45 matches were found
- Wednesday, June 13, 2007 #7364
How do I determine reach and frequency of a media mix that includes television, radio, print and online? Also, how do I calculate newspaper reach? Is it the same as coverage? Can I calculate by demo? Thank you
- The Media Guru Answers(Wednesday, June 13, 2007 ):
Once you have the pieces, combine using "random probability"
Coverage (a HH percentage) is the same as rating or one-time reach. Demographic reach may be determined in Newspapers.
See our eTelmar for reach calculation tools.
- Monday, May 28, 2007 #7339
What is the "Sainsbury" formula and is there a difference between unique circulation and paid circulation? Thank you!
- The Media Guru Answers(Monday, May 28, 2007 ):
The Sainsbury formula is a method for combining the reaches of schedules in different media. It varies from the simple random probability method which is based on the generally accepted assumption that there is no particular correlation between exposure to one medium and another. Sainsbury varies by adding a small adjustment to account for an assumed slightly more-than-just-random probability that those exposed to an advertiser's schedule in one medium will also be exposed to its schedule in the next medium. Typically, the adjustment is about a 5% deduction from the result of the random combination.
As you will see at the link shown, we may vary in our arithmetic expression of the probability equation (for the same result), so we can express the Sainsbury formula as (0.95 x random probability).
Unique circulation and paid circulation are unrelated terms.
- Paid circulation refers to the number of copies which are actually bought for money at the newsstand or by paid subscription, rather than distributed free or at such a discounted rate that the circulation auditor no longer qualifies the copies as "paid."
- "Unique" is more commonly an intrnet audience term. Perhaps you are thinking of "unduplicated" audience which only counts readers once, if they read two or more issues.
- Wednesday, April 04, 2007 #7309
Who is the founder of random Media Combination theory? Thank you
- The Media Guru Answers(Wednesday, April 04, 2007 ):
The Guru would not consider this a "theory" per se, nor that there is a "founder."
The "random" concept is basic statistics: when the probablity of occurrence of two unrelated phenomena is known, then the probablity of both occurring is calculated by the product of the two probabilites. In media terms, this means:
Multiplying the two reaches together gives the probability of both reaching the target, which is not reach, but duplication.
Of course, the reach of different media schedules reaching the same target are not truly unrelated phenomena, and various adjustments to pure random have been promulgated by many practitioners.
- Thursday, January 26, 2006 #7083
We are a very small online ad network.....we buy and sell and run various online campaigns. Are there established methods to acquire new business? We hired a sales person, but it seems most of the time is spent trying to get accurate info or research.....while that is inevitable,we are very inefficient in making progress.Any resources or suggestions on the "how to" in the growing of a young company?
- The Media Guru Answers(Saturday, January 28, 2006 ):
.If your network is just a random collection of small sites and you simply sell by low cpm, then any means of advertsing targetting inline media planners will be useful. AMIC itself is one good option, as are MediaWeek, online specialty trade print and online-topic ad trade email newsletters.
If your network is made up of a specifc kind of sites, like hoppyists sites or professional business sites, then find other trae media in those categories.
Around here, we believe advertising works, - as presumably, do you.
- Tuesday, September 13, 2005 #7010
I have a client whose target audience consists only of Insurance Brokers. I am planning a print campaign and need to justify it to the owner who wants to know only the ROI. I have one weekly business publication and two trade pubs in three markets. I also only have the circulation numbers and nothing else. What is the best way to calculate reach and frequency with this information? Can I also figure the percentage of awareness from these numbers?
- The Media Guru Answers(Saturday, September 17, 2005 ):
In trade publications, circulation tends to be heavy among qualified readers, pass-along is often minimal. A conservative audience estimate is audience = circulation. Again, because of this distibution pattern, issue-to-issue cume is minimal. As far as duplication between titles, random probability
is a safe estimate, but may be a bit high.
Reach becomes a maximum measure of awareness; you need to estimate the required frequency which generates awareness as well. Ad Awareness can't exceed reach.
- Friday, August 12, 2005 #6993
Back when I started in media (1984) I learned a way to hand-calculate combined reach/frequency. Say adding together a tv and radio schedule. I unlearned it over the years, especially when I had software that did it for me.
Now I don't have the software anymore and have been asked to do a combined R/F. Do you have that formula?
