rethinking the rules of thumb

37
Is PrimeTime Worth it? 11/5/14

Upload: 0ptimusconsulting

Post on 15-Feb-2017

573 views

Category:

News & Politics


1 download

TRANSCRIPT

Is PrimeTime Worth it?

11/5/14

Introduction

• Much of when and where campaigns, issue advocacy efforts, and non-profits place their TV ads is based on a combination of rudimentary data (Nielsen), and experience (found both in campaign staff and media buyers). Let’s be very clear, there is a lot of value that has been produced over the years by leaning on these resources.

• That being said, after having worked with several campaigns on maximizing the impact of their TV buys, we have found several “rules of thumb” aren’t quite as “full proof” as you might think. In fact we have consistently found (across numerous geographies and race types) certain rules of thumb may be incorrect, over-generalized, or conditionally valid.

• While we can’t share our clients data or findings with you, we have taken the time to reproduce some of the key insights in a market where we did not have a paying client (Houston, TX). Rest assured, the cases we present in this deck serve as good examples of our experience across multiple markets, race types, and channels.

Executive summary

This analysis concentrates on three “Rules of Thumb” we frequently hear, and if they should hold sway

• Rule of thumb #1: To achieve penetration with TV ads, it is smart to buy primetime, because A.) primetime ads talk to a lot of people so they are worth a high cost B.)particularly desirable and elusive segments of the electorate we would want to target watch prime time, and C.) these ads hit the “large” group of voters that watch only primetime.

- Using analysis of data , we find:

A lot of prime-time is inefficient as a vehicle for delivering impressions relative to time other times of the day

The desirable segments of the electorate we want to talk to aren’t elusive, and primetime has no inherent advantage in reaching these segments

The number of people who watch only prime time is not “large”, its actually very very small

• Rule of thumb #2: Broadcast news (e.g. local news or NBC nightly news) is not only a great place to reach lots of viewers, you are also getting high concentrations of voters that we want to talk to on these programs.

- Using analysis of data, we find:

Some news is very efficient at talking to desirable segments of voters (and should be bought), some news is not (and should not be bought)

• Rule of thumb #3: Buy football as a rule. It’s got tons of live viewers, and a very high proportion of them are live-viewers who will actually see the ads (unlike a lot of other programming).

-Using analysis of data, we find:

We agree football does have big audiences and that the majority are live-viewers. That being said, the decision on whether an ad during football is worth buying demands a lot more nuanced analysis dependent on some key information, like what price do we have to pay? When in the football game will the ad air? Does the game we are going to buy follow another with the market’s home team playing in it?

Agenda

Prime-time Deep Dive

Broadcast News Program Deep dive

Football deep dive

Is prime time worth it?

Context

• Targeted-race campaigns currently spend the majority of their money on TV ad buys

• According to one popular Rule-Of-Thumb, to achieve penetration with TV ads, it is smart to buy prime time. The reasons given traditionally are:

- There are a lot of people who watch primetime, so its worth paying high prices for it

- Key elusive parts of the electorate that you want to target watch prime time

- Primetime is the only time to buy the “large” group of voters who watch only primetime

• The only problem with just going along with this Rule-Of-Thumb is that primetime is very expensive; a campaign can easily blow through the majority of its TV budget buying a few primetime spots a week

• In the slides that follow, we examine reasons that are traditionally given to buy prime time, and analyze the data to discover if they should hold weight

Prime time is a very expensive time to buy ads

• Consider the graph on the left: it is showing you the price your would pay on a Wednesday in October to buy an ad on NBC in Houston (Q4, 2014), by time of day from 5 in the morning to midnight. We have highlighted primetime in yellow.

• As you can see, the price to buy time on NBC is in the $200-$1000 range (for each 30 second spot) for most of the day, until is shoots up in price for afternoon news, and then for prime time.

• Note that prime time is costing a ton of money --$4000 per TV spot as at its peak.

Weds, NBC, Houston, 4QCost to buy Ad

Investigating Claim #1: buying primetime is justified, because A lot of people watch primetime, so that makes it “worth it”

• “Worth it” is a value statement. An ad is “worth it” if it delivers enough value to justify the cost. But what does “value” mean in our context?

