How to measure billboard advertising effectiveness by location

To measure billboard advertising effectiveness by location, compare the stores a board can actually reach with similar stores that have no board, before and after the posting goes up. The gap, measured against a weighted twin of the board-free stores, is the lift. Everything else, from audience impressions to phone-location reports, counts people who passed the sign rather than people it sent to you.

Out-of-home is a large, growing channel. US out-of-home ad revenue reached a record $9.46 billion in 2025, up 3.6%, with digital formats at 36.3% of the total (Source: OAAA, via MediaPost, 2026). For multi-location brands, billboards are often the most local media they buy: one board, one road, one store. That makes them easier to measure than most people think, provided you set the test up around how boards are actually sold.

Why billboard effectiveness is hard to measure

Billboards have three habits that defeat ordinary reporting.

• They report audiences, not customers. The plant operator will quote weekly impressions, usually from Geopath audience ratings. That number estimates how many people were likely to see the board. It’s useful for pricing, and it says nothing about what those people did next.
• They don’t produce clicks. Vanity phone numbers and URLs catch a handful of people. Most drivers who act on a board just turn off at the next exit.
• They’re bought in blocks. Static boards are usually sold in multi-week posting periods, often four weeks, and good locations get locked into long contracts. You can’t flip them on and off like a search campaign, and directional boards (“Exit 14, turn right”) tend to stay up for years without anyone checking what they do.

None of this makes billboards unmeasurable. It means the measurement has to come from your locations, and the test design has to fit the posting calendar.

Start with the trade area, not the market

A TV or radio buy covers a whole metro. A billboard covers a corridor. So the first job is deciding which of your locations a given board can plausibly affect, and which ones it can’t.

• Distance. Draw each location’s trade area from your own data if you have customer ZIP codes, or use a drive-time radius if you don’t. A board well outside every trade area is a brand board, and it belongs in a market-level read.
• Direction. Inventory listings say which side of the road a board stands on and which way it faces. A board that faces drivers heading toward your exit is a different test from one that faces drivers who have already passed it.
• Overlap. If two of your locations share a trade area, treat them as one unit. Otherwise a board near one will quietly lift, or steal from, the other.

Three billboard test designs that fit posting periods

Pick the design that matches the decision in front of you. All three compare locations near boards with a weighted twin built from locations without them.

1. Matched sites: new boards near some locations

You are adding boards. Choose more candidate locations than you can afford to post, pair similar ones, and put boards near one of each pair. The rest become the comparison. This is the cleanest design, and the one in the worked example below.

2. Staggered start: boards go up in waves

You have committed to boards everywhere, but the plant can’t post them all at once anyway. Put half up first and the other half a few periods later. While wave two is still dark it’s extra comparison for wave one, and you get a second read when wave two goes live.

3. Go-dark at renewal: drop some boards and watch

This is the design for long-running directional boards. At renewal, keep most and drop a few, chosen by coin flip among similar sites. If customers at the dropped sites fall below their twin over the next few months, the boards were working. If they don’t, you have just found money.

Digital boards add a fourth option. Because the creative rotates on a schedule, you can often buy specific weeks or dayparts and alternate weeks on and off, even at one site. Treat that like any on/off test: decide the schedule in advance and run it for at least eight weeks.

What data you need to measure billboards by location

• Weekly outcomes by location. Transactions, guests, washes, new patients or booked jobs from your POS or booking system. A year of history is ideal, six months is the practical minimum.
• A board log. For each board: location, facing, the location it targets, posting dates by period, cost per period and the creative. Spreadsheets are fine. Memory is not.
• Everything else that changed. Price changes, local promotions, road works (a closed exit kills a board and a store at the same time), openings and other media that ran in only some trade areas.

Worked example: six new boards for a 30-site car wash chain

This example is illustrative. A 30-site express car wash chain posts one bulletin near each of six sites, each facing traffic heading toward the site. Each board costs $3,500 per four-week period, and the test runs four periods, so 16 weeks and $84,000 in total. The other 24 sites have no boards and form the comparison. The outcome is paid washes per site, per week.

The arithmetic for site A: 2,600 washes × 4.5% = 117 extra a week, × 16 weeks = 1,872. Its board cost $14,000, so $14,000 ÷ 1,872 = $7.48 per extra wash. Pooled, the six sites added about 6,461 washes for $84,000, or $13.00 each.

Now the judgement. Say a paid wash averages $15 and leaves about $11 after chemicals, water, labor and card fees. At $13 per extra wash the pooled program doesn’t pay on single washes. Sites A to C clearly do. Site F doesn’t, and when someone drives out to look, its board turns out to sit past the exit (in this example, the same mistake as Board B above). D and E are close calls whose value depends on how many of those extra customers buy a membership.

A sensible decision: renew A to C, move F’s face to the inbound side or drop it, and run D and E for another two periods with membership sign-ups as a second outcome. That’s a better use of $84,000 of learning than “billboards work, renew everything.”

Should you judge each billboard on its own?

Mostly no. One location’s weekly counts bounce around by several percent for reasons that have nothing to do with a sign on the highway. In the example, site E’s range runs from −0.7% to +5.5%. You can’t defend a decision about one board on that.

Decide whether billboards earn their place on the pooled result, then use site reads for two narrower jobs: catching boards that are plainly broken (wrong facing, blocked sightline, a closed exit) and choosing which sites to extend. If a board’s read is flat and nobody can find a physical reason, give it another two periods before you kill it.

What about mobile-location attribution reports for billboards?

Many out-of-home vendors now offer “attribution” reports. They take phones seen near the board, check which of them later showed up at your location, and compare that with phones that weren’t near the board. It sounds direct. It has the same flaw as most exposure studies: people who drive past a board near your store were more likely to visit your store anyway, because they live, work and commute nearby.

Use those reports to compare creative or check which board gets the most local traffic. Don’t use them to decide renewals. If the vendor shows you a large visit rate for exposed phones, ask what the same comparison looked like in the four weeks before the board went up. For more on why exposure doesn’t equal effect, see our guide to baseline versus incremental sales.

Common mistakes when measuring billboard effectiveness

• Reading before-and-after with no comparison. The board went up in April and washes rose. So did washes everywhere, because it was April.
• Picking test sites by hope. The sites operators lobby for are usually the strongest. Pair similar sites and let a coin decide.
• Ignoring which way the board faces. Facing and side of road matter as much as distance. Write them into the board log.
• Testing one board. One board at one site can’t produce a result you can act on. Six is a reasonable minimum for a pooled read.
• Stopping after one period. Four weeks is too short for a medium that works by repetition. Plan three or four periods.
• Letting the comparison sites see boards. A comparison location whose trade area overlaps a test board isn’t a comparison. Drop it.
• Never testing the old boards. Directional boards that have been up for years are the likeliest money to recover. A go-dark at renewal is cheap to run.

When billboard effectiveness can’t be measured by location

Some buys don’t fit this method. A metro-wide brand campaign with dozens of faces across the whole DMA reaches every location, so there’s no local comparison. Measure that at market level instead, with whole markets as test and holdout, the way you would TV. Our guide to matched-market tests versus geo holdouts covers the trade-offs.

And if all your locations sit on the same corridor, any board reaches most of them. Then a go-dark test across the whole corridor, with locations elsewhere as the twin, is the only honest option.

Questions people ask about billboard measurement

Billboards reward the brands that log every face, test in groups and are willing to drop the sentimental favorites. The road doesn’t care how long a sign has been up.

Written by The Countlift team at First Byte in The Woodlands, Texas. Questions or corrections? Email hello@countlift.com.