- The Media Guru Answers(Friday, August 12, 2005 ):
Click here to see past Guru responses about combining reaches
- Wednesday, May 25, 2005 #6937
Thank you for your answer about internet grps(#6936). Another question would be, can we calculate reach for internet? and how about reach for a hispanic target
- The Media Guru Answers(Wednesday, May 25, 2005 ):
Calculating reach is ususally done in one of two ways:
Using a respondent level audience measurement such as Nielsen//Netratings
or comScore-MediaMetrix, one tracks the actual use by the relevant demographic within the sample against the schedule run.
More practically, one obtains a series of such measurements and builds a model, so that one can then genralize from schedules run in the future, using variable such as # of impressions, number of sites in the mix, share of page loads on the sites, etc.
The issues are what portion of the sites' reach does your schedule get and what is the duplication between sites' audiences.
For example, Yahoo might reach 40% of all those online in a month, but your buy will probably appear in less than 0.1% of all Yahoo page loads. And how many of the persons exposed to your Yahoo buy will also be exposed to your buy on MiGente.com?
Since Hispanic audience is measure by both services, the Hispanic issues are no more difficult in this scenario.
As a ballpark sort of estimate, most major sites ought to be able to tell you the number of unique visitors exposed to your schedule. This number, divided by the relevant universe will give you an estimate of reach on that site. You can combine sites' reaches by random probability unless you can get site duplication estimates from the sites.
- Tuesday, April 19, 2005 #6903
Dear Media Guru,
Please help me figure out the combined reach and frequency of a multi-station TV buy. To find the average frequency, would I add the frequencies of the stations and then divide by the number of stations? Do I then multiply it by the total number of GRPs to calculate the reach? Since I do not have access to a software program, I need to calculate this manually. I'm in desparate need of your help since these figures are due soon.
- The Media Guru Answers(Sunday, April 24, 2005 ):
Frequencies are never additive. A specific model is the only accurate way to combine station reaches. The reach may be divided into the sum of GRP to calculate the average frequency.
A very rough estimation of combined reach might be calculated by a string of random probabilty parings (i.e. pair two stations and then combine the next with those and combine the next with the cume of the three and so on). Because this is "random" and does not account for the greater likelihood of any TV viewers to view other tv, the result will be overstated, by at least 10%.
- Wednesday, February 09, 2005 #6787
What is the formula for combining two R&Fs?
- The Media Guru Answers(Wednesday, February 09, 2005 ):
Click here to see past Guru responses about "random probability"
- Thursday, May 20, 2004 #6500
Using the OTS formula (GRP/Net Reach), if we set an OTS target with a predetermined reach, can we arrive at the required GRP for differrent OTS targets.
Why effective frequency is more popular over OTS when setting frequency objective. In my experience we need to achieve more GRP's to achieve a predetermined reach for an effective frequency over OTS target, any reason for that methamatical relationship.
- The Media Guru Answers(Thursday, May 20, 2004 ):
As a matter of simple arithmetic, Reach and GRP are inextricably linked by a multiplying factor which can just as readily be effective frequency. This does not mean that you can set any reach goal at random and assume a given GRP number will relate back with a specifc OTS. Different mixes of dayparts and media elements have different capabilities in reach / effective frequency generation.
Why more GRP for an effective reach level? Again, simple arithmetic explains it. "Reach" in an ordinary "reach and frequency" calculation, means reach 1 or more times. In other words, a frequency of 1 is treated as "effective." Typically, when we talk about "effective reach" we are working on an assumption that 3 or more frequency is needed for effective communications so that only those reached at least 3 times count. Naturally, more GRP are needed to get a given reach 3 tiems than only once.
- Tuesday, May 18, 2004 #6497
What is the difference between qualatitive and quanatative research?
- The Media Guru Answers(Thursday, May 20, 2004 ):
The great difference is "projectability."
- Quantitative research uses random sampling and may be analyzed statisctically to evaluate its degree of error in terms of predicitng behavior of a group like the one sampled. Surveys are the typical quantitative tool, and we learn what percent of a population group eats cornflakes or Kellogg's cornflakes. Quantitative results may be discussed in terms of "reliability," or how often the same research with the given sample would produce the same result within a given range of tolerance. This is different than "validity" which is about Whether what was supposed to be measured was actually measured.
- Qualitative research is aimed at understanding how people feel about things or react to ideas. It is the pursuit of insight rather than measurement. Focus groups are the most typical type of qualitative research. In this groups of 8 or 10 people are encouraged to discuss how they shop for a product how they feel about a package or a commercial, how they learn about products. The insights gained may influence copy or the design of a subsequent survey. Quantitative research may not be projected to behavior of populations. Anytime yoiu hear someone say "in the focus groups, X% said ___", beware!