• Campaigns are buying TV ads to persuade voters, so we care about getting our ad in front of voters so they can be persuaded

• To help us understand if an ad placement does this, we have now inserted a new line on the graph, in red. This line represents confirmed viewers on the channel

- What does “confirmed” mean? We have observed a set of 37k Houstonians and what their cable set-top-boxes are actually tuned to (literally, we can see what these boxes are tuned to every second of every day)

• So what do we make of buying at 11:30AM vs. 8:00PM?

WEds, NBC, Houston, 4Q

Note: Viewership figures above based on a second-by-second observed data set of 37,730 Houston set-top boxes

Viewers on channel

Cost to buy Ad

Investigating Claim #1: buying primetime is justified, because A lot of people watch primetime, so that makes it “worth it”

• We realize that if we buy at 11:30 AM, we are going to pay $200, and we are going to get 32k viewers to see that ad. This means we are paying six tenths of a cent for each impression we achieve at 11:30AM

• If however, we buy at 8:00PM, we realize that we are getting 32k viewers, but we are paying for $3500 for it. This means we are paying almost 11 cents an impression for each viewer who sees our ad.

• This means we are paying 18 times more per impression at 8:00PM than we are at 11:30AM

• Would you pay 18 times the price for a piece of persuasion mail if the mail man delivered it to the box at 8:00PM instead of 11:30AM? Probably not.

Note: Viewership figures above based on a second-by-second observed data set of 37,730 Houston set-top boxes

Weds, NBC, Houston, 4Q Viewers on channel

Cost to buy Ad

Investigating Claim #1: buying primetime is justified, because A lot of people watch primetime, so that makes it “worth it”

• What we are getting at here is efficiency -- it turns out, you can buy TV much more efficiently on some shows that others.

• We like to think of efficiency in terms of cost-per-impression (CPI). In the case of this example:

- 8PM buy: $3500 cost / 32,000 viewers = 10.9 cents

- 11:30AM buy: $200 cost / 32,000 viewers = 0.6 cents

• So we can say the CPI for buying at 8PM is 10.9 cents, and the CPI for buying at 11:30AM is 0.6 cents.

• But what about all the other times we could have bought during the day? Would those times have been efficient? How much would we pay per impression then?

Note: Viewership figures above based on a second-by-second observed data set of 37,730 Houston set-top boxes

Weds, NBC, Houston, 4Q Viewers on channel

Cost to buy Ad

Investigating Claim #1: buying primetime is justified, because A lot of people watch primetime, so that makes it “worth it”

• By simply dividing the price we would have to pay for each half-hour segment by the number of viewers for each half hour segment, we can quickly get to cost-per-impression for all timeslots throughout the day.

• Looking at this, we realize prime time is one of the worst times we could have bought this channel, in terms of efficiency.

• Why care about efficiency? If you are able to buy impressions at one quarter the cost of what you would have otherwise bought at, you can make the same money go four times further. That means you can buy four times the ads against your targets. That means you can show them more messages, or the same messages more often, for the same amount of money!

Note: Viewership figures above based on a second-by-second observed data set of 37,730 Houston set-top boxes

Weds, NBC, Houston, 4QCost per impression

Investigating Claim #1: buying primetime is justified, because A lot of people watch primetime, so that makes it “worth it”

• What this all means: lots of primetime is not “worth it”. We could take the same $4k we might spend on a prime time ad and put it behind lots of other buys that would yield lots more impressions on the voters we are trying to persuade

Note: Viewership figures above based on a second-by-second observed data set of 37,730 Houston set-top boxes

Weds, NBC, Houston, 4QCost per impression

Investigating Claim #2: buying primetime is justified, because lots of desirable and elusive segments of voters watch it

• To determine the validity of this second justification for buying primetime, we need to look at several types of voters we might want to target with ads (desirable segments), and determine if we could find them elsewhere (are they really elusive?)

• Let’s say we’re running a Republican primary race. One obvious desirable target would be Republican Primary Voters. Because we are observing individuals (37 thousand of them in Houston in this case), we can pick out only those that are Republican Primary Voters, and plot their viewership on the left in green.

