Reading KeHE and UNFI movement data in SPINS

Why this matters

KeHE and UNFI movement data in SPINS is the syndicated coverage mechanism for the long tail of independent natural retailers — the single largest analytical surface in the natural channel after the direct-scan retailers. Reading it well is a skill; reading it as if it were direct POS data is a common analyst mistake.

A natural-products brand watches its KeHE/UNFI movement number in SPINS swing from +18% one week to -22% the next. Distribution didn't change. Velocity at the direct-scan retailers (Sprouts, Natural Grocers) was stable across both weeks. Yet the KeHE/UNFI cut is behaving like a teenager.

The temptation is to call it noise and ignore it. The actual answer is more useful: the KeHE/UNFI numbers in SPINS aren't measuring what most analysts assume they're measuring. They're measuring distributor shipments to retailers, not retailer sales to shoppers. And shipment data has structural quirks that direct-scan data doesn't.

This page covers what those quirks are, when they matter, and how to read KeHE/UNFI movement data without being misled by the artifacts.

What KeHE/UNFI movement data actually is

For the broader context, see What is SPINS data? and Syndicated vs. panel data. The KeHE/UNFI-specific summary:

SPINS' coverage of the independent natural retailer channel — the long tail of regional natural co-ops, single-store naturals, and specialty grocers that don't license POS individually — comes from distributor-flow data. The two relevant distributors:

• KeHE Distributors — major natural / specialty food distributor, large footprint across natural retailers.
• UNFI (United Natural Foods, Inc.) — major natural / specialty distributor, historically the primary distributor to Whole Foods Market though that relationship has shifted, with KeHE expanding share at WFM in recent years.

What SPINS receives from these distributors is a record of what was shipped from the distributor warehouse to each retailer, by SKU and week. SPINS uses this to estimate retailer sales — but the intermediate quantity is shipment, not sale.

The implicit assumption: over a long enough window, what's shipped ≈ what's sold. Retailers don't endlessly accumulate or shed inventory. Week-over-week, that assumption breaks more than analysts expect.

The four quirks that move the number

1. Retailer inventory cycles

A regional natural chain might order four weeks of stock in one shipment to take advantage of a distributor promo, then not order again for a month. The KeHE/UNFI movement read shows a huge spike in the order week followed by zero or near-zero for three weeks. The underlying consumer sell-through was steady the whole time.

This is the single biggest source of week-to-week noise in distributor-flow data. Smaller retailers tend to order in larger, less-frequent batches, which produces lumpy movement reads at the individual-retailer level even when the consumer demand is flat.

2. Distributor pricing promo windows

When the distributor runs a deal-of-the-month, retailers opportunistically stock up. The shipment number spikes for the month, then drops in the following month as retailers work through inventory. This is real economic activity but it isn't consumer demand change — it's a working-capital decision by the retailer.

3. New-retailer ramp and de-ramp

A new independent natural co-op adds the brand. The first shipment to that co-op might be 2–3 weeks of inventory (initial stocking) rather than one week of throughput. The KeHE/UNFI movement shows an elevated number for the first month or two as retailers ramp the brand into their inventory cycles. Reading those first weeks as "sales velocity" overstates the actual consumer pull.

The same applies in reverse — a retailer that delists the brand shows a final order, then nothing. Past performance at that retailer continues to drag the trend for a month even after the delist hits.

4. Backfill from distributor data delivery

KeHE and UNFI deliver data to SPINS on schedules that aren't always weekly. When a data delivery is delayed (vacation, system issue, holiday week), the affected weeks show zero. When the data arrives, several weeks of activity backfill at once. Read naively, this is a trough followed by a spike. The underlying activity was normal the whole time.

The diagnostic is the same one Reading SPINS panel coverage outlines for direct-scan suppression: did ACV change with the sales drop? If yes, it's likely a real movement; if no, it's a delivery artifact.

Reading KeHE/UNFI movement data correctly

Five practices that filter the noise:

1. Use a 4-week rolling average

Distributor-flow data at the individual retailer or single-week level has too much variance to read directly. The right unit is 4-week rolling — long enough to smooth the order-cycle lumpiness, short enough to catch real demand changes.

A brand reading "KeHE/UNFI flat 4-week rolling" over a 13-week period is reading correctly. Looking at the same brand week-by-week will show big swings that aren't there in the rolling read.

2. Cross-check against direct-scan retailers

The direct-scan retailers in the Natural channel cut — Sprouts, Natural Grocers, regional naturals above the SPINS coverage threshold — are measuring POS. They're noise-free relative to distributor flow. If a brand's Sprouts and Natural Grocers reads are stable but the KeHE/UNFI cut is swinging, the swing is distributor-flow artifact, not a real demand change at the long-tail naturals.

3. Separate KeHE and UNFI when possible

KeHE and UNFI cover overlapping but distinct retailer sets — KeHE has historically been stronger in specialty and gourmet, UNFI in core natural, with shifts over time. Reporting them separately catches retailer-set-specific issues (e.g., a UNFI delivery delay affects one half of the long tail but not the other).