- Tuesday, December 16, 2003 #6315
How do you calculate combined frequency. If I have a cable plan in a market with a frequency of 2.6 and a broadcast tv plan with a frequency of 6.6 - what is the combined frequency?
- The Media Guru Answers(Saturday, December 20, 2003 ):
The Guru will assume you are referring to average frequency, typically considered for a four week period. One actually calculates the combined Reach and GRPs and then figures the "combined" frequency. Consider the following table. If you had run 400 GRP in broadcast and had 61 reach there would be 6.6 average frequency. If you also had 100 GRP of cable and a reach of 38, there would be an average frequency of 2.6.
GRPs are simply additive for a total of 500. Reaches must be combined by a system that recognizes duplication; "random probability" will overstate a bit when you are working with two related elements such as different kinds of TV. Probability might have estimated a combined reach of 76 here but let's suppose your algorithm estimates 72.
In any case, the combined average frequency is calculated thus: divide the combined GRP (500) by the combined Reach (72) which equals 6.9; see below:
- Tuesday, August 19, 2003 #6130
I would like to know if there is any possibility to measure additional reach that is generated by adding other medium to usual media mix based on TV. For a couple of years we have been running almost 100% TV based campaigns for FMCG client and we have recently observed that there are no changes in building brands awareness nor other crucial barnds parameters. We are now deeply thinking about moving some part of the budghet to other mediums as we intuitively think we can be gaining some incremental value by this shift. Is there any way to measure what additional percentage we can gain by using TV + radio or print instead of using only TV? Is there any research on what incremental campaign reach or brand awareness can we get by this decision? We have difference of mediums research methodology (telemetric for TV vs declarative for radio and print) so perhaps you could indicate some findings or research. Thank you a lot
- The Media Guru Answers(Saturday, August 23, 2003 ):
The Guru infers from your query that you are in a country without media measurement or at least without reach models.
Reach models we have are consistent in demonstrating that adding weight to a base plan in a new medium increases total reach more than adding comparable weight in the base medium already in use. Reach models within single media generally are based on measured "curves" of growth. However the reach added by a new medium can typically be estimated by simple random probablilty calculations.
Click here to see Guru explanation of calculating reach by probability.
- Thursday, May 15, 2003 #5974
I can use random probability to calculate reach. Is there any way I can create a complete frquency distribution? For instance, Is it correct to add up probabilities for an individual for 4 publications (Say 0.75, 0.75, 0.5 and 0.25) and say that frequency for the person is 2.25?
- The Media Guru Answers(Sunday, May 18, 2003 ):
The Guru follows neither your math nor your logic here.
If your example means the the indivisual has a 75% chance of reading the first publication, etc how would that give the person a 2.25 frequency for the 4? You are working with almost unrelated data, not to mention the overstatement of random probability in calculating reach of related media. Further, frequency distribution deals with the numbers of persons who experienced each integral frequency, i.e. how many had one exposure, how many had two exposures. No individual may have a fractional exosure.
- Tuesday, April 22, 2003 #5943
i believe there is a rule of thumb when calculating the reach of trade publications. something like the first major pub in the buy gets over like 75 or 80% and then there is a average increase per added pub. It's just a rule of thumb, but it sure would be useful since we cannot define the size of the overall industry's target universe. IF you do not know this rule of thumb, how would you suggest we calculate the reach and frequency of 5 trade pubs bought with differing levels of insertions over a year. Thanks for any help!
- The Media Guru Answers(Saturday, April 26, 2003 ):
Crudely: Calculate each book's circulation's percentage of the sum of the goups' circs and make this each one's individual reach. Start with the largest and calculate the added reach contribution of the others by random probability.
- Friday, February 07, 2003 #5815
We've been asked to estimate reach/frequency/etc. for a plan that includes USA Today, newspapers in 8-10 major markets, spot radio in 5 markets, metro traffic in 8-10 markets, and national magazines. I think this is impossible, but can you think of any way I can provide the client with a decent estimate? I was thinking I could start by pulling delivery for USAT, magazines, New York Times, and then somehow estimating the rest.
- The Media Guru Answers(Saturday, February 08, 2003 ):
The Guru sees no problem, and so does not quit understand your question perhaps. Assuming you know what reach and frequency is, you can readily determine the reach of each one of the media you mention. Most simply, you can combine them by random Probability . Most reach and frequency systems on the market, like our own eTelmar, can do this for you. The only "trick" is accounting for the different geographies, but that's just artithmetic, and easy if you look at all the percentage reaches as their equivalent in thousands.