• Or, what if we are running a targeted Republican vs. Democrat general election race? Well a desirable group to talk to would those who are not hard partisans. We could define that group as those voters we don’t see registered to a major party and or taken part in a primary where they pull a Republican or Democrat ballot. We plot their viewership in blue.

• Even at first glance, the claim is starting to look dubious. These desirable segments don’t seem that elusive.

Weds, NBC, Houston, 4QViewers on

channel

Investigating Claim #2: buying primetime is justified, because lots of desirable and elusive segments of voters watch it

Weds, NBC, Houston, 4QCost per

impression• We can calculate the cost-per-impressions for each

of these segments of the electorate, just like we did for slide 10, and we have plotted the results on graph to the left.

• Even when we take into account going after “desirable and elusive segments”, primetime still looks pretty inefficient.

• The takeaway: sub-segments of the electorate are not that elusive, primetime is not the only way to get to these voters on TV, in fact we show it can be a very inefficient way to get to these segments. Targeting based on efficiency would allow a given TV budget to achieve a lot more impressions on target voters.

Investigating Claim # 3: buying primetime is the only way to get to the “large” group of voters who watch only primetime

• To assess this claim, we need to take a look at prime-time viewers to determine if they are exclusively primetime viewers.

• How are we going to do that?

Weds, NBC, Houston, 4QViewers on

channel

John DoeJane DoeSally DoePete Doe

Investigating Claim # 3: buying primetime is the only way to get to the “large” group of voters who watch only primetime

Mon 2pm Mon 3pm Mon 4 pm Weds 8pm Weds 9pm Thurs 11AM Thurs 12AM Thurs 1PM... ... ... ...

NBCNBC

NBC OFF

NBCNBC

NBCTNT

OFF

HGTV

OFF

OFF

OFF

HGTV

BRVO

OFF

OFF

OFF

CNN

ABC

OFF

HGTV

OFF

OFF

OFF

HGTV

BRVO

OFF

OFF

OFF

CNN

ABC

The tuners are all knowing…

First, we collect all the tuners data for those tuners that turned on NBC in the prime-time window

highlighted on the previous slide

John DoeJane DoeSally DoePete Doe

Investigating Claim # 3: buying primetime is the only way to get to the “large” group of voters who watch only primetime

Mon 2pm Mon 3pm Mon 4 pm Weds 8pm Weds 9pm Thurs 11AM Thurs 12AM Thurs 1PM... ... ... ...

NBCNBC

NBC OFF

NBCNBC

NBCTNT

OFF

HGTV

OFF

OFF

OFF

HGTV

BRVO

OFF

OFF

OFF

CNN

ABC

A&E

OFF

FOX

OFF

TNT

OFF

NBC

OFF

TNT

OFF

NBC

FX

The tuners are all knowing…

Then, we look throughout rest of week to figure out if we could have put an ad in

front of them somewhere that is not prime-time broadcast

Investigating Claim # 3: buying primetime is the only way to get to the “large” group of voters who watch only primetime

All viewers

Primary Republican Voters

Non-partisan general election

only voters

Wednesday night NBC primetime Houston viewers broken out by how many non-prime-broadcast + cable segments (30 min) were watched that week • Contrary to popular belief,

the number of people who only watch primetime is very small, as we can see primetime viewers watch lots of other non-primetime segments throughout the week.

• The takeaway: the number of individuals who only watch primetime is very small, so this fails as a justification for it’s purpose

Agenda

Prime-time Deep Dive

Broadcast News Program Deep dive

Football deep dive

Campaigns & buying news on the broadcast Big 4: The status quo

Context

• When it comes to broadcast news (e.g. your local 6am news, NBC nightly news, your local late night news), campaigns often fall back on a “rule of thumb” which says that buying news is a good thing.

- The reasoning goes that those people who care about local news, care about what is going on in the communities/state/country they live in, and so are likely to vote. This means you are targeting your buys efficiently when you buy broadcast news.

