The First Thing Television Ever Showed Was a Dollar Sign

Philo Farnsworth’s investors wanted to know when the machine would pay. It took ninety-nine years to answer them.

In the spring of 1921, a fourteen-year-old named Philo Farnsworth was driving a harrow across a potato field outside Rigby, Idaho. Up one side, down the other, cutting the dirt into long parallel rows, for hours, the way farm kids did before anyone thought to ask them what they were thinking about.

What Farnsworth was thinking about was light.

He had been reading about the problem every electrical engineer in the world was chasing at the time: how to send a moving picture through the air. The leading approach involved spinning mechanical disks with holes punched in them, and it was terrible, and it was never going to scale. Farnsworth looked back at the field behind him and saw the answer sitting in the dirt. You do not have to move a picture all at once. You take it apart into lines, you throw the lines through the air one after another, fast enough that nobody notices, and you put them back together on the far end. A field, scanned by electrons.

He explained it to his chemistry teacher, Justin Tolman, on a classroom blackboard. Tolman kept the drawing, which turns out to have been the single most consequential act of teacher packrattery in American history.

Five years later, a pair of San Francisco money men put up the first real capital and set Farnsworth up in a loft at 202 Green Street. On September 7, 1927, his team transmitted a single straight line from one room to the next. He was twenty-one years old, and he had just invented electronic television.

The line was a miracle. The line was also, to the people paying for it, a line.

So the backers asked the question backers have always asked, and one of them finally said it out loud: when are we going to see some dollars in this thing?

Farnsworth put a dollar sign in front of the camera and sent that.

It is the best joke in the history of the medium, and it is also the entire argument of this article, compressed into a single 60-line video frame. The first commercially motivated image ever broadcast was not a face or a landscape or a sunset. It was a demand for ROI, answered with a punchline, in a rented loft, by a guy who grew up without electricity in the house.

The rest of the story is less funny. David Sarnoff sent RCA’s people out to San Francisco, offered $100,000 for the whole invention, and got turned down. Years of patent litigation followed. Farnsworth won in 1935, largely because a chemistry teacher in Idaho could still reproduce a blackboard sketch from 1922. RCA began paying him royalties in 1939, reportedly the first time the company had ever paid rather than collected. Then the war shut down television manufacturing cold, and by the time America started buying sets by the millions, his key patents had expired.

In 1957 he went on the CBS game show I’ve Got a Secret as “Doctor X.” The panel had to guess what he had done. They could not. He walked away with eighty dollars and a carton of cigarettes, having invented television, on television, anonymously.

He was not thrilled with how it turned out. He told his family he had created a monster and mostly kept the set off at home. Then in July 1969 he watched Apollo 11 land on the machine he built, and told his wife it had made the whole thing worthwhile.

Here is the part that should bother anyone who buys media for a living: the investor’s question never actually got answered. For the next ninety-nine years, television sold reach, frequency, gross rating points, and a brand lift deck that landed two quarters after the money was spent. Everything that visibly moved a deal forward moved in search and in email. TV got credit for “awareness,” which is the word an industry reaches for when it cannot produce a number.

That era is over. Not because television got more artistic, but because it got addressable, cheap, and instrumented. The dollar sign finally has a figure behind it.

Your mental price tag is off by four zeros

When someone says TV advertising, almost everyone pictures the same scene. A Super Bowl slot. A crew on location in Iceland. A celebrity with an agent, and an agency taking its cut for assembling the whole circus. Eight million dollars and a nine-month timeline before a single human being sees the spot.

That is one way to buy television. It has close to nothing to do with how streaming actually gets bought. Connected TV inventory sells programmatically, in auctions, the same way display and search do. You set a budget, pick an audience, and your spot runs inside Hulu or Peacock or Paramount+ against real viewers in real households. The floor is fifty dollars. The campaign can be live the same afternoon you decide to run it.