- Monday, August 05, 2002 #5446
As a new publication what is the best way to combat MRI. Currently our website has skyrocketed on alexa versus our competeting books! What is the best approach. In my eyes MRI is old news and alexa is the wave! Would love your opinion.
- The Media Guru Answers(Wednesday, August 07, 2002 ):
Alexa neither measures nor purports to measure print media. Alexa is a web page usage report based on 7 million users who have downloaded Alexa software. These are neither a random sample nor representative of web users.
MRI is a measurement of magazine audience, based on a large scale sample of U.S. adults and statistically projectable to the population and various demographic cells.
Web traffic to the web sites of magazines is probably not indicative of the magazine's audience among its target, anyway, and if it were there are better measures of web audience, such as comScore / MediaMetrix.
Alexa is not the wave, in the Guru's opinion.
One does not "combat" MRI. When a magazine has not yet been measured, it may talk of circulaiton, subscribers or single copy sales.
- Wednesday, May 22, 2002 #5301
In one of your responses to advantages of media mix and multimeedia strategies u have mentioned "Better distribution of frequency of exposure" as the advantage of using a media mix
Can u pls elaborate on this
Thanks for the help
- The Media Guru Answers(Thursday, May 23, 2002 ):
Each medium has heavier and lighter users. The heavier and lighter viewers of each medium duplicate at random, so that heavier print readers may be the lighter TV viewers. Consider the graph below, comparing a TV + print plan (1) to an all-TV plan (2). At the same budget, Plan 1 had a reach / frequency of 89.5 / 6.7 while plan 2 achieved 78.6 / 5.7.
Not only does plan 1 have better total reach and average frequency, but the portion of the target exposed to each number of ads (in the bar graph) is greater for plan 1. The proportional margin increases as number of exposures grows.
- Tuesday, May 21, 2002 #5295
Avoiding media jargon: provided client have 2 spot (each 30") and he would
like to hear recommendation of rotation. Particularly he asked of any
proven examples of tactics when one spot is placed 1st in commercial block
and the second spot is placed last in block- he called it "top-&-tail"
(any research saying it is more effective than random placing in block).
The other side of the story is - I suggested to rotate as follows: 1 spot
- any other commercial- 2nd spot- other commercials (I heard it is called
"consecutive spots". Both spots are in form of reportage- this connects
them and therefore I recommended such rotation. I read in couple of books
that qualitatively placing 1st /last in break is more effective, as peaple
switch to other stations after 1st spot and than come back likely at the
end of the block, but it was not proved by any examples, case studies,
which is requested by my client.
- The Media Guru Answers(Wednesday, May 22, 2002 ):
There is research showing first and last in break are more likely to be viewed. The alternate 'consecutive' might be more effective in awareness / recall building. among those who view the entire set.
So the first issue is to set a standard of "success."
Is it most viewers or most recall? Or is it really sales in resonse to audience size versus audience impact?
Best research resources are ESOMAR, the European Survey, Opinion and Market Research Organization and The Advertising Research Foundation InfoCenter. For details about the InfoCenter, call 212-751-5656, extension 230.
- Thursday, May 16, 2002 #5288
Hi, do You know of any publishing companies/print
media which are using variable advertising pricing
according to the reach of the media so that fixed CPT
is offered instead of fixed rates? Are there any print
media where You can buy GRP-s?
If Yes; how is it done?
- The Media Guru Answers(Saturday, May 18, 2002 ):
The physical form of print media don't really allow this. You may buy geographic pieces of circulation, demographic editions or A/B (every other copy) splits of the circulation. You seem to want a random placement akin to online.
You can evaluate print media by GRPs but not buy audience chunks this way.
- Monday, February 04, 2002 #5050
Is Telmar's multi-basing system the same thing as Fusion? And, if I'm currently doing the random probability formula to get total reach percent, what is the difference between that and Telmars calculations? Thanks.
- The Media Guru Answers(Wednesday, February 06, 2002 ):
According to Telmar:
Multibasing preserves the integrity of a survey. It does not ascribe
answers, and as such, avoids what we call "regression to the mean",
washing away everything to averages. It preserves the leverage of a media
element against any target group, not just those that leverage on
Telmar's R&F formulas use the actual turnover and duplication
between media that are inherent in the survey. When there is real
data, we use it.