• What this means is that campaigns, either explicitly or implicitly, instruct their buyer to buy news programs on the big 4 broadcast channels (FOX, NBC, CBS, ABC).

• In the slides that follow, we examine whether broadcast news is worth buying

Investigating Claim: you are buying efficiently if you buy broadcast news

• Consider the Houston NBC affiliate in Houston (KPRC), whose news programs and prices to buy spots during those new programs are shown in the chart to the left

• If this is all you see, it’s pretty hard to tell whether you should buy any off this, other than doing so because of the “Rule of Thumb” which says you should buy broadcast news

Investigating Claim: you are buying efficiently if you buy broadcast news

• But what if, using set-top-box data, we can understand how many viewers were actualy watching a given news program, on any day of the week at any time.

• Further, if we know a news program has 20,000 viewers, and it would cost us $800 to buy a spot during that news program, we can calculate that program’s cost-per-impression for all viewers. We now know the “value” of all programs relative to each other.

• But wait, a campaign doesn’t care about all viewers. This is an election, we care about the viewers we want to persuade.

- We don’t care about making impressions on non-registered voters

- We don’t care about making impressions on those voters who are registered but aren’t going to turn out

- We don’t care about making impressions on hard partisans who already know who they are going to vote for.

Investigating Claim: you are buying efficiently if you buy broadcast news

• So instead of figuring out how efficient it is to talk to “all viewers”, figure out how efficient it is to talk to just the voters a targeted R vs. D campaign cares about

• For example, we can create this new universe displayed on the right. Unlike “All Viewers”, this universe, called “Non-Partisan General Election Voters”, looks a lot more like what we want to target in a targeted R vs D election:

- Only registered voters

- Only voters likely to show up in the coming election

- Only those voters who are not hard partisans, because hard partisans already know who they are going to vote for regardless of what they are shown.

• Now we know exactly what an impression is costing us to talk to the voters we actually want to talk to. This allows to

- Select which news programs we buy based on efficiency. This allows us to maximize the impact of a given TV budget, by stretching dollar for maximum impressions on targeted voters

Example of good value

Example of bad value

Investigating Claim: you are buying efficiently if you buy broadcast news

• But what if this isn’t a R vs. D race, what if this is a Republican primary? How am I supposed to figure out which news programs to buy then?

• Answer: just define a new target, and solve for each news program’s efficiency

• On left, we have inserted Republican Primary Voter efficiency calculations.

Campaigns & buying news on the broadcast Big 4: The status quo

• We can now visualize this data, and see exactly what number of impressions on targeted voters a program will get us, for what price, and at what efficiency (cost per impression)

Co

st t

o b

uy

an

ad

on

pro

gra

m

Cost per impression on targeted voter paid when buying a program

= 1,000 Targeted Voter Impressions = 5,000 Targeted Voter Impressions

Note: some news programs were left off this chart because they were off-the-chart bad

Houston’s NBC News program Inventory – Republican Primary Targeting

Campaigns & buying news on the broadcast Big 4: The status quo

• We can now visualize this data, and see exactly what number of impressions on targeted voters a program will get us, for what price, and at what efficiency (cost per impression)

• We can now understand what most desirable inventory looks like, and select the best buys for the budget we have. On left, we have put the possible buys into categories using shades of green

• By the way -this trick (buying on value in terms of cost-per-impression-on-target voter) works on all TV programs, not just news. It is the basis of modern campaign impression buying.

• And of course, there is more than one channel offering news inventory..

Co

st t

o b

uy

an

ad

on

pro

gra

m

Cost per impression on targeted voter paid when buying a program

= 1,000 Targeted Voter Impressions = 5,000 Targeted Voter Impressions

Note: some news programs were left off this chart because they were off-the-chart bad

Be

st b

uys

Nex

t b

est

bu

ys

Dec

ent

Bu

ys

Stre

tch

bu

ys

Har

d t

o ju

stif

y

Houston’s NBC News program Inventory – Republican Primary Targeting

Campaigns & buying news on the broadcast Big 4: The status quo

• When we add in other ABC and FOX into the mix, we can very quickly see that there is plenty of inventory to buy at very good value

• Q: So what about CBS? Or what about a list of which programs to buy specifically (you didn’t label the bubbles!). A: This is part of what we provide campaigns who work with us, so hire us.

