The data everyone talks about vs. the data that actually moves the needle

Every ad company says the same thing:
“Data is important.”

But in digital advertising, “data” often means:

• audience segments

• broad demographics

• store lists

• estimated visits

• interest groups

Useful for targeting.
Terrible for driving real commerce.

In retail media, this gap becomes painfully obvious.

Because the only data that can actually change what people buy is:
transactional truth — POS.

This is the moment retail media stops being “digital billboards inside stores”
and becomes something completely new:
a commerce optimization engine.

To understand why, we need to stop thinking about POS as a reporting tool —
and start seeing it as the operating system of retail media.

2.1 The Power of First-Party Retail Data (With Numbers Behind It)

The most accurate, future-proof signal in the industry

The world is moving toward first-party data because it’s the only signal that will survive privacy changes.

And retailers possess the most valuable form of it:

• 100% verifiable purchases

• SKU-level precision

• timing accuracy within seconds

• store-level granularity

• basket combinations revealing behavior

• seasonal + neighborhood patterns

According to McKinsey:

• Retailers have 10–50× more purchase-relevant signals than digital ad platforms.

• First-party purchase data improves predictive accuracy by 30–70% across categories.

This is why over 80% of global retailers (Walmart, Tesco, Kroger, Boots, Aeon) now run or are building a retail media network.

But here’s the key point:
Having data is not enough.
Using it to change behavior is the real advantage.

Brands want to know:

• Which customers actually buy?

• Which stores drive the uplift?

• Which promotions work? Which don’t?

• What products grow when this SKU is promoted?

These are questions Google and Meta cannot answer —
because they do not see purchases.

But retailers can.

And ReceiptRoller builds directly on that foundation.

2.2 The Big Gap Between Impressions and Purchases

Clicks measure curiosity. POS measures behavior.

Digital advertisers long assumed:

“If someone clicks or views something, they’re interested.”

In reality:

• 78% of digital clicks show no correlation with purchase (Nielsen).

• Up to 40% of display ad impressions are never visible (ISBA).

• A shopper may ignore an online ad but buy instantly when they see a shelf.

Clicks describe noise.
Purchases describe truth.

And truth behaves very differently.

Examples from retail data we’ve seen:

• A drink category showed 0.3% CTR, yet 11% sales uplift when shown in-store.

• A snack product with a high video completion rate online showed no uplift at all without shelf visibility.

• A seasonal SKU purchased 70% of the time only after physical exposure — not because of any online signals.

This is why POS-backed retail media is not a small improvement —
it’s a paradigm shift.

Digital media measures what people look at.
Retail media measures what people do.

That difference drives results.

2.3 POS Is Not a Feature — It’s the Foundation

Importing POS for reports ≠ using POS as the decision engine

A lot of platforms say:

“We support POS — upload it to our dashboard!”

That’s reporting.
Reporting doesn’t change sales behavior.

ReceiptRoller does something different:
we feed POS data directly into the decision engine that chooses what to display, and when.

Our system evaluates four layers of context every few seconds:

1. Real-time commercial context

• Stock levels (updated every 1–10 minutes)

• Velocity of sales today

• Price elasticity curves

• Recent uplift signals

• Promotion windows

If inventory drops below a threshold (ex. 20 units),
the system automatically stops promoting the SKU.

2. Behavioral context

Using tens of thousands of transactions per store per month, we model:

• time-of-day purchase patterns

• cross-category triggers

• “basket partners” (products that rise together)

• repeat purchase intervals

For example:

• Showing coffee between 7:00–9:00 AM can increase pastry sales by 6–12%.

• A sandwich SKU often lifts beverage sales by 15–25% when paired.

3. Environmental context

We integrate real-world variables:

• weather (temperature, rain, humidity)

• time of week

• school schedules

• nearby events

Rain can increase hot drink sales by 20–40%
and reduce cold beverage sales by 10–30%.

4. Demand elasticity

Not all categories react the same way to advertising.

• Highly elastic categories can jump 20–50% with small pushes.

• Low-elastic categories respond only under specific conditions.

We remove noise, normalize signals, and focus on the features that drive real uplift.

This is not just “showing ads.”
This is real-time commerce optimization.

2.4 The Closed-Loop Advantage

(What Digital Advertising Has Tried and Failed to Solve for 20 Years)**

Most ad platforms dream of linking:

1. exposure

2. behavior

3. purchase

4. feedback

Digital platforms get stuck at step 1 or 2.

Retailers can see all four.

Here's what closed-loop measurement enables:

• Identifying which SKUs respond best to creative

• Measuring incremental sales per store, per hour

• Distinguishing “signal vs. noise” across campaigns

• Detecting emerging micro-trends (ex. flavor shifts in specific neighborhoods)

• Adaptive learning that improves with scale

In one pilot:

• A category-level campaign increased revenue by 8.4%,

• but only 5 of 23 SKUs drove 90% of the uplift.

This insight is impossible without POS linkage.

With our system:

• models improve every 24 hours

• uplift accuracy increases by 15–25% over time

• SKU-level forecasts become more reliable as data compounds

This is how retail media becomes self-improving —
and why scale accelerates value.

2.5 The Structural Moat: Why Competitors Can’t Catch Up

Tech advantages can be copied. Structural advantages cannot.

The advantage of POS-grounded retail media is not technical — it’s structural.

Here’s what makes it nearly impossible to replicate:

1. Retailers do not give POS data to ad networks

This ensures:

• privacy

• compliance

• strategic independence

And creates a permanent moat.

2. Data granularity compounds exponentially

After 12 months, a retailer may have:

• tens of millions of transaction rows

• SKU-level seasonality curves

• uplift patterns across hundreds of stores

• cross-category maps

• substitution likelihood models

Competitors starting later cannot “catch up” without the same historical depth.

3. Integration requires multi-year retailer cooperation

This is:

• operational

• technical

• political

• long-term

You cannot build trust in 3 months.
You cannot fix data quality overnight.
You cannot understand retail behavior without time.

4. Click-based DSPs cannot simply switch to purchase-based models

The entire architecture — pipelines, schema, attribution methods — is incompatible.

You cannot bolt causality onto a system built for clicks.

5. The flywheel strengthens over time

More data → better predictions → higher ROI → more retailer adoption → more surfaces → more data.

This is why retail media is not just another ad channel.
It’s a self-reinforcing commerce ecosystem.

And this is why ReceiptRoller is built on the only data that truly matters:

what people actually buy — and why.