CTV zip-code targeting for dealerships, explained

Ask a rep how connected TV targeting works and you will hear the word precision several times and a number you cannot check. That is not much use when you are deciding which zips to buy and how much to put behind them.

The mechanics are worth understanding, because they decide what you can and cannot do. Most of the money wasted on streaming and CTV is wasted on a targeting decision made before the campaign ever launched.

This is the how. If you are still deciding whether the channel is affordable at all, that argument is a separate post.

How the targeting actually works

Broadcast is sold against a transmitter. The signal covers what it covers, and you buy the whole footprint whether or not those households would ever drive to you. That footprint is the DMA, and it is why television priced out single-point stores for decades.

Connected TV is not sold that way. The ad is delivered to a device over the internet, so the identifier is the IP address that the smart TV or streaming stick is connecting from. That address resolves to a household, and a household sits in a postal geography.

That one difference is the whole thing. You are not buying a coverage area any more. You are buying a list of households, and a zip code is just a convenient way to describe the list.

Be honest about the limits of that resolution. IP-to-household matching is good and it is not perfect. Apartment buildings share connections, some households route through a VPN, and a device that moves carries its targeting with it. Treat zip targeting as accurate at the neighborhood level rather than the doorstep.

Build the list from your own sales, not a radius

Most zip lists start as a circle. Somebody opens a map, draws twenty-five miles around the rooftop, and hands over the zips inside it. That circle is a guess dressed up as data.

You already have the real answer. Pull every sold unit from the last twenty-four months, group by the customer's zip, and sort by count. That list is your trade area as it actually behaves, not as it looks on a map.

Two things usually fall out of that exercise, and both of them cost money if you never run it.

• A zip inside the circle that has never bought a car from you. Often there is a river, a bridge with a toll, or a competitor sitting between you and it.
• A zip well outside the circle that over-delivers, almost always because a highway makes the drive shorter than the map suggests.
• A zip that only ever buys one kind of unit, which is a message decision as much as a targeting one.

Drive time beats distance

Nobody has ever decided where to buy a car by measuring a straight line. They decide by how long the drive feels on a Saturday.

When you are sorting the list, rank by drive time rather than miles. Two zips the same distance out can be twenty minutes apart in practice, and the further-feeling one will underperform no matter how much you spend against it.

This is the single most common correction we make to an inherited zip list, and it costs nothing to fix.

Frequency is what breaks a small zip list

Here is the trap that catches stores who do the targeting work properly. You tighten the list, waste drops, cost per completed view improves, and everyone is pleased. Then response falls off a cliff in week three.

Tightening the geography shrinks the pool of households you can reach. The budget does not shrink with it. So the same households see the same spot again and again, which is frequency, and past a certain point frequency stops building memory and starts building irritation.

This is the same arithmetic that burns through social creative in a small trade area, and it has the same fix: cap the frequency, and change the ad before the market can recite it.

A tight zip list and a single spot running for a quarter is not a targeting strategy. It is a slow way to annoy your best customers.

The households you cannot reach

Not every household in a zip is reachable on ad-supported streaming. Some subscribe only to ad-free tiers, some watch almost no television, and some sit behind a connection that will not resolve cleanly.

Do not chase full coverage. The temptation is to keep adding budget until the reach number in the platform looks complete, and all that does is push frequency up among the households you were already reaching.

Buy the reachable households at a sensible frequency, then put the next dollar into a different channel rather than into the same one harder.

What to ask before you sign anything

Connected TV is sold by a lot of people, and the quality of what you are buying varies more than the pitch suggests. Four questions separate a real geo buy from a repackaged one.

• Can I see the app-level report? Some inventory is premium streaming apps and some is a long tail of games and screensaver channels. Both count as CTV. Only one of them is somebody sitting on a sofa.
• What is the minimum geography you can actually buy? A rep who says zip and then delivers at metro level has sold you a DMA buy with extra steps.
• How do you handle frequency across the campaign, and can I set a cap? If the answer is vague, the frequency will not be.
• What happens to the households that fall outside my zips? On some platforms the answer is that spend quietly leaks to them to hit delivery targets.

Do not buy the same household twice

One more trap worth naming, because it is invisible in every dashboard. If you are running connected TV and YouTube and social video against the same tight zip list, you are frequently reaching the same person on three screens and paying three times for it.

That is not automatically wrong. A buyer who sees a consistent message in three places forms a stronger impression than one who sees it once. It becomes wrong when nobody planned it, because then the frequency you think you set on one channel is really three times that in the household.

The fix is not complicated. Decide the total number of times you want a household to see you in a week, then divide it across channels deliberately rather than letting each platform optimize its own delivery in isolation.

Measuring a geo buy without fooling yourself

The platform will report completion rate, and it will look excellent. Completion rate tells you the ad played to the end. It does not tell you anybody drove to your store.

The honest read on a zip-targeted buy is a zip-level one. Take sold units by zip for the ninety days before the campaign and the ninety days during it, and compare the targeted zips against the ones you deliberately left out.

Say plainly what that comparison is and is not. Sold units at a single rooftop are small numbers, and small numbers are noisy. A quarter is a signal, not proof, and anything shorter than that is mostly weather and payday timing.

That is still a better instrument than a completion rate, because it is measured in the unit you actually sell.

Where to start

You do not need a sophisticated plan to run this properly. You need a defensible list and the discipline to leave it alone long enough to read.

Take your top ten zips by sold units. Hold back one or two comparable zips as a control. Run one clear message with a frequency cap, and give it a full quarter before you judge it. Size the spend against units and cost per unit rather than against whatever the rep quoted, which is what the ad budget calculator is for.

Then treat the zip list as a living document. Trade areas move when a competitor closes, a bridge shuts for a season, or a new employer opens forty minutes away. A list built once and inherited three general managers later is just a circle again.