Billboards, Radio, Print and Sponsorships: When Offline Marketing Makes Sense for a Startup

1. Blog
2. Demand & Channels

By Saroj Jha · October 24, 2026 · 26 min read

Offline media sells reach in a place or in a publication, and it is easy to overpay for people who will never buy. Give each offline buy one job, price it per buyer you care about rather than per person who passes by, and set up the measurement before you sign. The nine steps below show how, with one example company from start to finish.

This guide is for founders and marketers at Seed to Series B startups who have been offered a billboard, a radio package, a trade-magazine ad or a sponsorship. I wrote it to give you a test for when offline fits, a plain account of how each channel is bought, a way to work out the cost per buyer from any rate card, a measurement plan that works without clicks, a scorecard for sponsorships, and the contract and advertising-rule basics to check before you sign. Events have their own guide, Conferences and Trade Shows for Startups, and online ads are covered in the Digital Advertising Playbook.

Running case: Example Co. Fictional

Example Co. is a seed-stage B2B SaaS company. It sells scheduling and dispatch software to heating, cooling and plumbing contractors with 5 to 50 technicians. Its 140 customers are concentrated in three metro areas, and the buyer is usually the owner. Annual contract value is about $6,000. In one month the founder is offered four things: a digital billboard near a trade-supply district, a sign package with a local minor-league baseball team, a sponsorship at the state contractors’ association meeting, and host-read ads on a podcast for contractors. Example Co., its names, prices and results are all invented.

Step 1: Decide which job offline media should do

Offline media (billboards and other out-of-home, radio, print, and sponsorships) is usually bought for one of four jobs:

• Reach in one place: many people in one area at once.
• Local presence: being seen around town, so a local market knows you are there.
• Credibility: appearing in places and publications your buyers already trust. Whether that works for your company is something to test, not assume.
• Category awareness: teaching a market that a type of product exists before buyers know what to search for.

Offline media is sold by place, format or publication audience, not by named account, and a billboard or radio spot produces no click, so measurement takes more work (Step 7). The controlled evidence I could find is also thin. A 1990 review in the Journal of the Academy of Marketing Science concluded that the limited field-experiment evidence suggested outdoor advertising “is likely to increase sales”. But it found that “only two published and no unpublished studies” used true experimental or quasi-experimental designs, and it called for more field studies^[3]. A review that old, resting on two studies, is weak evidence either way: it doesn’t show that billboards work today, and it doesn’t show they don’t. That is a reason to run your own test, as Step 7 describes.

Example Co., step 1 Fictional

The founder writes one job per offer. Billboard: awareness among contractor owners in one metro. Baseball team: local presence. Association meeting: conversations with owners. Podcast: credibility with owners who already listen. Only the association meeting is directly about pipeline, so the team will hold the other three to a stricter measurement standard.

Do this

Write one sentence per offer: “This buy is for ___, among ___, and I will count ___.” If you can’t fill in the second blank with your buyer, stop.

Step 2: Is a billboard worth it for a B2B startup?

Three situations make a billboard worth pricing. Each one is about your buyers being concentrated in a place:

1. Dense buyer geographies. Your buyers work in a small area, such as a trade-supply district, an industrial park or a hospital campus, so a board on their route could reach a real share of them.
2. Event cities. A major conference fills one city with your buyers for a few days. A board near the venue or on the airport road is a short, timed buy.
3. Recruiting. A board near where engineers or clinicians work is aimed at candidates. Measure it on applications, not pipeline: applications per role in the area before and during the buy, and how many applicants say where they heard of you.

The limits:

• Most passers-by are not buyers. Vendor audience numbers count the people who pass, not the buyers among them (Step 5 shows the arithmetic).
• A board is read from a distance. OAAA defines billboards as displays for viewing from distances “generally more than 50 feet” (Step 3), so write for a glance, not a page.
• There is no click. You need a geo comparison and indirect signals (Step 7), and small counts swing by chance.
• One board is a small sample. The sizing table in Step 7 shows how large a swing a small area needs before chance stops being a likely explanation.

Before you price a billboard, write down three things: the place your buyers pass, your estimate of the share of them who pass it, and the second metro you will compare against. If you cannot fill one in, the test in Step 7 cannot run.

Example Co., step 2 Fictional

The board sits on a road into a trade-supply district where many contractors buy parts, so it passes the “dense geography” test. Example Co. also has customers in two other metros, which can serve as comparisons. The founder keeps the billboard on the list, subject to the cost check in Step 5.

