the figures circulating in this category can be traced to anything at all.

What is in-store advertising ROI?

In-store advertising ROI is the incremental gross profit a campaign produced, minus its full cost, divided by that cost. The word doing the work is incremental: profit that would not have happened without the campaign, measured against comparable stores that did not run it.

Written as a formula:

ROI = (incremental units × gross margin per unit − total campaign cost) ÷ total campaign cost

Every term in that line is a place where programmes go wrong. “Incremental units” gets replaced with total units sold. “Gross margin” gets replaced with revenue, which inflates the result by whatever the margin rate is. And “total campaign cost” usually means the hardware invoice, leaving out the three cost lines nobody signed a purchase order for.

A worked example, with round numbers so the mechanics are visible rather than the answer. Twenty test stores sell 4,200 more units of a featured item over eight weeks than twenty matched holdout stores. Gross margin is £4 per unit, so £16,800 of incremental gross profit. The campaign cost £11,000 across hardware amortisation, content production and the analyst time to join play logs to sales. ROI is (16,800 − 11,000) ÷ 11,000, or 53%. Change the margin assumption to £2 and the same campaign loses money. That sensitivity is the real finding, and it is why the margin of the featured item matters more than the creative.

Why does the store still matter if retail moved online?

Because most retail spending never went online. E-commerce was about 16% of total United States retail sales in Q4 2024, according to the U.S. Census Bureau’s quarterly retail e-commerce estimates, leaving the large majority of sales in buildings with doors. The same source puts total retail and food services sales above $7 trillion for 2024.

That is the pool in-store advertising works inside, and it explains the pressure on measurement. Shelf space that carries a message is now expected to earn like media, with the reporting that implies. Most stores cannot produce that reporting yet, which is why so many ROI conversations end in an anecdote about a screen somebody liked.

What are the formats, and what can each be measured against?

Each format serves a different moment and leaves a different trail. The useful distinction is not screen against paper. It is whether the message can change on the day, and whether it leaves a record you can join to a sale.

Printed and fixed material (posters, shelf wobblers, endcap headers, floor graphics) is cheap per unit, slow to change, and effectively invisible to measurement. Digital formats change from one place and produce logs. Phone-triggered messaging is the only family that knows who it is talking to, which is both its advantage and the reason it carries a consent requirement the others do not.

What belongs in the cost side of the calculation?

Four cost lines, and only one of them is the screen. Hardware and installation are visible on an invoice. The other three arrive later, are usually funded out of someone else’s budget, and are the reason a programme that looked profitable on paper does not survive its second year.

Content operations is the line that kills programmes. A screen network needs someone producing and scheduling material every week. When that role is unfunded, the network ages into a slideshow of expired offers, which is worse than a printed poster because the shopper can tell it is stale.

Log-to-sales integration deserves its own note, because it is the cost of being able to calculate ROI at all. Play logs sit in one system and transactions in another, and joining them on store ID and timestamp is a real piece of engineering work. A programme that skips it has bought a media network it cannot report on, and the first ROI review will be a slide of impressions.

Which in-store advertising claims should you not trust?

Most of the widely quoted figures in this category cannot be traced to a published study. The familiar claim that a large majority of purchases are unplanned circulates in trade summaries that cite each other, and the original methodology, sample and date are usually missing.

The same is true of sales-lift numbers. A double-digit lift in pilot stores is a plausible result and a meaningless sentence without four things attached:

• Control stores. Lift against last year is not lift against a comparable store running no campaign.
• A pre-period. A store that was already growing will show a lift on the day you switch a screen on.
• Category-level reporting. A featured item can gain units while the category stays flat, which means the campaign moved a choice rather than creating a sale.
• The cost side. Hardware, installation, network and content labour belong in the same table as the lift.

Unsourced figures have been removed from this page rather than restated. If a number here has no link next to it, it is not a measurement. When a supplier offers you a lift figure without the four items above, treat it as a marketing asset rather than evidence, and ask which stores were the controls.

How do you measure in-store advertising without fooling yourself?

Use a holdout. Pick comparable stores, run the campaign in some and not others, and compare the same weeks across both groups, because everything else measures the season rather than the screen.

Matching the two groups is most of the work. Pair stores on weekly footfall, format, catchment and the pre-period sales of the category you are about to promote, then check that the two groups tracked each other for the eight weeks before the campaign. If they did not, the pairing is wrong and no amount of post-hoc adjustment will fix it. Twenty stores a side is a working minimum for a single-digit lift; a handful of stores can only detect an effect large enough that you would not need a test to see it.

The metric set is short, and each item answers a different question:

Footfall and dwell come from indoor positioning or a sensor; interaction rate from kiosk or QR logs; the unit figures from point of sale. Signage production spend comes from finance, and belongs in the review because a fall in print cycles is a real return even when unit sales do not move.

The pairing that matters is the last two against each other. Featured-item units rising while category units stay flat means the campaign redistributed demand. That can still be the goal if the featured item carries a better margin, but it is not growth and should not be reported as growth.

Set targets as a number, a zone and a deadline. “Raise footfall to the rear homeware aisle by 15% in eight weeks, measured against four holdout stores” is testable. “Improve engagement” is not.

Where should an in-store screen go, and how would you know?

Placement should follow measured movement, not the store plan. A floor plan tells you where the aisles are. It does not tell you where shoppers slow down, where they turn back, or which corner they walk past without looking at.

Three signals matter, and they are not the same thing. Footfall counts passing bodies, and high footfall with no dwell is a corridor, which is where content goes unread. Dwell time identifies where a decision is already being made: a shelf where people stand still is a shelf where a comparison is happening. Path shows the sequence, and the same offer reads differently at the third stop than at the first.