- Tuesday, November 20, 2001 #4900
I am trying to estimate past Reach & Frequency for a
transportation trade industry print campaign -- and based on that set R&F goals for 2002. I have gathered the following information: Target universe in US, Asia and Europe; each publication's circulation to that target (where available); duplication (very limited availability of this from these pubs). Given this information, what formula could I use to (gu)estimate Reach & Frequency for this Trade plan? Alternatively, what other measures could I offer to my client to measure a recommended media plans effectiveness (i.e. Competitive SOV)?
- The Media Guru Answers(Wednesday, November 21, 2001 ):
The simple formula begins by calculating audience-divided-by-universe to estimate ratings (probability of exposure). Multiplying together all the negative probabilities gives you the reach, disregarding specific duplication. In other words, if you get a rating of 14% of target, the negative probability is 86%. Then, two issues of that publication have a combined negative probability of 0.86 X 0.86 or 0.7396. Thus the probable "reach" is 1 - 0.7386 or 26%. This reflects a rando likelihood of dulication of roughly 14%. In reality, there is more than just this random duplication between two issues of the same trade title, probably 50%+, so a better estimate of the reach would be 14% + 50% of 14%, or 21% reach.
For a good guestimate, combine all your insertions this way, using 60% duplication between repeats in the same title and 30% between different titles. Use judgement about titles from different countries which may have virtually no mutual duplication.
SOV is another comparitive tool. Going beyond relative communication and relative spending gets quite speculative.
- Saturday, October 13, 2001 #4791
I have been trying to find a clear, easy-to-understand way of explaining to account services why Nielsen isn't an absolute, exact science, but see it for the base measurement tool that it is. Any way you could sum it up for me? I have pointed out many of the flaws and error measurements to the methodology but they can't seem to grasp why it isn't black and white.
- The Media Guru Answers(Saturday, October 13, 2001 ):
One would devoutly hope that an account professional would have some experience with survey research, like attitude and usage studies, but then again, many of them believe you can project porduct usage from a couple of focus groups of ten people.
Nielsen is like any other research based on sampling, it is subject to sampling error. Projecting behavior of an entire population based on a large sample is a well established scientific procedure, and Nilesen has a good overview on their site. The short answer is, when well conducted research is based on a random sample, results are at best within a certain +/- range. When you hear that a national political poll regarding two candidates is '+/- 3 points' that really means that a 50% projection is between 47 and 53% at a certain confidence level like 95%. Or, that id the poll was redone the same way 100 times, 95 times the results for that question would be from 47 to 50%. But 50% has the largest possible sample portion behind it ( if the answer was 75%, it also means that 25% said the opposite and both must be equally reliable.
TV ratings are more often numbers like 2% or 7% because there are so many viewing choices and non-viewer as well. With the same sample as the political poll above, the 2% answer (or rating) is subject to much less absolute error than the 50% answer, perhaps 0.4 points rather than 3 points, But as a relative error that is very big, 20% of the audeince reported. Over time, if a show' real rating is 2.0 and rthe aduience doesn't change, there will be some weeks that get a 2.4 rating and others with a 1.6. They'll average out to the 2.0, but depending on how you read Nielsen, you need to account for that. If you by a weekly program expecting the 2.0, and happen to be in a 1.6 week, you hope (realistically) that some other program you bought has an 'up' week. Over a whole quarterly schedule, with a lot of announcements, if the GRPs are far from expectation, look elsewhere than Nielsen for blame.
- Thursday, July 26, 2001 #4610
What is your opinion of Mobil Trak radio research and the methodology they employ to generate radio statistics.
- The Media Guru Answers(Thursday, July 26, 2001 ):
1. Their method can be valuable in certain, specific, retail cases. If a store wants to know what station is on the radio of persons passing their location, MobilTrak can be useful.
2. The limitations are severe, however. MobilTrak's claim that the majority of all radio listening is in-car is strongly contradicted by Arbitron data, which shows only 30% of total listening is in car. Arbitron, unlike MobilTrak, has no vested interest in location-of-listening results.
In multi-store mall situations, a store can't even be sure that passing cars are their own customers.
For market-wide measures, the MobilTrak technique is not only limited by the car-only problem, but potentially by lack of randomness even among car listeners.
- Friday, June 29, 2001 #4538
I have two questions about calculating reach and frequency that I have been unable to find in the archives of past responses. Perhaps you can help?