• Big takeaway: the rule of thumb that says “broadcast news is worth buying” is wrong. Some news is worth buying, some is definitively not worth buying.

- To separate out which is worth it and which is not, you have to drill down to cost-per-impression –targeted-voter.

• When you buy based on value, you can maximize the impressions a given TV budget will deliver you, and increase the odds your candidate wins

Co

st t

o b

uy

an

ad

on

pro

gra

m

Cost per impression on targeted voter paid when buying a program

= 1,000 Targeted Voter Impressions = 5,000 Targeted Voter Impressions

Note: some news programs were left off this chart because they were off-the-chart bad

Houston’s NBC, Abc, & Fox News program Inventory – Republican Primary Targeting

Agenda

Prime-time Deep Dive

Broadcast News Program Deep dive

Football deep dive

Status Quo – to buy or not to buy football

Context

• When it comes to football, a Rule-Of-Thumb we hear says that buying football is a no-brainer

- The reasoning is usually rather simple, people live-view (as opposed to DVR) football at greater rates than other programs, and lots of people watch them, so they are worth buying

• In the slides that follow, we examine whether football buys are universally smart

Investigating Claim: buying football is a general good practice, it has a lot of viewers and since most of them are live viewers, it is “worth” it

• On the graph to the left, you see the Houston viewership of two NFL games played at the same time on Saturday Oct 5th , by minute

• First-glance thoughts:

- Viewership in these gamers is not steady, so I am going to be very interested not only in the price I might pay for an ad, but also that ad’s location in the game

Houston NFL Viewers(Oct 5, Mid-afternoon Games)

Halftime Start

Halftime starts

Halftime Return

Halftime Return

Viewers

Investigating Claim: buying football is a general good practice, it has a lot of viewers and since most of them are live viewers, it is “worth” it

• Some observations from these two games:

• It helps to have a good lead-in

- In this case, the Chiefs/49ers game benefited from the Cowboys/Texans game (a local game to this market) that played right before it on same channel

• Half-time penalty

- Games lost 15-30% of their audience after half time, so you better not pay the same rate for a 1st half as and a second half ad

• Games can get a second wind from hour-breaks and other games ending

- This wind will be “caught” if game is close (Chiefs/49ers), and quickly lost if it’s a blowout (Cardinals/Broncos

• All this serves to say, the rule for buying football simply can’t be “it’s all good”

Houston NFL Viewers(Oct 5, Mid-afternoon Games)

Halftime Start

Halftime starts

Halftime Return

Halftime Return

Viewers

Investigating Claim: buying football is a general good practice, it has a lot of viewers and since most of them are live viewers, it is “worth” it

• Beyond tactical observations that suggest not all football buys are good, or at least that some buys are a lot better than others, we need to dig deep to figure out if buying football is, as claimed, “worth it”

• The graph to the left shows the viewership of sub-segments of the electorate by minute in the Cardinals Broncos game. In this case, the sub-segments are types of voters by party and types of elections they turn out for. We could just as easily observe a custom group we have created for a given election, e.g. “likely undecided voters”, or “soft-oppose-ballot-issue-3 voters”

• By looking at the people we care to truly target in a race, we start to understand what we would “get” from a campaigns perspective, for placing an ad at any given minute during the game.

Houston NFL Viewers – Cardinals Vs. BroncosBy sub-segment of electorateViewers

Investigating Claim: buying football is a general good practice, it has a lot of viewers and since most of them are live viewers, it is “worth” it

• Now, let’s assume we are running a traditional Republican vs. Democrat general election campaign.

• A desirable segment to target in this election might be the non-hard partisans. We have blown up their viewership from the previous slide in the graph to the right

• Now that we know how many target voters we are going to be reaching with a given ad, we can calculate efficiency to reach that audience during the game…

Houston NFL Viewers – Cardinals Vs. BroncosBy sub-segment of electorateViewers

Investigating Claim: buying football is a general good practice, it has a lot of viewers and since most of them are live viewers, it is “worth” it

• In fact, in this case, we are going to show what efficiency would be based on different price points offered to us. We do this because we have noticed the cost for buying ads in football games can vary wildly.