Indoor positioning is what turns those from guesses into numbers. Mapsted’s retail and mall deployments use blue-dot indoor-outdoor navigation with horizontal and vertical positioning for multi-level venues, and the same positioning layer produces heat maps, dwell times and movement patterns for the floor. It uses existing infrastructure, with no additional on-site hardware installation, which keeps the measurement layer out of the capital line of your ROI calculation.

Two practical rules follow. Do not put a screen where you want traffic to be; put it where traffic already slows, and use it to point at the zone you want to fill. And resist density. A store with a screen in every sightline trains shoppers to ignore screens, which is an expensive way to reach nobody.

What do privacy rules allow for phone-triggered offers?

Phone-triggered advertising needs consent in most markets a chain operates in, and that consent has to be specific rather than bundled. Location data tied to a device is personal data, which is settled law rather than a matter of interpretation.

In the European Union, Regulation (EU) 2016/679 requires a lawful basis for processing, and Article 4(11) defines consent as freely given, specific, informed and unambiguous. Separately, Article 5(3) of the ePrivacy Directive 2002/58/EC requires consent before storing or accessing information on a shopper’s own device, which is what a beacon-listening app does. In California, precise geolocation is sensitive personal information, defined in the California Consumer Privacy Act as locating a consumer within a radius of 1,850 feet or less, and consumers can limit its use; the California Privacy Protection Agency publishes the notice and opt-out duties that follow.

The design consequence is simple, and it has an ROI dimension. Aggregate analytics and personalised messaging are two different products with two different legal footings. A project that bundles them into one consent prompt loses the analytics every time a shopper declines the offers, and the analytics are what you were going to measure the campaign with.

What makes an in-store screen legible and accessible?

Two published standards decide most of it, and neither is optional if the store serves the public. Legibility is a contrast and type-size question; reach is a physical dimension question. Both have numbers you can check with a meter and a tape measure.

For anything with text, WCAG 2.1 Level AA Success Criterion 1.4.3 asks for a contrast ratio of at least 4.5:1 for normal text and 3:1 for large text. That single check rules out a large share of the promotional templates brands send to retailers, particularly white type over photography.

For anything a shopper touches, the 2010 ADA Standards for Accessible Design cap unobstructed forward reach at 48 inches above the floor and low reach at 15 inches, and limit the force needed to operate a control to 5 pounds. A kiosk mounted for the convenience of the installer fails this quietly and permanently, and a failed kiosk returns nothing while still carrying its full cost.

Content rules that follow from both: one idea per screen, type sized for the distance at which the screen is actually read rather than the distance it was designed at, and no message that depends on colour alone to be understood.

Where is the trade-off honest?

The trade-offs are real, and a business case that names them in its first version survives its second year. One that promises a per-screen sales figure does not.

• Print is still right for some jobs. Static categories with stable prices do not need a display refresh cycle, and the ROI of not buying a screen is 100% of a cost you avoided.
• Beacons are sometimes the right answer. A single promotional zone with a hardware budget and no app footprint is a reasonable fit. The trade is ongoing battery and maintenance work across the estate, which belongs in the cost line.

The other two limits are about the ceiling on what any of this can return, and they are arithmetic rather than opinion.

• Phone messaging has a ceiling. It reaches app users who consented, which is a fraction of the shoppers in the building. No technology choice changes that.
• Attribution will be imperfect. A holdout design gives you a defensible estimate for a group of stores, not a per-screen number.

The last one is worth being blunt about internally. If someone needs a return figure for every individual display, the honest answer is that no method in this category produces one reliably, and promising it is how a programme loses credibility in year two.

Your screens are already placed somewhere. The question is whether that somewhere is where shoppers slow down. Mapsted maps your store, shows how people actually move through it with heat maps and dwell times, and places contextual messaging against that movement instead of against a floor plan. Book a scoping call and we will walk your layout with you.

Frequently Asked Questions

How do you calculate in-store advertising ROI?

Multiply incremental units by gross margin per unit, subtract total campaign cost, and divide by that cost. Incremental means the difference against matched holdout stores that did not run the campaign, not the difference against last year. Total cost means hardware, installation, network work, weekly content operations and the integration effort to join play logs to point-of-sale data.

What is a realistic in-store advertising ROI?

There is no credible benchmark figure, and any single number quoted without its controls, pre-period and cost basis should be treated as marketing material. The result depends almost entirely on the gross margin of the featured item and on whether content operations were funded. Run a holdout test in your own stores; the answer is specific to your margins and your footfall.

Does in-store digital advertising really increase sales?

It can, and the only way to know for your stores is a holdout test. Run the campaign in some stores and not in comparable ones, compare the same weeks, and check featured-item units against category units. If the category stays flat, the campaign moved a choice rather than creating a sale, which may still be worth doing on margin grounds.

Do I need consent to send offers to a shopper’s phone in my store?

In the European Union, yes. Regulation (EU) 2016/679 requires a lawful basis for processing location data, and Article 5(3) of the ePrivacy Directive 2002/58/EC requires consent before accessing information on the shopper’s own device. In California, precise geolocation is sensitive personal information and consumers can limit its use. Keep analytics consent separate from messaging consent.

How many stores do I need for a valid lift test?

Around twenty per group is a working minimum for detecting a single-digit lift, matched on weekly footfall, format, catchment and pre-period category sales. Fewer stores can only detect effects large enough to be visible without a test. Check that both groups tracked each other for the eight weeks before launch; if they did not, the matching is wrong.