1. I normally use the formula (a+b)-(.a*b) to determine combined reach of two mediums, such as radio and print. How do I calculate the combined reach of more than two? The plan I am working on includes spot TV, spot radio and local newspaper.
2. Is it possible to determine a combined reach for more than one market or should each market be reported separately? In the past, I have provided separate delivery for each market in the same plan with a total number of gross impressions for the whole plan. Is this correct?
Thanks in advance!
- The Media Guru Answers(Friday, June 29, 2001 ):
1. This common formula is based on an assumption that different media duplicate their audiences according to random probability. Therefore if you follow this assumption, media may be added to combinations of media in a "chain" of the same formula. So, once you have combined TV and Radio, you can use this combination as your "a" and then combine it with newspaper as "b."
2. You can combine reaches across markets by doing a weighted average. Multiply the reach in each market by the percent of U.S. in each market. Add all the products and divide by the sum of the % U.S.
- Wednesday, June 06, 2001 #4458
I'm working on a plan that includes cable and network television. I have been asked to present a rational for different schedules on three levels of spending. If i know the programs rating point, the average CPP and the cost per spot, how can I use this information to put together the total reach/frequency of sample schedules. I'm trying to get general information at this point without contacting reps to run several schedules. I need to know how to do the math by hand without a program if it's possible. Thanks
- The Media Guru Answers(Wednesday, June 06, 2001 ):
It's no longer really reasonable to do the math by hand. The Guru has described calculating reach by "random probability" in the past. But the unique duplication patterns within tv schedules need to be accounted for either with tables reflecting many schedules' reaches or computer models.
Our own eTelmar offers low cost, single use, online reach calcuation.
You might try the R&F generator at U. Texas .
- Friday, May 04, 2001 #4368
Media Guru, please help. How do I calculate reach and frequency for a two-week, two-newspaper buy? We are placing 4 ads per week (total of 8 ads for the schedule) on Newspaper #1, which has a maximum reach of 9% of our target. Newspaper #2 will carry 2 ads per week (4 ads for the schedule) with a maximum reach of 23% of our target. Please advise. Thanks!
- The Media Guru Answers(Sunday, May 06, 2001 ):
Find some example newspaper R&Fs at The Newspaper Advertsing Associations Marketscope site.
In very general terms, you can estimate some parameters. If newspaper A has a 9% maximum reach, it probably has a single copy reach of around 7%.
If B has a maximum of 23%, then it likely has single copy reach around 20%. So the outside bounds of reach for your schedule are a minimum of 20, but more likely closer to 25, the random combination of the two papers' single copies. The outside maximum is 32 ( the 9% plus the 23% maxima), but more likely closer to 30 (random again).
A solid estimate of 25-30 reach for your schedule should be good enough, but you could use the eTelmar pay-per-use system for a specific calculation.
Frequency, of course, will be the sum of the single copy audeinces of all insertions (GRP) ÷ the reach estimate.
- Wednesday, December 13, 2000 #4041
My question is regarding print measurement. For a consumer print campaign (magazines, regional) I've been asked to provide a pithy statement (to be read by a board of directors with limited marketing savvy) adressing the effectiveness of the proposed print campaign. Our account planner asked for reach and frequency, which I don't believe I can provide. I can provide circulation and readership (which would equate to reach, I believe, but that doesn't account for duplication). I am to complete the sentence "This plan results in..." Am I missing something? Thanks!
- The Media Guru Answers(Sunday, December 17, 2000 ):
You have not made clear why you believe you cannot provide reach and frequency. Once you have the readership of individual publications you can begin to combine their audiences in a rough way, by "random probability." This method will understate duplication somewhat, because related publications and particularly multiple issues of the same publication duplicate more than merely randomly. Using duplication between simialr national magazines, as documented by services like MRI, you can reasonable estimate the duplication in your own schedule and thereby estimate your reach and frequency.
- Wednesday, November 15, 2000 #3972
I'm a newcomer to the site and I very much enjoy your bright responses.
Re recency, you write >a core concept of recency is that once the third exposure is delivered, all additonal exposures are at 3+.<
That concept belongs to Herb Krugman, ("Why Three Exposures May Be Enough.")whose work was misread as supporting effective frequency.
The corresponding core concept of recency is a single exposure within a short planning interval is most cost-effective.
These results in moderate TRP's and more weeks of advertising. When heavier weight is called for (i.e., new product introductions), instead of accepting random frequency, recency shortens the planning interval and maintains a solus reach goal. Planning for continuous reach produces a better distribution of frequency.
