# Reading the shelf through point-of-sale and syndicated panel data
The category review deck is up on the screen. Your Nielsen slide says your brand grew 3.2% last quarter. The next slide, sourced from Circana (formerly IRI), says you shrank 1.1%. Same brand, same quarter, same shelf. The buyer across the table wants to know which number is real before she decides your promotional plan for next year.
Welcome to the most common and most misunderstood argument in FMCG data. Both numbers can be right. They measure different things, from different sources, using different math. Once you understand why, you stop panicking about the gap and start reading it for signal.
There are two fundamentally different ways to know what sold.
Measured (retail) data comes from store scanners. Every time a barcode beeps at checkout, that transaction is captured. Retailers share this data (directly or through providers), and it gets aggregated into what the industry calls point-of-sale (POS) data: actual units and dollars scanned at the register.
Panel data comes from people, not stores. A recruited panel of households scans everything they buy at home, or grants access to their loyalty and receipt data. Providers then project that sample up to represent the national population.
Nielsen (now NielsenIQ) and Circana both sell measured retail data AND household panel data. The classic "Nielsen vs IRI" disagreement usually comes down to different retailer coverage, different projection methods, and different category definitions.
Three culprits explain most gaps:
1. Retail coverage. If one provider has a data-sharing deal with a large club or discount retailer and the other does not, their totals differ. A brand skewing to club stores can look strong in one dataset and flat in the other.
2. Projection. Measured data still has to estimate stores it does not directly capture. Two providers using different store universes and weighting will project differently.
3. Category definition. Is a protein bar in "snacks," "nutrition," or "candy"? If the two providers slot your product differently, your share and your competitive set change.
None of these means someone is wrong. It means you must know what each number is counting before you quote it.
Here is the single most important concept for reading the shelf: ACV, or All Commodity Volume.
ACV measures distribution weighted by store size. A store's ACV is its total dollar sales across all products, not just your category. So a hypermarket counts far more than a corner store.
When your data says your brand has 60% ACV distribution, it does NOT mean you are in 60% of stores. It means you are in stores that represent 60% of total retail dollar volume. You could be in only 30% of physical stores but hit 60% ACV because you are in the big ones.
Split your sales growth into two levers:
A brand can grow sales purely by adding distribution while velocity quietly declines. That is fragile growth. When ACV maxes out (you are already in the big chains), the growth stops cold unless velocity improves.
Always ask: is this distribution growth or velocity growth? The category review deck that only shows total sales is hiding the answer.
A useful derived metric is sales per point of distribution (SPPD): your dollar sales divided by your % ACV. It normalizes for how widely you are stocked, so you can compare a new regional brand against a national one fairly.
SPPD = Total category-adjusted sales / % ACV distribution
Example (illustrative):
Brand A: $12M sales / 60 ACV points = $0.20M per point
Brand B: $12M sales / 30 ACV points = $0.40M per point
Same sales. Brand B sells twice as hard per point of distribution
and has more room to grow by expanding ACV.The next trap is volume. If your brand sells a 330ml can and your competitor sells a 2 liter bottle, "units" tells you nothing useful. One unit is not one unit.
Equivalized volume (often "eq" or "EQ") converts everything to a common base unit so you can compare fairly. In beverages it might be liters or "equivalized cases" (for example, a standard case of a set volume). In laundry it might be wash loads. In tissue it might be sheets.
Why it matters:
Rule of thumb: for anything involving price or true consumption, work in equivalized volume. For anything involving shelf presence and purchase frequency, units matter too.
Back to that deck. Here is how a sharp analyst reads it.
Step 1: Reconcile the sources. Ask which retailers each dataset covers and how the category is defined. If Nielsen includes a discounter that Circana omits, and your brand indexes there, the gap is explained, not alarming.
Step 2: Separate distribution from velocity. Pull ACV alongside sales. If sales rose 3% but ACV rose 5%, your velocity actually fell. The growth is borrowed from shelf expansion that will not repeat.
Step 3: Convert to equivalized volume for pricing. Check price per eq unit versus the category. A rising price per liter with falling volume share often signals you have priced yourself off the shelf.
Step 4: Cross-check with panel. Measured data tells you WHAT sold. Panel tells you WHO bought and whether they came back. A sales bump with flat repeat rate in the panel is a promotional sugar high, not a franchise gain.
For a solid free primer on how these metrics fit together, the Category Management Association resources and general retail analytics explainers are a good starting point.
🎬 [VIDEO: "Understanding ACV and Distribution Metrics in CPG" — youtube.com — a clear walkthrough of ACV, distribution, and velocity for consumer goods analysts]
Knowledge check
1. A brand's Nielsen data shows growth while Circana shows decline for the same quarter and shelf. What is the most professionally sound interpretation of this gap?
2. What is the fundamental distinction between measured (retail) data and panel data?
3. A brand sells disproportionately through club stores. One provider has a data-sharing deal with a major club retailer and the other does not. What effect would you expect on the two datasets?
4. Select ALL correct answers. Which factors commonly explain why two providers' numbers for the same brand and quarter diverge?
Select all the correct answers.
5. Select ALL correct answers. Which statements about projection in measured retail data are accurate?
Select all the correct answers.
Measured data is precise about transactions but blind to shoppers. It cannot tell you whether 1,000 units went to 1,000 new buyers or 100 loyalists stocking up.
Panel data answers the questions that decide long-term brand health:
A well-known principle from marketing science (associated with the Ehrenberg-Bass Institute) is that most brand growth comes from increasing penetration, not from squeezing more out of existing buyers. Panel data is where you verify that. If your brand is growing sales but penetration is flat, you are leaning on a shrinking loyal base, which is risky.
The catch: panel is a projected sample, so small brands and small regions carry wide error margins. Treat panel trends as directional, not decimal-precise, especially below a certain sales threshold.