There are two ways to measure it: device-based attribution through a pixel on your site, and direct-response mechanics such as vanity URLs and promo codes. Neither has a hard cutoff you must clear. Your site traffic and your spend concentration decide how good the read will be, not whether you are allowed one.
The two measurement approaches
Device-based attribution: devices exposed to your units are matched against site visits through a pixel and compared to a control group. This is the rigorous method. It captures everyone who visited after exposure, however they got there, and it reports lift.
Direct-response mechanics: vanity URLs, campaign landing pages, and promo codes capture visitors who deliberately typed what they saw. Simple and legible, but structurally undercounting, because most people who see a billboard and later visit go straight to your main site or search your brand name.
Which one should you use
Both, for different jobs. Device-based attribution answers whether the campaign worked. Vanity URLs and promo codes give you a visible signal that stakeholders intuitively trust. Just never judge the campaign on vanity URL volume alone; it is the tip of the iceberg by design.
Which pages to pixel
There is no limit. The baseline is your homepage plus the page that represents the conversion you care about: checkout, confirmation, form submission, signup complete. Two pixels answer most questions.
Pixels can coexist with tags you already run for radio, podcast, or TV vendors. There is no conflict.
Site-wide, for web attribution
Web attribution measures exposed-device site visits across your whole site rather than counting specific events, so it needs the pixel on every page. If you want both, install site-wide and designate the conversion pages.
Why more pages is not automatically better
There is no cap, and there is a tradeoff anyway. Pixeling deep into the funnel shrinks the measurable pool, because fewer exposed devices reach a page three clicks in. Pixeling everything makes it harder to isolate what OOH specifically influenced.
Choose the pages that correspond to decisions you would act on. That is usually fewer than the number you first think of.
How much site traffic you need
Site traffic sets how good the read will be, not whether you are permitted one. The read gets noisier as traffic falls, in four recognisable bands.
500,000+ monthly visitors: roughly 20,000 a day. The clean, recommended line, with full depth and accuracy.
250,000 to 500,000: attemptable, with a warning. Expect reduced accuracy and less depth in the cuts.
100,000 to 250,000: attemptable, with a stronger warning. Treat the output as directional.
Under 100,000: effectively inconclusive. We will tell you so rather than sell you a study.
Below the top band you will be asked to acknowledge the added risk before web attribution is added to the campaign. Causal lift on a daily event file has no traffic requirement whatsoever and is usually the better route. For lower-traffic sites, branded search volume and direct traffic trends in exposed markets often give a cleaner read than page-level analytics.
Two spend gates, and they are not the same gate
$25,000 in total campaign spend qualifies a campaign for analytics access at all. This is a commercial guideline rather than a methodological limit, and it exists to keep non-viable campaigns out rather than to ration reports.
$25,000 per market is what web attribution wants in each market being measured, because the exposed set is built market by market.
A $75,000 campaign clears the first gate comfortably and clears the second one in three markets, not in fifteen. Spend concentration matters as much as the total: the same budget across three markets measures better than across fifteen.
Why the minimums exist at all
Not to ration access. Device coverage is partial, typically a 10 to 15 percent match rate, so a small campaign produces a small and unstable sample. The result is not unavailable, it is unreliable, which is worse: a number you cannot trust is more damaging than no number. Minimums are about whether the output will survive scrutiny.
What the report gives you back
Attributed visits, conversions, conversion rate, how many times someone was exposed before converting, how long they took, and which units performed best. Market-level views compare exposed against holdout geographies.
There is no published conversion-rate benchmark to compare your number against. A benchmark study is being scoped; until it exists, read your rate against your own other markets and your own other campaigns rather than against an industry figure.
Setting it up
Three planning-stage tasks: the pixel installed before launch, the vanity URL or promo code decided before creative is produced because it has to be printed, and baseline traffic captured so there is something to compare against.
Common misconceptions
Not "there is a cap on how many pages can be pixeled, as with some radio and podcast vendors." There is no cap. The constraint is analytical, not technical.
Not "there is a site traffic threshold you have to clear." Traffic is a quality ladder, not a cutoff. Lower traffic means a noisier read, not a closed door.
Not "there is no minimum spend for attribution reports." The $25,000 minimum is on campaign spend and it gates analytics access generally.
Not "small campaigns cannot be measured at all." Brand lift and causal lift both work at small scale. It is web attribution and device matching that need volume.