Creative used to be the other wall, and it was a tall one. A fifteen-second spot meant a shoot, an editor, a voiceover session, three weeks of revisions, and a five-figure invoice. Generative models have taken most of that apart. You describe the product and the audience, you get a finished spot, and you iterate on it over an afternoon instead of scheduling a reshoot. A small business can now test four different messages for less than a single day of production used to cost.

So the honest comparison is not television versus your entire marketing budget. It is television versus the two hundred dollars a day you are already handing Google to bid against every competitor in your category for the same twelve keywords.

The mid-funnel case is embarrassingly easy

Start with the simplest version. Someone visits your pricing page and leaves. Today you retarget them on Meta, where your ad sits next to fourteen others in a feed they are thumbing past at roughly the speed of sound.

You can also put that same person on Hulu on Tuesday night. Full screen, no skip button, sound on, in their living room, with their shoes off.

Same audience. Same intent signal. A completely different quality of attention. The pixel that fires on your pricing page builds a segment, that segment gets matched to streaming inventory, and a person who was already considering you sees you again in the one format nobody has figured out how to scroll past. That is consideration work. It has nothing to do with the top of the funnel, and it never did.

Upload a customer list and the channel changes species

Feed in a CRM file, and CTV stops being an acquisition channel at all. Churn-risk accounts are a segment. Customers who bought once and vanished are a segment. The creative for each of those looks nothing like your prospecting spot, because the job is nothing like prospecting.

Expansion is the same story. If you sell three products and most of your accounts have one, you have an obvious use for addressable television that never touches a new prospect in its life.

For B2B it gets more surgical. Run streaming against a target account list and time the flight to land in the same weeks your outbound sequence hits those same accounts. Nobody clicks a television ad. That is fine. The rep calling into a warmed account is having a different conversation than the rep calling cold, and the reply rate on the email moves. Farnsworth’s scanning lines, pointed at forty named companies.

Attribution is where teams talk themselves out of it

Here is how CTV shows up in your dashboard: branded search, direct traffic, and organic signups. If your model credits last click, Google collects the credit for demand that streaming manufactured, and the CTV line looks dead on arrival. So the team cuts it. Two months later cheap branded search volume falls off a cliff, and nobody connects the two events, because the two events are sixty days and one dashboard apart.

The fix is unglamorous, which is presumably why so few teams run it. Geo holdout. Pick matched markets, turn streaming off in half of them, compare total pipeline across both sets. Branded search volume and direct traffic are your fast indicators, since they move inside two weeks. Click-through rate on a television screen is close to meaningless as a signal, so if you are running lead gen, watch form fill rate by market instead.

The remote control is a conversion event

The newer ad formats close the gap on their own. A QR code or “Text CallMe to 93746” on screen, or a remote-control action that pulls up a form, gives you a direct line from an impression to something a person actually did. Which means the channel can finally be judged on cost per lead, the way you judge everything else you buy.

That changes who is allowed to buy television at all. A plumbing company does not care about reach curves. It cares whether the phone rings on Wednesday. When a spot produces a trackable lead for fifty dollars, television becomes buyable by people who were never, under any circumstances, going to fund a brand campaign.

What to actually do with the budget

Take streaming out of the brand budget and off the quarterly review. Put it on the same dashboard as paid search and paid social, with four audiences running at once, each mapped to a stage: prospecting, site retargeting, CRM winback, target accounts.

Then give it a measurement window that matches how it actually works. Search converts in hours. Streaming converts over weeks. Judging a two-week flight on same-day attribution will tell you to shut it off every single time, and it will be wrong every single time.

Ninety-nine years later

Farnsworth’s backers wanted to know when they would see some dollars in the thing. He answered with a gag, because in 1927 a gag was the only answer available. The machine could carry an image. It could not carry a number back.

It can now. A QR code or a Text-To message on a Peacock midroll is the same joke, told straight, with the receipt attached. Somewhere in Rigby, Idaho, a kid on a harrow is entitled to feel vindicated.