My apology for this somewhat truncated explanation. I can provide greater detail if you'd like.
- The Media Guru Answers(Thursday, November 16, 2000 ):
As the leading industry writer on the topic, your comments are greatly appreciated, and you'll have to excuse the Guru for using your own writings in his reply.
Maybe "seminal" concept would be a better term than "core" concept when the Guru cites this Krugman principal, since it is more part of the evolution than structure of recency.
Perhaps connecting the concepts himself, but gathering them from your own articles, such as Learned Any Ads Lately?, the Guru sees the concept that all additional exposure are at 3+, as part of the underpinnings of Recency. Because this idea gets us past the effective frequency issue, the -- superior, in the Guru's opinion -- Recency theory surmounts objections from the effective frequency camp.
- Wednesday, August 09, 2000 #3691
Dear Guru, what does the terms Bi-weekly, Bi-monthly actually mean. My assumption is that bi-weekly means every other week and bi-monthly is every other month. My collegue assumes bi-weekly is twice a week and bi-monthly is 2X per month. Can you clear this up for us. Thank you.
- The Media Guru Answers(Wednesday, August 09, 2000 ):
This is a semantic question, more than a media question, but interesting nonetheless. The Guru consulted his random House "Webster's Encyclopedic Unabridged Dictionary."
The first meaning of bi-monthly is "occurring every two months." But the second meaning is "occurring twice monthly." Bi-weekly has the same pairing. In both cases, the second meanings are given an alternate definition of semi-monthly / semi-weekly. And, the "semi's" only mean twice per time period.
So, on balance, the preferred usage of "bi" would be once every two time cycles. But a publisher could mean either. A wise publisher will avoid the terms and say "twice a week" or "every other month" so that there is no confusion.
- Tuesday, April 25, 2000 #3420
We are putting together a sponsorship package that incorporates TV spots, our company newsletter, our website and our fleet vehicles -- is it possible to estimate a combined reach/frequency for all four mediums combined?
- The Media Guru Answers(Tuesday, April 25, 2000 ):
The TV is easy, using standard methods, of which you are probably aware.
The other estimates must start from simple counts of the newsletter circulation, web traffic and - the toughie - persons exposed to your fleet. Most simply, after getting a standard TV reach, convert the other media
impressions to ratings and combine by "random probability."
- Thursday, March 16, 2000 #3326
I would like to know if there is any equation to calculate media mix reach?
- The Media Guru Answers(Thursday, March 16, 2000 ):
There are several, equivalent ways to express the arithmetic to combines media according to random probability, which has been found generally adequate for the purpose of multimedia combination.
Here's an easy one:
- Work with two reaches at a time
- Treat the reach of each medium as a decimal (50 reach is 0.5)
- Add reach of medium A and medium B
- Multiply reach of medium A by Reach of medium B
- Subtract the product of the multiplication from the sum of the addition
- Reach of medium A = 40, reach of medium B = 55
- 0.4 + 0.55 = 0.95
- 0.40 x 0.55 = 0.22
- 0.95 - 0.22 = 0.73
- Combined reach is 73
To add additional media, treat the combination as medium A and the next medium as B.
In some cases, a planner may have access to research which shows that an adjustment should be made for actual, measured, duplication between different media, rather than use the "random probability" formula above. In that case, more sophisticated reach calculating software packages, such as those from
Telmar allow you to make the calculation and build in known adjustments.
- Friday, November 12, 1999 #2964
Can research determine which media is best to drive traffic to a local retail business? If there is a particular medium or media, what research approach can best determine which media works? Please consider that this a local business that currently advertises in radio, newspaper and billboards and is very successful at driving traffic to retai outlets that are not highly visible in their marketplace.
- The Media Guru Answers(Tuesday, November 16, 1999 ):
A specific individual business can use research to determine this.
Most simply, it is easy to include something in the advertising which makes people want to tell the business's operators where they heard of them. Or staff can be instructed to inquire where customers heard. More expensively, a commissioned study can probe awareness and shopping behavior from a random sample or a customer database sample. In any of these cases, the research must be carefullly studied and interpreted, to distinguish the results of branding efforts from promotions.
If the business has a long history of establishing its name and offerings in the community through radio and outdoor campaigns, the research might still find that "What brought the cusotmer in today" was a newspaper or Yellow Pages ad. Analysis might well show that the newspaper or Yellow Pages ad would have had little impact without the other media's branding effects.