• Using this graph, we can now determine what cost-per-impression on targeted voter is going to cost us, based on the price a channel is asking us to pay and when our ad airs

- For example, if the station is asking for $12,500 for the spot, and our spot airs at 4:15, we know we will be paying $1.50 per impression

- Now, we can make rational assessment:

Is $1.50 per that impression worth it, or would we rather have 3 mail pieces to those voters?

Is $1.50 for that impression worth it, or would we rather acquire 15X the impressions against these targeted voters at 10 cents an impression purchased during other shows during the week?

Houston NFL Viewers – Cardinals Vs. BroncosCost per impression based on spot price

Co

st p

er im

pre

ssio

n o

n t

arg

et v

ote

r

Investigating Claim: buying football is a general good practice, it has a lot of viewers and since most of them are live viewers, it is “worth” it

• This graph also allows us to ask some tough questions.

• Earlier in this deck, we demonstrated you could buy this audience on the news for well under 20 cents an impression.

• While it is true that football is live-viewed at rates higher than other programming, the cost does not seem justified.

• Consider the case of an ad coting you $7500, one of many quotes we received for a game airing in Houston. Based on the blue line to the left, let’s say we paid $1 per impression.

• To make football worth it, relative to news efficiency, we’d have to believe that only 1 in 5 viewers of the news is watching live, and thus actually watching our commercials. That just does not seem plausible. Who records the 5:30 AM news to watch that night?

Houston NFL Viewers – Cardinals Vs. BroncosCost per impression based on spot price

Co

st p

er im

pre

ssio

n o

n t

arg

et v

ote

r

Investigating Claim: buying football is a general good practice, it has a lot of viewers and since most of them are live viewers, it is “worth” it

• So to review, football does have a lot of viewers, and it does have the advantage that a very high portion of it’s viewers are live-watching the program.

• That being said, tactical considerations, like time ad airs in game, which game we are talking about, quality of lead in, have a lot to do with a particular ad placement’s worth

• Further, and perhaps more importantly, price still matters, even for football. Football demands a high premium. And even when taken into account the live-viewer advantage, it many cases it is going to be hard-pressed to deliver the value we would need to see.

• The takeaway: “Buy Football” is too general a rule. The nuances surrounding a potential buy must be taken into account to determine if a football ad would be good relative to other buys that could have been made with the money

Houston NFL Viewers – Cardinals Vs. BroncosCost per impression based on spot price

Co

st p

er im

pre

ssio

n o

n t

arg

et v

ote

r

Agenda

Prime-time Deep Dive

Broadcast News Program Deep dive

Football deep dive

Conclusion

Conclusion

• Primetime analysis takeaways:

- Lots of primetime is not “worth it”. We could take the same $4k we might spend on a prime time ad and put it behind lots of other buys that would yield lots more impressions on the voters we are trying to persuade

- Politically desirable sub-segments of the electorate are not that elusive, primetime is not the only way to get to these voters on TV, and in fact we show it can be a very inefficient way to get to these voters. Targeting based on efficiency would allow a given TV budget to achieve a lot more impressions on target voters.

- The number of individuals who only watch primetime is very small, so this fails as a justification for it’s purpose

• Broadcast news analysis takeaways:

- The rule of thumb that says “broadcast news is worth buying” is wrong; some news is worth buying, some is definitively not worth buying

- To separate out which is worth it and which is not, you have to drill down to cost-per-impression-on-targeted-voter for each show.

• Football analysis takeaway:

- “Buy Football” is too general a rule. The nuances surrounding a potential buy must be taken into account to determine if a football ad would be good relative to other buys that could have been made with the money

• Biggest Message: the rules of thumb we typically hear around campaigns as it relates to buying TV were generated in an era before we had the ability to evaluate and value every single possible buy relative to our political audiences; upon deep examination, we have found these rules to be some combination of incorrect, over-generalized, or conditionally valid.