As much as I'd like to think of myself as a numbers guy, my education is so far in the past that I've forgotten 90% of what I think I once knew. Not so with reader Brendan, he's in the thick of it right now and produced a backward looking analysis of what factors seem to determine advertising prices for primetime broadcast shows. While it will appeal to only a fraction of our readers, for them it may make their day.
Here is his summary, but you can read his entire paper here. The text below is all from Brendan:
I love poring over the yearly Advertising Age survey that estimates the cost of a 30 second advertisement for each primetime show. So I did a statistical analysis on what factors influenced these estimated ad prices for the 2009-2010 season.
I examined the following factors for 65 returning shows:
- total viewers (2008-2009 seasonal average in ratings points)
- 18-49 viewers (2008-2009 seasonal average in ratings points)
- age skew (the ratio of 18-49 viewers to total viewers)
- affluent viewers (whether or not the show was among the top ten “upscale series with highest indices” according to a recent MediaPost article)
- day of the week (when the show airs)
- network (where the show airs)
- timeslot (what time the show airs)
- scripted (whether the show is scripted or unscripted)
- DVR viewers (whether or not the show had one of the top twenty 18-49 demo increases due to DVR)
I concluded that the following combination of factors best explain the ad cost for a given show: the amount of 18-49 viewers, how young the show skews, the affluence of the viewers, whether the show is scripted or unscripted, and whether or not the show airs during the 10:00 hour. The other factors (notably DVR viewers) did not seem to have a significant effect.
18-49 viewers
If you visit the site often, it should be obvious that ad cost would be mostly determined by how many 18-49 viewers the show can attract each week. Advertisers seem willing to pay about $50,000 for each demo point. This factor effectively sets a “base price” for ad costs, while other factors may be worth paying a little more for.
Age skew
Advertisers also seem to favor shows that skew young, as an advertisement on a show with a higher percentage of 18-49 viewers will cost more. From the data, it appears that shows with more total viewers tend to skew older, so I believe this just indicates that the most efficient way to reach 18-49 year olds is to advertise on shows that skew young.
Affluent viewers
The application of the data I obtained to evaluate affluence is dubious, but the results correspond with intuition. Essentially, an advertisement on a show with a more upscale audience is pricier than a comparable show with less affluent audience (by about $10,000).
Scripted or unscripted
Likewise, advertising time for a scripted show is pricier than that of a comparable unscripted show (by about $10,000). My best guess is that this is an indication of a higher viewer engagement with scripted shows.
10:00 PM time slot
The data suggests that a show that airs at 10:00 PM will earn more than a comparable show in another time slot (about $5000-$10,000 more). This confused me at first, but I do have a (perhaps tenuous) hypothesis as to why this might be. If advertising buyers predicted low ratings from The Jay Leno Show and thus decreased competition in the time slot five nights a week, they may have expected 10:00 shows on the other networks to benefit from increased ratings. This is a bit of a reach, so I welcome any better explanations.
This study was fairly simple, and not particularly rigorous, so I would not recommend reading too much into it. Having said that, I do think it is safe to base estimates of the value of a show almost entirely on 18-49 ratings and that (at least for this season) each 18-49 ratings point is worth about $50,000 per 30 second ad.
Methods
I used gretl, a statistical software package, to create and evaluate various linear regression models. The final concluded model is a weighted least squares regression that attempts to account for heteroskedasticity given by:
Ad Cost = -68.22 + 47.63(18-49 Viewers) + 0.91(Age Skew) + 10.81(Affluent Viewers) + 12.73(Scripted) + 7.53(10:00 Timeslot)
So here is a question that has bee bugging me for a while. Let’s say you have two shows, one has a 3.2-3.6 demo and consistently between 7 and 8 million viewers. Another show has a 3.2-3.6 demo but consistently has between 10 and 12 million viewers.