Step 3: How is out-of-home advertising bought and measured?

Out-of-home (OOH) is any advertising people see outside the home. The US industry association, the Out of Home Advertising Association of America (OAAA), groups formats into four categories^[4]:

• Billboards: “Standardized large-format advertising displays intended for viewing from extended distances, generally more than 50 feet.”
• Street furniture: “Advertising displays, many that provide a public amenity, positioned in close proximity to pedestrians”, such as bus shelters and kiosks.
• Transit: “Advertising displays affixed to moving vehicles or positioned in the common areas of transit stations”.
• Place-based: OAAA’s description is “Just about anything you could imagine.” Its list includes arena and stadium advertising, cinema, gas stations and shopping malls. (OAAA lists airports under transit.)

Digital billboards are a shared format. OAAA notes that digital bulletins “offer static messages that rotate every few seconds with typically six to eight advertisers”^[4]. You are buying a share of the screen’s time, not the board.

How audiences are measured. OAAA calls Geopath’s Audience Measurement Ratings “a cornerstone of OOH metrics”^[5]. Geopath states: “Geopath measures audience in terms of people, not cars”^[6]. Its estimates draw on inputs including a mobile device panel, road network data, points-of-interest data and population demographics^[7]. Its three core metrics are the likelihood-to-see impression, audience composition, and reach and frequency, and the same page names target rating points as a measure calculated from them^[8]. On the likelihood-to-see page, Geopath says that “the larger the space that the media takes up within the audience’s field of view, the more likely the content is to be seen”^[9].

What to ask any OOH vendor before you price anything

• Is the unit measured, and by whom? Ask for the answer in writing, with the measurement period.
• What are the impressions for my target audience, not adults 18+? A general-audience number tells you about traffic, not buyers.
• For digital, how many advertisers share the loop, and how long is my share?
• What will I see as proof? Photos of the live ad, a log of when it ran, and a credit if it did not.
• What does production cost, and who approves the creative? Production can be billed separately, as on the East Carolina University bus-shelter page in Step 5.

The two small local rate cards opened for Step 5, a campus bus shelter and a county transit system, give no audience figure, so for inventory like that you price on your own estimate.

Example Co., step 3 Fictional

The digital board is measured. The vendor quotes 400,000 likelihood-to-see impressions over four weeks for adults 18+, with seven advertisers in rotation. The founder asks for the audience by business owners, and the vendor has no breakdown that narrow. That gap becomes an assumption in Step 5.

Step 4: How do radio, podcasts and print work?

Broadcast radio is priced by spot length, daypart and number of spots; the one station rate card I opened, in Step 5, lists drive time and broad rotation, with a lower price per spot for more spots and a weekly package. Ask for the audience estimate for your target, the daypart it applies to and a written schedule. That rate card gives a price for each daypart and no audience for any of them.

Host-read podcast ads are read in the host’s own words. Ask where the read sits in the episode, whether it stays in the episode permanently or is inserted for a set period, and how the show counts a download. Podcasts for B2B Startups covers when to guest, sponsor or host.

Print and trade publications sell ad sizes priced per insertion, and some also sell newsletter placements; the AASA kit in Step 5 lists both. Ask for circulation, recipients by job title, and how the numbers are verified.

Example Co., step 4 Fictional

The contractors’ podcast offers four host-read ads, one per weekly episode, attached permanently to each episode. The founder asks the host to try the product first, so the read can be honest (Step 9).

Step 5: What does offline advertising cost per buyer reached?

The only cost that matters is the cost per buyer you care about. Rate cards give a price per placement. Turn it into a cost per thousand (CPM), then into a cost per target buyer.

The arithmetic:

• CPM = price ÷ (impressions or copies ÷ 1,000)
• Target CPM = price ÷ (impressions among your buyers ÷ 1,000)
• Cost per buyer reached = price ÷ number of distinct buyers likely to see it

Real rate cards, as examples only. These are prices published by the organisations themselves, which I opened on 24 October 2026. They will change, and they are not benchmarks:

