Tracking Your Television Marketing Performance

person looking on their smartphone while watching tv

Does TV impact marketing performance? And if so, how?

Every marketer with a budget large enough to support a healthy TV media buy has been asked this question by a CFO. The two follow up questions are:

“What is my point of diminishing returns?”

“If we had to cut your budget by 20%, what would you cut?”

Up until 2015, my canned answers were all I needed to get through a budgeting conversation:

  • “Yes, it does work.”
  • “It drives awareness which in turn increases digital volumes – eventually leading to sales.”
  • “How do I know? Let’s go dark and find out how much it impacts our sales.”

That was typically why TV budgets grew 5%-10% each year even as digital investments started to grow. Then came attribution.

Now I can provide much better answers when planning for future investments. The path to those answers is quite simple. Generate a baseline, monitor variance to baseline, and overlay station logs. While it’s not quite the same as having a phone number to track, it provides visibility into offline activity on online behavior. Notice I say behavior, not simply web traffic. After a few years of tracking website volume, I started to realize that’s not the goal. The goal is driving potential sales, not traffic.

couple watching television

Setting a Baseline

There are many decisions that must be made when creating a baseline and, unfortunately, each client is different. For discussion’s sake let’s assume a competitive set, marketing mix, and overall purchasing behaviors are relatively consistent. There are those who review performance minute by minute over a 30-day view and those that prefer minute by minute over the previous 8 weeks. I prefer the second as it eliminates weekend vs weekday noise and holidays are easily removed from the calculation. That being said, we will compare the website traffic at 1 PM on Thursday over an 8 week period to get an average return. Let’s say it is 100 visitors. We will observe the following Thursday to review the variance. Hopefully, we see 120. We will overlay the traffic for the following Thursday to see the traditional media execution that led to the increase. Easy enough. Now let’s go look for spikes.

If all things are working well, we see various spikes throughout the day indicating something is driving people to the site at greater rates than expected. Next, overlay all the airing times of your traditional media. Let’s hope the spikes are created within a 20-minute window of the TV/radio spots aired!

Spot Airing & Online Activity

How much time should we observe between spot airing and online activity happening? Should it be immediate or should there be a slight lag of :30 seconds? How long should we utilize as a window for capturing unattributed activity? Is it 15 minutes, 30 minutes, or an hour? How to decipher between the TV commercial on ESPN vs the radio commercial on ESPN radio that happened 3 minutes apart?

As I mentioned, it differs by client and product. What I have found to be most informative is the following:

  • Start observation at the exact time the spot begins
  • The impact lasts 20 minutes (short term repeatable result)
  • Commercial impact:
    • Either divvy traffic up by the size of the spot (impression delivery) and provide partial credit pending proximity to spot airing with a sliding scale toward the 20-minute timeline.
    • Or, which is my favorite, provide credit to all stations. Sounds crazy, but what develops over time is that the low-cost, high-frequency spots show poor delivery whenever the low-frequency, high-cost spots are not present to provide impact. The low-frequency, high-cost spots seem to have a very similar impact each time.

Combining the traffic impact divided by the cost of spot starts to generate a narrative of the station-by-station performance. The cherry on the top is that it delivers variance of performance by daypart. Now I have my media buy ready to go, and I know which spots are present when there is a spike or even when there is not.

Reviewing Website Traffic

Now the real kicker – let’s go back and review the website traffic that we generated within 20 minutes by the network. Is it following the same journey as your eventual customers? Are they visiting the same pages in a similar order? Probably not.

What I find is the large burst of traffic is typically people who are not ready to buy. We didn’t just remind them to place an order, we informed them of our existence. They are here to learn more, not purchase. Compare the incremental traffic to your average traffic and you can determine the right message for retargeting to this TV audience.

Articulating the Impact of TV Spots to Leadership

It’s nice to look a CFO in the face and be able to answer their questions with certainty:

  • In Q4 of last year, TV had a 23% incremental impact on web traffic which led to an observed 8% increase in sales over the same time period.
  • The point of diminishing returns was at 83% of budget implementation which delivered a 1% decrease in impact for the next 10%, and the final 13% observed a 22% decrease in return
  • If you were to eliminate 20% of the TV budget it would impact the overall drop in sales of 2.7% in Q4 if the media mix remains consistent
  • I would eliminate the HGTV, TLC, and A&E from the TV buy as they accounted for .01% of sales increase, traffic generated poor user journeys and soaked up 14% of the total budget. The remaining 6% would be isolated to daypart investment by the network.

If you’re not able to answer your CFO with this type of certainty – give PIN a call, because our clients can. And they didn’t even have to go dark!