Foot-Traffic vs Credit-Card Panels: When to Use Which Signal for CPG and Retail Reads

CPG and retail buyers face two purchase-adjacent panels: mobility visits and card transactions. Both cost money; neither substitutes for the other. Wrong panel for the question is the segment's most common mistake. This framework covers measures, failure modes, triangulation, and procurement. Pair foot-traffic panel sizing, CPG transaction deep dive, and CPG 71M DAU signals.

What Each Panel Actually Measures

Foot-traffic sees a consented device near a POI, giving visit and dwell, sometimes origin-destination. Cannot see basket, payment, or loyalty attribution. Browse-without-buy equals buy visit in mobility-only read. Global Mobility at scale.

Card panels (issuer/processor settlement) merchant, MCC (merchant category code, per Federal Reserve Regulation II interchange fee standards), amount, time. Invisible: cash (the Diary of Consumer Payment Choice reports 6 cash payments out of 47 per month), browse-only, gift purchase at A vs redemption at B.

Where the Two Panels Systematically Diverge

Divergence concentrates where visit-transaction and merchant-of-record loosen:

• Cash-heavy (c-store, QSR, small format) card undercounts.
• Gift cards: purchase at issuer merchant, redemption visit elsewhere.
• Loyalty redemptions: zero/odd tender vs normal visit.
• Bill-pay and money orders at mass retail: transaction not category insight.

Triangulation: Why Buyers License Both

Same chain, period, geography: card transactions ÷ foot-traffic visits ≈ conversion. Flat revenue + visits up 8% → conversion down about 7% (1 ÷ 1.08): pricing or assortment signal invisible in either panel alone. Multi-panel triangulation is the standard way to reduce single-panel bias.

Align market-level scaling; cash-heavy categories distort ratio: document caveats in QBR.

If You Can Only License One

Heuristic:

1. Foot-traffic (site selection, competitive visitation, trade area, RE/mall analytics) spatial question.
2. Card-panel (demand forecast, share-of-wallet, basket, pricing) financial question.
3. Neither alone for visit-to-conversion bridge: need both or misread market.

CPG: card for category share; foot-traffic for retailer-account reads. REITs: foot-traffic for tenant mix; card for lease sensitivity. Finance: often both. CPG Feed + Global Mobility + Real Estate Data complement.

Procurement Diagnostics for Both Panels

Foot-traffic: DMA depth, polygon vs centroid, dwell rules, raw-to-verified visit rate, panel sizing guide. Card: issuer diversity, MCC completeness, cash visibility in target categories, SEC alt-data alert on single-source concentration. Both: methodology-change log mandatory.

Score combined stack via RFP scoring matrix before dual-license.

Panel Churn Is the Silent Break in Both Series

Both panel types share a failure that rarely appears in vendor documentation: the membership changes underneath the series. Devices drop out when an app is deleted or a permission is withdrawn, cards leave when an issuer relationship ends, and new members arrive with different characteristics from the ones they replace. A month-over-month change can therefore be entirely a change in who was measured.

The protection is a same-panel comparison. Ask whether a reported trend was computed on members present in both periods, or on whoever happened to be in the panel each month. The first is a real change in behavior; the second mixes behavior with composition and cannot be separated afterwards. Where a vendor cannot produce a stable-panel view, treat sharp month-over-month movements with suspicion, particularly around any date the vendor changed supply.

Panel Fit Is Category-Specific, Not Universal

Buyers tend to ask which panel is better, when the answerable question is which panel is better for a category. Foot traffic performs well where the visit is the commercial event and the location is discrete, and poorly where a store sits inside another venue or shares a footprint. Card panels perform well where the transaction is card-led and merchant identification is clean, and poorly where purchases route through marketplaces, third-party delivery, or a parent company whose merchant descriptor tells you nothing about the brand.

Test fit on your own categories before generalizing from a vendor case study, which will naturally have been chosen from the categories where the panel performs best. Take a handful of brands whose real performance you know from another source and check whether each panel reproduces the ordering and the direction of change. A panel that gets the ranking right is useful even when its levels are off, and a panel that cannot rank your category correctly will not become reliable through a larger sample.

Reconcile on a Schedule, Not After a Surprise

Most teams compare panels against internal truth once, during evaluation, and then run on the panel alone until something looks wrong. By that point there is no way to tell whether the divergence appeared last month or has been growing all year, and the investigation starts without a baseline. A standing reconciliation removes that problem for very little effort.

Pick a stable subset where you hold ground truth, your own locations or your own sales, and recompute the relationship between panel and truth every period. Track the ratio rather than the raw gap, since a stable ratio means the panel is tracking correctly even when it under-counts, which is the normal and workable case. A moving ratio is the early warning, and it usually appears a month or two before anyone notices a problem in a headline number.

Reconciling the Two Panels Before You Divide One by the Other

The conversion ratio only means something if both sides describe the same stores. Card panels key on the merchant of record, which is a settlement descriptor, and a single retail brand often settles under several of them (corporate units, franchised units, licensed departments, a fuel position on the same lot). Mobility keys on geometry and brand hierarchy, and its store list usually still contains locations that closed mid-period. Ask each vendor for the store list it actually observes rather than the list it claims to cover, then build the crosswalk at store level where descriptors allow and at chain level where they do not. Write down which stores sit in each denominator, because that is the assumption a QBR argument eventually turns on.

Time alignment is the second reconciliation. Visit timestamps are local and event-driven. Card records are frequently keyed to settlement rather than authorization, so weekend and holiday activity batches forward and Sunday lands on Monday. Measure that offset on a quiet week before you trust any daily read, and compare weekly until you have. Late arrivals deserve the same caution, since a panel that backfills for days makes the most recent week look soft, which reads as a conversion drop that reverses at the next refresh.

The third is restatement. Both panel types re-weight, gain and lose contributing sources, and change expansion factors. A ratio computed across such a change measures the change, not the shopper. Require a methodology-change log with effective dates from both vendors, and ask the question that matters more than the log: is history restated on the new basis, or spliced? Then compute the ratio for a period where you hold ground truth, your own point-of-sale or a retail partner's reported comparable sales, and check whether the error sits still or drifts with season. A ratio that is biased but stable still reads trend. One whose bias moves is a chart, not a measurement. Score both feeds on the same RFP matrix rows before committing to a dual license. To see per-chain foot-traffic reads for major brands before you license anything, start with the free foot-traffic tools.