In practice, SPINS' standard Natural channel cut blends the two; the separate distributor views are available in custom cuts.

4. Treat the absolute dollar level cautiously

Distributor-flow shipments to retailers don't perfectly equal retailer sales to shoppers. SPINS' projection methodology bridges the gap, but the bridge has uncertainty. Use distributor-flow movement for trend rather than absolute level when the absolute number matters.

5. Flag promo periods explicitly

For any promo run by the brand or by the distributor in the analysis window, the surrounding 4–6 weeks of KeHE/UNFI movement are contaminated by inventory effects. Annotate those weeks in the report and use them with caution.

Worked example — the apparent volatility

A wellness brand's weekly KeHE/UNFI movement reads across 8 weeks:

Reading the raw KeHE/UNFI column week-by-week: chaos. Numbers range from $28K to $68K — a 2.4x swing.

Reading the 4-week rolling column: gradual, stable, $42K–$53K range — roughly flat trend, no real demand change.

Reading the direct-scan column: stable across all 8 weeks, $179K–$184K. Consumer demand is flat.

Reconciliation: the KeHE/UNFI week-to-week variance is order-cycle and delivery-cadence artifact. The brand's actual demand in the independent natural channel is stable. A team reading the raw KeHE/UNFI weekly numbers would have spent the quarter chasing phantom movement. The team reading the 4-week rolling alongside direct-scan would have correctly concluded the brand was steady.

Where KeHE / UNFI movement data is the only signal

The trade-off matters but isn't one-sided. Distributor-flow data is the only way SPINS — or any syndicator — sees the long-tail independent natural retailers at all. The single-store naturals, the regional co-ops, the specialty grocers below the SPINS direct- scan coverage threshold: none of them license POS individually. Without KeHE / UNFI flow data, the natural channel cut would effectively be Sprouts + Natural Grocers + the handful of regional chains above the SPINS threshold — and the long-tail half of the channel would be invisible.

Four specific questions where distributor-flow is the better tool than direct-scan:

• Distribution breadth across independent naturals. Direct-scan doesn't see the long tail. Whether the brand is shipping to 200, 500, or 1,200 independent natural retailers is a number only distributor-flow can produce.
• Regional distribution patterns at the long-tail level. KeHE and UNFI flow data captures regional shipment patterns to independents that no other source sees.
• Distribution onboarding velocity for a new SKU. When a brand launches a new SKU into the natural channel, the distributor-flow data is the first read of which independent retailers picked up the SKU. Direct-scan won't see it until the product moves through the larger chains.
• Detecting brand-level distribution loss before it shows up at the chain. Independent retailers sometimes pull a brand before the chain buyer does. A trend break in KeHE / UNFI flow can be an early warning that direct-scan won't surface for another month or two.

For these questions, the distributor-flow noise that this page spends most of its time helping you filter is actually the information — what's "noise" for weekly demand reads is "signal" for distribution-trend reads.

Anti-patterns

• Reading single-week KeHE/UNFI movement as a sales signal. Too noisy. Use 4-week rolling or accept that you're going to chase artifacts.
• Comparing distributor-flow ACV to direct-scan ACV directly. ACV computed from shipment data has different timing than ACV computed from POS. A brand looking "down in ACV at the long-tail naturals" might be looking at a shipment lag, not a delisting.
• Aggregating KeHE/UNFI dollars with direct-scan dollars without noting the source mix. The combined Natural channel total is the right cut for most reporting, but when accuracy matters (board reports, investor reads), separate the two and flag the distributor-flow share of the total.
• Treating distributor-flow lift on a promo as comparable to direct-scan lift. A promo at a KeHE/UNFI-distributed retailer shows up in shipment data as retailer-stockup-then-sell-through; the lift profile is spread across 4–6 weeks rather than concentrated in the promo week. Direct-scan lift on the same promo at Sprouts is concentrated in the promo week. Comparing the two lift profiles without normalization will mislead.
• Assuming KeHE and UNFI distributor relationships are static. WFM-related KeHE/UNFI dynamics have shifted in recent years. Other major retailer-distributor pairings also evolve. Check the current relationship structure before doing a longitudinal comparison that crosses a relationship change.

Doing this in Scout

Scout exposes the KeHE/UNFI movement as a separate column from the direct-scan retailers in the Natural channel cut, with a default 4-week rolling option in the report view. The KeHE-vs-UNFI split is available where the SPINS extract contains it. Anomalies in the distributor-flow column that don't show up in direct-scan are visible side-by-side, which makes the "is this a real demand change or a shipment artifact" question a glance rather than a separate investigation.

Summary + further reading

• KeHE and UNFI movement data is distributor shipment data, not retailer POS. SPINS projects shipment to estimated sales, but the intermediate quantity has retailer-inventory-cycle quirks.
• Read at the 4-week rolling level, not single-week. Cross-check with direct-scan retailers to filter real demand changes from distributor-flow artifacts.
• Promo periods contaminate the surrounding 4–6 weeks of distributor-flow data; flag those windows explicitly.

Related: What is SPINS data? · Syndicated vs. panel data · Reading SPINS panel coverage