5 ms·

What kind of tool do you have to monitor traditional ad spend? According to op this should be impossible.

wtvanhest 10y ago
Most large companies use a quantitative model with various inputs and outputs. For example, if you know you spend X in market Y, but not market Z you can measure lift within some confidence level between the markets. Imagine a model with every market, and sku, plus the inputs also know where you spent ad dollars and on what product.
- robertlagrant 10y ago
  The "some confidence level" is the problem - how do you come close to quantifying and removing all the other effects that are involved? This stuff always just becomes something that can be debated away if the advertising lead is popular, or otherwise used as an excuse to fire him/her.
  - wtvanhest 10y ago
    Multifactor model
- busterarm 10y ago
  Definitely quantitative. And just like everything with advertising, you run experiments constantly. We also have this fun problem where nobody's actually been verifying that TV ads actually run when they're supposed to. We have physical offices in all of the locations we advertise for the most part, so we're talking about setting up slingboxes and programmatically verifying that our ads actually run.
  - trimbo 10y ago
    >We also have this fun problem where nobody's actually been verifying that TV ads actually run when they're supposed to Have you considered hiring Kantar to check if the ads are running? They do this.
    - busterarm 10y ago
      Thanks for the suggestion. We have a lot of markets that we'd have to check and I imagine it's cost prohibitive. It's actually a very easy thing to build. Moreover, we're kind of the big company in this space. We own many of the billboards we (and our competitors advertise on). We have multiple different platforms that we sell to our "competition". They're 10-20 years behind us in tech and the budget/desire to get there. It makes sense that we build this ourselves and keep it that way.
  - EwanToo 10y ago
    There's various companies that do this for you as a service, including verifying the reach across market research panels. If you want a fairly raw build-your-own solution you could speak to someone like TVEyes who'll give you an mpeg or mp3 feed from 100s of channels across the USA. If you insert a fingerprint into your advert audio, you can then auto-detect plays across the entire country.
ithinkinstereo 10y ago
Two ideas: 1) Coupons or discount codes in print and tv/radio ads (or using a unique URL, phone number, etc.) 2) Simply watching to see if leads traffic (however you define it: foot traffic, hits to website, calls, etc.) surges in the period post ad campaign (assuming you stagger your campaigns for better individual campaign tracking)
- busterarm 10y ago
  #2 is the right way to do it and how we do it. #1 isn't an option for us because we don't take upfront payment, or really any money directly, from our clients. Tons of companies try this and it's expensive and doesn't convert well for everyone. In certain kinds of sales pipelines this works, but many times it's just people searching for a coupon code online and you don't get accurate information about what converted.

Most large companies use a quantitative model with various inputs and outputs. For example, if you know you spend X in market Y, but not market Z you can measure lift within some confidence level between the markets. Imagine a model with every market, and sku, plus the inputs also know where you spent ad dollars and on what product.

robertlagrant 10y ago
The "some confidence level" is the problem - how do you come close to quantifying and removing all the other effects that are involved? This stuff always just becomes something that can be debated away if the advertising lead is popular, or otherwise used as an excuse to fire him/her.
- wtvanhest 10y ago
  Multifactor model
busterarm 10y ago
Definitely quantitative. And just like everything with advertising, you run experiments constantly. We also have this fun problem where nobody's actually been verifying that TV ads actually run when they're supposed to. We have physical offices in all of the locations we advertise for the most part, so we're talking about setting up slingboxes and programmatically verifying that our ads actually run.
- trimbo 10y ago
  >We also have this fun problem where nobody's actually been verifying that TV ads actually run when they're supposed to Have you considered hiring Kantar to check if the ads are running? They do this.
  - busterarm 10y ago
    Thanks for the suggestion. We have a lot of markets that we'd have to check and I imagine it's cost prohibitive. It's actually a very easy thing to build. Moreover, we're kind of the big company in this space. We own many of the billboards we (and our competitors advertise on). We have multiple different platforms that we sell to our "competition". They're 10-20 years behind us in tech and the budget/desire to get there. It makes sense that we build this ourselves and keep it that way.
- EwanToo 10y ago
  There's various companies that do this for you as a service, including verifying the reach across market research panels. If you want a fairly raw build-your-own solution you could speak to someone like TVEyes who'll give you an mpeg or mp3 feed from 100s of channels across the USA. If you insert a fingerprint into your advert audio, you can then auto-detect plays across the entire country.

ithinkinstereo 10y ago

Two ideas: 1) Coupons or discount codes in print and tv/radio ads (or using a unique URL, phone number, etc.) 2) Simply watching to see if leads traffic (however you define it: foot traffic, hits to website, calls, etc.) surges in the period post ad campaign (assuming you stagger your campaigns for better individual campaign tracking)

busterarm 10y ago
#2 is the right way to do it and how we do it. #1 isn't an option for us because we don't take upfront payment, or really any money directly, from our clients. Tons of companies try this and it's expensive and doesn't convert well for everyone. In certain kinds of sales pipelines this works, but many times it's just people searching for a coupon code online and you don't get accurate information about what converted.