(If it helps, think of these shows as “Fringe” and “The Mentalist”)
If you basically have the same demo for each show, but the one show has three million more viewers, shouldn’t more add revenue go to that show? I know that “it is all about the all-important demo” but in this case the demo is roughly the same – so shouldn’t advertisers pay a premium for the show with more viewers. After all they get the same demo penetration and an added bonus of extra viewers.
Or am I missing somthing here?
Glad to see some actual numbers of ad costs (page 16).
(affect, with an a)
zeker,
if my understanding of the analysis above is right, the opposite might actually be true. If the two shows have the same 18-49 demo, it would make sense for the show with less total viewers to skew younger.
It makes sense to me and it doesn’t. I agree that reaching more people in general would be worth more money. However, if a show skews younger and its audience grows, it would be much more likely for the demo to grow faster than an older skewing show. Sorta like a ‘get your foot in the door’ tax for being an advertiser for a show with demo potential.
Zeker, to the degree you trust the estimates from AdAge it doesn’t seem like there is any benefit to having more viewers. And while I don’t think it is probably quite as complex as Brendan’s analysis makes it, I also don’t think it’s a simple as we’d like to make it.
It could be true that a spot in How I Met Your Mother makes more than NCIS, but if so, there are forces at play that aren’t specific even to basic 18-49 demos.
Better w/18-34, in some gender demos and affluence could all potentially play a part in the price differences, but average audience doesn’t seem to.
Unfortunately, even more than last year or the year before, I really didn’t trust the AdAge estimates this year.
You can see them here on AdAge in easy grid format:
http://adage.com/article?article_id=139923
or in the table on page 16 of Brendan’s paper.
Great work. Thanks for posting.
You write: “If advertising buyers predicted low ratings from The Jay Leno Show and thus decreased competition in the time slot five nights a week, they may have expected 10:00 shows on the other networks to benefit from increased ratings. This is a bit of a reach, so I welcome any better explanations.”
I don’t find this to be a reach at all. I think this is precisely what happened. If you re-do your analysis next year I suspect this 10pm impact will wash out in year two of Leno.
Thanks, Mikey!
I wouldn’t be too quick to call the Leno theory fact. I should stress that I wouldn’t consider any of my results especially trustworthy, if only because the AdAge survey results are questionable.
But if the 10:00 effect is real, I also think it would be unique to this year because of some statistical anomaly. Whether or not that anomaly is named “Leno”, I am not as sure, though that was the best thing I could come up with.
Also, to echo, more total viewers shouldn’t hurt a show’s ad price, but it seems to be way down on the list of things that help it (if it has any positive effect at all).
Mikey, I find it a bit of a reach based on AdAge estimates that went into it and coupled with actual results.
Factoring out NBC itself , from M-F you had 10 time slots where it could occur. It happened in four (or 5, if Brendan used the preceding 9:30pm half hour of the estimates for Cougar Town rather than the one hour avg of Modern Family/Cougar Town). So there it happened based on the estimates 5 out of 10 times, but here are the five:
1.) The Forgotten @10p > Scrubs/BOT from 9p-10p. Here, I’d chalk it up to a known quantity in the ratings (which wasn’t good) for the comedies vs. unknown potential for The Forgotten. Leno often betters the Forgotten, so we now know the potential!
2.) The Good Wife > NCIS: LA if the AdAge estimates were any good and the rate differential is still holding up (two very iffy ifs), The Jay Leno hypothesis is in fact the only thing that makes any sense to me. While The Good Wife beats Leno by a wide margin, it is a lower than average performer for CBS.
3.)CSI: NY > Criminal Minds. Was also the case last year.
4. Eastwick > Cougar Town. Eastwick canceled (and routinely BEATEN by Jay Leno).
5. Numb3rs > Medium. This was the case last year as well with the 10pm vs. 9pm slot, though up until this year the Friday night CBS trend was high ratings at 8p, a drop off at 9pm and picking back up at 10pm. Numb3rs is down this year, though so is Ghost Whisperer so I wouldn’t try to blame Numb3rs’ drop on Jay Leno
In the other 5 possible slots, there was no premium for 10pm.
