This document discusses measuring the impact of TV advertising on digital engagement and conversions. It summarizes a methodology that evaluates minute-by-minute digital activity to establish a baseline and then measure the incremental impact of TV airings. TV spots are credited for a portion of subsequent digital actions in the hours after an airing based on modeling. Examples are given showing how the methodology provides insights such as which TV networks and device types drive the most response to optimize strategies.
What I learned from auditing over 1,000,000 websites - SERP Conf 2024 Patrick...
Measuring TV Effectiveness - TV Attribution
1. TV Attribution Case Study
Understanding the Impact of TV on Digital Engagement and Conversions
@caseycarey
2. Redefining Our View of TV Performance
FX
BRAVO
CNN
HIST
BBC
ESPNN
FNC
COMEDY
TV LAND
SPIKE
CMT
MSNBC
NBA
A&E
TNT
USA
TBS
TRU TV
NFL
DSC
IFC
HGTV
Effectiveness
Reach
3. 6am 8am 10am 12pm 2pm 4pm 6pm 8pm 10pm 12am
Digital Activity
Baseline
TV
TV
TV
TV
TV
TV
TV
TV
Measuring the Incremental Impact of TV Airings
How it Works
• Evaluate minute-by-
minute and hour-by-
hour activity
• Machine learning
establishes baseline
• Model incremental
impact of airings
4. TV Spot (T0) T1
TV
45%
Combining TV and Digital Conversion Paths
Direct
50%
T2
T0 to T1
= TV credited 45%
T1 to T2
= TV credited 35%
Direct
5%
TV
35%
Direct
15%
Display
25%
PPC
25%
$100
Display
25%
PPC
25%
$100
Display
25%
PPC
25%
$100
6. 1.6%
0.3%0.4%
1.0%
Long Form Brand Short Form Response
Category Search
Brand Search
Insight: Brand spots drive 4X greater interest in the
category; Align search strategies to response.
7. Insight: Mobile and tablet experience is critical to
effectively harvesting response from TV.
59%
30%
28%
51%
13% 18%
Baseline Response
Tablet
Mobile
Desktop
8. Insight: Underinvested in best performing
networks; Reallocate buys to match.
173
120
41
20
66
61
35
28
70
7
13
25
6
SPIKE
CMT
MS…
NBA
A&E
TNT
USA
TBS
TRU…
NFL
DSC
IFC
HGTV
Impressions (MM)
5.5
6.9
8
9
10.2
13
13.6
14.1
15.5
16.5
24.6
35.8
50.6
SPIKE
CMT
MS…
NBA
A&E
TNT
USA
TBS
TRU…
NFL
DSC
IFC
HGTV
Effectiveness Index
9. Insight: Creative 1 has a 30% higher brand
response and 4X greater overall response.
10.2%
7.8%
33.2%
18.8%
13.0%
3.1%
Creative 1 Creative 2
Brand Category Overall
1. Reach the right people.
2. Tell them a story.
3. Reconfirm that story over time. Build love and trust. Come back.
Borrowing pyramid from http://adwords.blogspot.com/2014/02/investing-in-better-measurement-for.html
Mass Media
Creates customer awareness
Drives customers to online and in-store
Digital Media
Captures customer segment not exposedto mass media
Drives website and in-store traffic
Supports mass media online response
Slide Objective
Key Talking Points
Direct nav (pointer icon)
Display (square display icon)
Paid Search (magnifying glance)
In addition to understanding the impact of TV on digital, it’s important to understand how TV fits into the full conversion path for a holistic view of attribution. If we were only to run the impact analysis of TV, we would still be over-counting conversion credit across digital and broadcast. The way that we accredit TV is by running a lift analysis across conversion paths at using probabilistic analysis to allocate a portion to TV. Here’s an example: In these paths, there is a direct visit, then a display view, then a search click, and finally a conversion, we would fractionally attribute value across these touch points using the methodology we just discussed. Now let's say that after running our TV attribution model we show that TV spot drove a lift in website visits during a particular period of time, essentially what we do is inject into that user-level path a kind of probabilistic event that's tied to that TV ad spot. So, the way to think about that is let's say that our model showed that there was a 40% probability that that website visit was driven by TV, what we then essentially do is take the fractional conversion attribution that would have been attributed to Direct Navigation, and we divide 40% back into the TV touch point (and I’m guessing this is divided over t1 and t2 based on decay?). This basically allows us to push down the TV insights into the Digital data, allowing the fractional attribution across both TV and the digital touch points and it allows for picking up some of the lag effect...where the TV spot may have driven website visits or a search click weeks prior to the actual conversion, but we're then able to tie that conversion event that happens much later back to that TV activity.
Green ad had 1.8M impressions – green bars show the lift for that
The yellow shows smaller ads – and you can see how the attribution is layered in