• Trade print (EdTech). AASA’s 2026 media kit for School Administrator lists a four-color full page at $7,505 for one insertion and a print circulation of 20,000. The publisher says it reaches “20,000 educators nationwide who specify, recommend or have the final say” in district purchasing^[1]. That’s about $375 per thousand copies, or about $179 per thousand if you use the stated total readership of 42,000. The “20,000 educators” description is the publisher’s own; ask how it is verified.
• Trade newsletter (EdTech). The same kit lists a leaderboard in the weekly News of the Nation at $5,700 per quarter, sent to 8,900 subscribers with a 48.8% average open rate^[1]. The kit says the newsletter is sent weekly but doesn’t state how many sends the quarter includes, so the cost per thousand opens can’t be computed from the page. Each send opens at about 4,300 (8,900 × 48.8%), so the cost per thousand opens is $5,700 ÷ (4.3 × the number of sends). Ask for the send count in writing.
• Association newsletter (MedTech, clinic operations). MGMA’s 2026 media kit lists a weekly leaderboard in MGMA Insights at $1,750 for non-members. The kit gives two audience figures: “more than 77,000 healthcare professionals” receive the newsletter weekly on page 18, and “47,000+ Newsletter Subscribers” on page 7^[2]. That’s $23 or $37 per thousand, depending on which you use. When a media kit gives two numbers, ask which one is in the contract.
• Radio. WYSU’s sponsorship page lists 15-second announcements at $22 each in drive time and $18 in broad rotation for the first 13 spots, falling to $19 and $14 for 27 to 52 spots. A 12-spot weekly news package is $228 a week for a flight of 1 to 13 weeks. The page claims “more than 35,000 weekly listeners” for the station as a whole, but gives no audience for a daypart or package, so you cannot compute a CPM from it^[12]. The page shows no effective date for the rates and says they are subject to change.
• Local street furniture (consumer or campus). East Carolina University’s transit system lists one off-campus bus shelter ad, one side, for one semester at $750, with design and printing billed separately^[11]. LeeTran, the Lee County transit system, lists shelters at $300 a month for one unit, falling to $225 a unit for five, benches at $150 a month, and a six-month contract for bus ads, on a card effective January 1, 2022^[10]. Neither page gives an audience figure, so you can’t compute a CPM. You have to estimate the audience yourself.

Two things stand out in the rate cards. First, the price of an audience varies by more than a factor of ten between formats before you ask whether the audience is yours. Second, price is tied to commitment: the radio page drops from $22 to $19 a spot as the flight grows from 13 to 27 spots or more, and the transit card steps down with more units and requires six months for a bus. That is why Step 9 asks what a short first flight costs against the volume price.

Example Co., step 5 Fictional

The digital board costs $8,000 for four weeks and delivers 400,000 impressions: a $20 CPM, which looks cheap. Now the target arithmetic. The metro has about 2,000 contractor businesses. The founder assumes 15% of their owners pass the board in four weeks (300 owners), about four times each (1,200 target impressions). Cost per target impression: $8,000 ÷ 1,200 ≈ $6.67, or about $6,667 per thousand. Cost per owner reached: $8,000 ÷ 300 ≈ $26.67. For comparison, the state association’s newsletter offers a month (four sends) for $3,000 to 6,000 subscribers, with an assumed 40% open rate: 9,600 opens, or about $312.50 per thousand opens, nearly all of them contractors. Per buyer, the board with the lower CPM costs far more than the newsletter. The board stays on the list only as a measured test (Step 7).

Do this

For every offer, write the price, the vendor’s audience number, your estimate of the share that are buyers, and the resulting cost per buyer reached. Make the assumption visible so you can revisit it.

Step 6: How should I write a billboard or other out-of-home ad?

Out-of-home is made to be read from a distance (Step 3). I found no primary study that sets a word limit, so I give none. Test the message instead:

• One idea. Write the one thing you want a passer-by to remember before you write the line, then run the real ad through the Five-Second Message Test to check that one idea comes through.
• Legible at a distance. Print or display the artwork at the size and viewing distance the placement will have, and check that someone who has not seen it can read it.
• Something to trace. Use a vanity URL or code that appears on this placement only, so Step 7 can count it.
• Radio and podcasts. The listener cannot reread, so have someone listen once and write down the URL or code. If they get it wrong, change it.

Example Co., step 6 Fictional

The board reads: “Every tech, every job, one screen. Example Co.” The vanity URL, examplego.com/phx (fictional), is used only on this board, so any visit to it can be traced to the board.

Step 7: How do I measure offline marketing without clicks?

Offline media rarely produces a click, so read a comparison and several indirect signals together.

1. Geo holdout or matched-market test. Run the buy in one area and compare it with similar areas where you don’t run it. Compare the change in each area against its own baseline before the campaign. Google researchers describe geo experiments in which “non-overlapping geographic regions are randomly assigned to a control or treatment condition”^[13]. Incrementality Testing for Startups covers the design and the difference-in-differences arithmetic. One billboard in one metro is not randomized, so treat it as a directional matched-market comparison. Keep everything else steady in both areas: don’t change digital spend, pricing or sales activity in the test metro during the window, or you can’t tell what moved the result.