New shows tend not to command much for ads because they’re unknown. If advertisers had bought upfront for the first seasons of Lost and Desperate Housewives, 30 second spots could have been had for a mere 100k.
Also, ad rates (at least for ads sold upfront, which are a huge chunk) are based off of last season’s ratings. So, to see correlation between this fall’s ad rates and the ratings for HIMYM, you have to look at last season’s ratings. And last season, HIMYM and NCIS were doing about the same rating in adults 18-49.
Do sitcoms command more? That would make the difference between HIMYM and NCIS.
Zeker –
There is an added premium above what the demo numbers for shows that generate blowaway numbers of total viewers, like Grey’s, House, AmIdol, and NCIS. There may also be premiums where a show really zones in on a particular demo, like South Park with Males under 35.
To be clear, because I assumed the upfront ad prices were mostly based on *last* season’s ratings, I only used returning shows in the study. So there is no prediction made about ad costs of new shows. Presumably, the AdAge survey is primarily based on “buzz” for new shows.
@Doug
There could be a sitcom factor (I actually should have tried to account for that), but I’m guessing it’s more of an age factor, as HIMYM has a much younger audience than NCIS.
Doug, indeed with regard to HIMYM vs NCIS last year HIMYM had a better 18-49 average. So that was a bad example to use. This year’s results are certainly unimportant to both last springs’ upfront sales and Brendan’s analysis.
edit: Like Brendan, my guess would be any premium for HIMYM vs. NCIS given similar 18-49 ratings is due to more 18-34. The comedy analysis would be interesting but especially if you throw the FOX animations into the mix, I don’t know that you’d be able to separate out the younger skewing when it came to the premiums vs. there being any premiums for comedies.
Scott R., unless you have better information than Ad Age, I’m not sure how true it is that shows get a premium for huge total viewer audiences. If that were true, NCIS and The Mentalist would have had much bigger price tags at the start of this season.
Additionally, House isn’t exactly setting the world on fire in terms of total viewers, particularly not last season when it didn’t get a total viewer boost from Idol like it had in previous seasons. House is a huge success in the demo, but not as impressive in total viewers.
I think it’s more likely that shows that have huge audiences also tend to have huge demo ratings.
I don’t get how some shows can be so expensive, such as Private Practice which costs $175K while House costs $183k and Criminal Minds $120k. House consistently pulls 5.0+ in the demo and CM is the second highest rated show in the night it airs. I do realize that PP has higher ratings in the Women demo, but really, is it worth +50k? Doesn’t make a lot of sense.
Static, those AdAge price numbers definitely have some unexplainable outliers.
The 10 PM effect is likely just an artifact of “skewing younger”. 20-30 somethings stay up later than 40, 50, and 60 somethings.
Unscripted (ie reality shows) skew young, but they also skew dumb. Scripted shows attract a higher IQ audience, which correlates with higher income. Reality does better with the bottom half of the bell curve.
I guess 10pm is better than 8pm.
Too bad FOX doesn’t program 10pm, They could potential make even more money.
In conclusion:
If Scott R is right then my mystery is solved. If everyone else is right I am back as square one. If Robert is right, we are beating our heads against a wall to even figure this out.
Thanks Guys!
So basically NBC is hemorrhaging money?
And wow, someone has actually come up with a formula as to how much ads cost? I always thought it was just a network would accept or decline bids for ads placed in the show of choice…
Here’s the only reason why Jeff Zucker isn’t going to sleep with a horses’ head in his bed anytime soon – and it isn’t prime-time…
…it’s Today, which is significantly in the lead in demos and viewers (yes, viewers may be only for PR purposes, but it’s nice to use any PR you can when your primetime is fourth on many nights)…
…although if Bill Watterston could be persuaded to make a “Calvin and Hobbes” morning show, NBC would become fourth in morning news ratings too
…mmmmmm – chewy on the inside, crunchy on the outside, 100 percent of the RDA of caffeine…
…Chocolate Frosted Sugar Bombs brings you “Calvin and Hobbes: The News”