The 2026 Geo-Panel Audit: What Mobility Coverage Honestly Looks Like Post-FTC

The honest 2026 answer to "how big is your mobility panel" is smaller than 2022 decks claimed: and the difference is structural, not marketing spin. X-Mode/Outlogic, InMarket Media, and Mobilewalla consent orders pulled sensitive-category supply, forced deletion of non-consented historical tranches, and tightened SDK consent lanes. Buyers evaluating global mobility must audit daily uniques inside their geographies, not headline device totals. Pair with data brokers post-FTC orders and foot-traffic panel sizing.

Procurement teams that skip panel math discover coverage gaps only after production joins fail. Run seed geography tests on identical polygons before comparing vendor narratives.

Panel Math: Historical vs Observable

Vendors quote cumulative MAIDs, lifetime devices, or global installs: metrics that do not predict daily uniques inside your POI polygons. The audit number is daily unique devices observed with documented consent in target geography during a representative week. Compare vendors on that denominator after sensitive-category exclusions apply.

How FTC Orders Reshaped Supply

Orders banned or restricted sensitive-category sales, required affirmative consent for advertising use of precise location, and mandated historical deletion: shrinking panels in step-functions. Vendors rebranding post-order without pipeline change carry residual risk. Read primary order text in diligence memos; see FTC location enforcement.

Where Replacement Supply Comes From

Consented mobile SDKs: smaller but cleaner; decay continues under Apple's App Tracking Transparency rules. Carrier-panel aggregates: larger reach, coarser granularity, different compliance envelope. CTV household resolution: complementary for exposure and cross-device, not a foot-traffic substitute, the same way the IAB Tech Lab cross-device measurement framework treats the two signals. Architecture should assign each lane a primary use-case rather than blending incompatible panels into one visit metric.

Four Questions That Surface Real Coverage

1. What is SDK or sourcing provenance by platform, and what was removed post-FTC?
2. What sensitive categories are excluded before the feed reaches us: show production logic?
3. What are daily uniques inside our polygon list last week, not global totals?
4. What deletion and opt-out proof exists for historical tranches orders required removed?

Geo-Panel Audit Workflow for 2026 RFPs

Attach polygon list or DMA list to RFP. Require DUAs-by-site tables, platform mix, hour-of-day coverage, and composition stability month over month. Score with RFP matrix and re-run seed match tests annually. Mobility coverage without governance controls is not institutional-grade.

Foot-traffic products live or die on panel honesty. Teams that document observable cohorts, exclusion policies, and geographic DUAs before signing avoid the QBR arguments that follow deck-inflated expectations.

Composition Tells You More Than Size

Panel headline counts invite a comparison that does not mean much, because two panels of similar size can behave completely differently depending on where the supply comes from. A panel concentrated in a small number of applications inherits those applications' user base, which is rarely a cross-section of the population and often skews by age, income, geography, and time of day. Size does not correct that, and a larger concentrated panel can be less representative than a smaller diversified one.

Ask how concentrated the supply is, specifically what share of observations comes from the largest contributing source. High concentration is not disqualifying, but it is a risk that belongs in the decision, because a single supply relationship ending can remove a large share of the panel at once. That is a continuity question as much as a quality one, and it is the reason two panels with identical current coverage can carry very different odds of still being comparable in a year.

Audit Against Your Own Footprint, Not a Sample Market

Vendor evaluation packs are built around markets where the panel performs well, which is a reasonable thing for a vendor to do and a poor basis for a decision. The audit that matters uses your own locations, including the rural sites, the quiet dayparts, and the regions you know are hardest, because those determine whether the data supports the decisions you actually make.

Request observation counts per location per period across your real footprint during evaluation rather than a national summary, then read the weakest decile rather than the average. Two vendors can look equivalent nationally and differ sharply where you operate. Where a vendor will not supply per-location figures for a defined evaluation footprint, that is itself informative, because any vendor able to serve the account can produce them.

Make Supply Changes a Notifiable Event

Mobility supply changes for reasons outside anyone’s control, and a panel that gains or loses a large source will show a step change in observed volume that looks exactly like a change in the world. Without notification, the first sign is usually an analyst investigating a movement that has no market explanation, some weeks after it began and after the affected numbers have circulated.

Write notification into the contract: advance warning where a change is planned, prompt notice where it is not, and a version stamp on every delivery so a series can be reconstructed later. Then keep a small set of reference locations whose real behavior you can verify independently, and check them whenever a headline number moves sharply. If the references move on the same date as the subject, the panel changed rather than the market. The foot-traffic panel sizing guide covers the depth side of the same question.

What a Panel Break Does to Trend Lines You Already Published

The audit above answers a point-in-time coverage question. The harder one arrives afterward. If a supplier deleted historical tranches to comply, the history behind your dashboards changed after you built them, and a year-over-year visit index now compares two different panels. The movement you are reading may be composition. Ask each supplier for the effective date of every consent-lane change and every deletion event touching the periods in your series, then mark those dates inside your own tables rather than in a slide footnote nobody opens.

There are two defensible treatments and one that is not. Restate the series, recomputing history using only the devices and lanes that survive today, which costs engineering time and yields a single comparable line. Or break the series at the change date and refuse to compare across it, which is cheap and honest. Splicing, where pre-change and post-change values share a trend line with no marker, is the one to reject. Ask which of the three the vendor did inside its own indexed products, because a normalized visit index hides whether the base was restated or simply continued.

Pick metrics that degrade gracefully while you sort this out. Absolute visit counts move with panel size. Within-period share metrics, such as a chain's share of visits across a defined competitive set, survive uniform shrinkage but not composition shifts, so they still break when one platform, one app category, or one region leaves the panel. That distinction is the diagnostic worth running. Recompute a period you consider stable on both the old and the new panel, then decompose the difference by platform and geography. If the gap concentrates in one lane you have a composition problem and share metrics will not rescue it. If it spreads evenly, rebasing at the break is enough. Re-run whenever seed match tests or an SDK partner change tell you the panel moved again.