2. Search lift. Offline ads can send people to search engines. A study in Management Science found that TV advertising for financial services brands increased related Google searches and searchers’ tendency to use branded keywords. It estimated an elasticity of total brand searches to TV advertising of 0.17^[14]. That was TV, not billboards or radio, so it supports the method of tracking branded search, not the effect of any offline channel. Compare branded search in test areas against comparison areas.

3. Direct traffic and demo requests by area, before and during the campaign. Set this up before launch with the Tracking Plan and UTM Builder.

4. Self-reported attribution. Add a free-text “How did you hear about us?” field to your demo or signup form, and code the answers consistently with the Self-Reported Attribution Coder. Leave the field open rather than a drop-down, so a person can write “saw a billboard on Route 9” or “heard it on the contractors’ podcast”. Treat the answers as one signal among several and read them with the others.

5. Promo codes and vanity URLs, one per placement. A code or URL records only the people who use it, so read the count as a floor, but each use is direct evidence that a placement reached someone.

How big a swing counts as more than chance?

Small offline tests fail most often on size, not on design. Demos arrive in small whole numbers, and small counts swing by chance. As a rough check, the table below compares the test area’s count with its own baseline average, using a Poisson distribution, a standard model for counts of events that arrive independently. It shows the count a test area needs before chance alone becomes an unlikely explanation (a 5% chance or less), and how often a real lift of 50% or 100% would clear that bar. This is arithmetic that AAJ ran, not a published finding. It ignores the comparison areas and any trend, so it shows the least you can detect, not a full analysis.

Example Co., step 7 Fictional

The team takes eight weeks of baseline in all three metros. Then it runs the board for four weeks and keeps reading for four more. Demo requests per four weeks: the board’s metro goes from 12 to 18 (+50%), the two comparison metros together from 20 to 22 (+10%). The difference in differences is about 40 percentage points, or about 5 extra demos (12 × 0.40 = 4.8), at roughly $1,700 per extra demo ($8,000 ÷ 4.8). Three of the 18 demo forms mention “billboard”, and the vanity URL got 41 visits. With a baseline of 12, a count of 18 or more turns up by chance about 6 times in 100, and 19 or more would be the first count below 5 in 100, so the swing is suggestive but not proof. The team reads it as “not disproven” and decides to retest in a second metro before buying more boards.

Do this

Write down before you sign which areas are test and comparison, which signals you will read, for how long, and what result would make you stop or continue.

Step 8: Should we sponsor a local sports team, an event, a newsletter or a podcast?

Rate every sponsorship, whether a local sports team, community event, conference, newsletter or podcast, the same way, against the job you set in Step 1. Fit matters, and explaining the fit can help. A 2025 meta-analysis of 57 effect sizes from 31 studies, covering 15,744 participants, found that “sports event-brand fit has a positive effect on consumers’ cognitive, emotional, and behavioral response”^[15]. Three cued-recall experiments on sponsorship memory reported that sponsors that naturally match an event have a memory advantage. They also reported that explaining the link between sponsor and event (“articulation”) can improve memory even for sponsors that do not match, though the gain was affected by the presence of competing brands and by how memory was accessed^[16]. Both are consumer research. The meta-analysis is about sports sponsorship and measures consumer responses, and neither study measures B2B buying, so treat them as support for the idea of fit, not as proof that a B2B sponsorship will pay.

The sponsorship scorecard. Rate each offer on five criteria as strong, partial or weak, using your own facts, and write the evidence next to each rating. I give no weights and no pass mark, because I have no source for any. Decide yours from your own best channel, and write down why before you rate the offers.

Then decide from the pattern: sign an offer that is strong on fit and measurement and has an activation plan, run a smaller test where an answer is partial, and pass where fit is weak or you can’t measure it.

Activation beyond logo placement. A logo on a banner is a placement, not a plan. Write the activation plan before you sign. Ways to activate:

• host a small session or breakfast for the buyers in the audience;
• offer something useful that uses the product, such as a free schedule review;
• book meetings with attendees before the event;
• give the organizer a story about why you sponsor (the “articulation” in the research above);
• turn it into content: a recorded talk or a joint guide with the organizer.

For conference sponsorships, the Conferences and Trade Shows guide covers cheaper ways to show up.

Example Co., step 8 Fictional

The team rates four offers against the job in Step 1. Cost per qualified contact is the price divided by the founder’s estimate of qualified conversations (invented here). The association meeting is the only offer rated strong on fit, activation and measurement, so the team signs it, with a breakfast for owners as the activation. It runs the newsletter, whose activation is weak, and the podcast as small tests with their own URLs. It declines the baseball team for demand generation, where fit, activation and measurement are all weak. If the job were hiring technicians for its own support team, the founder would rescore it with audience fit defined as local job seekers.

Should we sponsor a local sports team to build awareness? Only if the job is local presence, recruiting or community goodwill in a market where your buyers live, and you have an activation plan. Ask the team to show who attends, and rate it on audience fit and cost per qualified contact like any other offer. If your customers are local people, as for a consumer app or a clinic marketing to patients, the fit question is easier to answer yes.

From offer to decision in four weeks

Offers can arrive with a deadline. A short process keeps you from deciding under pressure:

Step 9: What should a sponsorship or media contract include, and which rules apply?

This section is orientation on contract terms and advertising rules, not legal advice.

Contract basics to ask about:

• Deliverables: every asset in writing, with dates, placements, number of spots or reads, and proof of performance (photos, airing logs, episode links).
• Make-goods: what you get if something doesn’t run as agreed, such as replacement exposure or a credit.
• Exclusivity: whether you are the only sponsor in your category, and whether you are restricted elsewhere.
• Creative and production: who approves creative, the deadlines, and whether production is included (it was not on the bus-shelter page in Step 5).
• Term and renewal: the length of the commitment, what a short first flight costs against the volume price, and any automatic renewal.
• Contact data, cancellation and payment: what data you receive and on what consent basis, notice periods and refunds.

Rules to know about:

• Broadcast sponsorship identification. FCC rules require a broadcast station that airs matter for which “money, service, or other valuable consideration is” paid or promised to announce that it was sponsored^[17].
• Endorsements and host reads. The FTC’s Endorsement Guides include a podcast example: “Near the beginning of a podcast, the host reads what is obviously a commercial for a product.” The example says listeners would likely still expect that the podcaster was compensated, so a disclosure of payment is not needed for that commercial. It adds that “the audience may believe that the host is expressing their own views in the commercial”, and in that case the host must actually hold the views expressed^[18].
• Claims need a basis. The FTC’s 1984 policy statement on advertising substantiation says advertisers and agencies must have “a reasonable basis for advertising claims before they are disseminated”, and that a claim that states a level of support, such as “tests prove”, needs at least that level^[19]. A billboard line is an advertising claim: if the board says “cuts scheduling time in half”, have the evidence first.
• Roadside signs. Federal rules aim at “effective State control of outdoor advertising in areas adjacent to Interstate and Federal-aid primary highways”, and states may impose stricter controls^[20]. Local sign rules may also apply; ask the media owner which permits cover the board.
• Health claims. If your product makes health claims, the FTC’s Health Products Compliance Guidance says that “advertising” in the guide “refers not only to traditional TV, radio, print, and internet ads” but more broadly to marketing and promotion methods, and it covers health-related products including health equipment^[21]. If you sell a medical device, have your regulatory adviser check every offline claim before it runs. The MedTech Marketing Playbook covers this in more depth.
• Schools. If you are offering anything to a school or district, read that district’s own policy first, and see How to Sell EdTech to School Districts.

This is general information, not legal, financial or medical advice. Check with a qualified professional.

What changes with your volume?

• A few demos a month. The sizing table in Step 7 shows that a small area needs a large swing to be readable. Give each placement its own URL or code (Step 7), and compare cost per qualified contact with your digital channels.
• More demos in several areas. A smaller swing becomes readable, so a comparison between areas gets more useful. The MMM-Lite Starter Kit helps fold the results into a media-mix view.

What mistakes should I avoid?

• Buying on CPM alone. In Example Co.’s case (Step 5), a $20 CPM became $6,667 per thousand among contractor owners, against $312.50 per thousand opens for the newsletter.
• Buying before setting up measurement. Without a baseline and comparison areas, you can’t measure it afterwards.
• One board, one month, then a verdict. At small counts, only a large swing is readable (Step 7).
• Logo-only sponsorships. Without activation, you have paid for a name on a banner.
• Signing without a deliverables list. If it isn’t in writing, you can’t claim a make-good.
• Taking audience numbers at face value. Ask who measured them and which number is in the contract.
• Changing other spend during the test. If digital spend or pricing moves in the test area, the comparison no longer isolates the offline buy.