Different businesses enjoy different effects from various media. A roadside, impulse business, like a highway restaurant chain can get immediate results from highway billboards which would have much less benefit for an in-town, white tablecloth eatery. A branding-oriented newspaper campaign for the latter would likely be more effective than one for the highway chain.
- Tuesday, September 14, 1999 #2792
What can you tell me about reach-based planning?
Thank you in advance.
- The Media Guru Answers(Tuesday, September 14, 1999 ):
The usual assumption is that print and broadcast duplicate with random probability, there is no special, greater or lesser likelihood that persons in the audience of the print schedule will also be or not be in the audience of the broadcast schedule.
Mechanically. the combination may be calculated in a few equivalent ways. The Guru finds it easiest to consider the reaches as decimals (50% reach = 0.50).
Subtract the reach of print from 1 and multiply this by 1minus the reach of broadcast. Suppose print has a 40% reach and broadcast has 55%.
By subtracting 0.4 from 1 (1 - 0.4 = 0.6), you have the probabilty of the target not being exposed to print. Subtract 0.55 from 1 to get the probability of not being exposed to broadcast (1 - 0.55 = 0.45)
Multiply these two together (0.6 * 0.45 = 0.27) and you have determined there is a 27% probability of people not being exposed to either of the combined media, or a 73% reach.
This formula is typically used in media software to combine different media.
Certainly there are cases where there is a somewhat better than random probabilty of media duplication, such as TV Guide combining with a TV schedule, but that's the exception, calling for judgement.
- Wednesday, September 01, 1999 #2759
Is the random probability formula used to combine reach for different media also valid when looking at effective reach (i.e. 4+ level)?
- The Media Guru Answers(Thursday, September 02, 1999 ):
If you mean, can you combine the 4+ reach of one medium with the 4+ reach of another medium to get the 4+ reach of the two combined media, the answer is no.
Among those who were reached 2 or 3 times by each medium, some will now be reached 4 or more times and some will not, yet these people are not considered by combining only the two four+ groups. There are also those reached only once by the first medium and three times by the other, etc. A new, overall calculation of the frequency distribution must be done, to determine the 4+ of the combination.
- Thursday, August 19, 1999 #2727
The formula for calculating the reach of media vehicles is (a+b)-a*b.
Please tell me the "N" formula for it,
or you have a different formula for calculating reach?
- The Media Guru Answers(Saturday, August 21, 1999 ):
Your formula is for " random probabilty," which is used to combine two different media, based on the assumption that their audience duplication is purely at random. This formula is not appropriate to combining different vehicles in the same medium, which typically have more than merely random duplication.
There are various, quite complex formulae for computing reach of various vehicles of the same medium, among them the Beta Binomial, Lamda function, and others. The Guru is not familiar with your reference to "the 'N' formula."
- Friday, April 30, 1999 #2481
Is there any way to calculate duplication across a media plan using several media (e.g. print and radio and TV), or can I only get a duplication analysis within a media (radio duplicaton and then another duplication factor for print, etc , etc)
I use telmar for research with simmons and arbitron access and we also use JDS for buys.
- The Media Guru Answers(Friday, April 30, 1999 ):
The standard assumption in media planning is that duplication between different media is purely at random. Therefore, the random probability formula is used:
- Express the reach of each medium as a decimal (50% reach = 0.5)
- Multiply the reach of one medium by another to determine the duplication.
- Subtract the duplication from the sum of the two reaches to get the net reach
So, if you have a 40% reach in TV and a 55% reach in Print, multiply
0.4 x 0.55 to get 0.22
subtract 0.22 from 0.4+.55 and get 0.73 or
73% reach of the combined media.
There are a variety of ways to do the calculation. The Guru actually prefers to use the probablilty of not seeing each medium (reach as a decimal subtracted from 1.0) When these are multiplied they give the net probability of not seeing any of the media. When this result is subtracted from 1, the final result is net reach. This style is particulary useful for combining several media at once.The example would combine this way:
- 1-0.4 = 0.6
- 1-0.55 = 0.45
- 0.6 x 0.45 = 0.27
- 1-0.27 = 0.73 or
Telmar's "Media Mix" program uses these assumptions.
- Friday, March 26, 1999 #2415
The target audience for my schedule is "Own a dog or cat and took your pet to the vet at least once in the last 12 months". I have done the Xtab in Telmar and know which states this audience lives. I am trying to get more specific and find out what DMA the target lives in.
What is the best way to find this information? I do have information from the Lifestyle Analyst book provided by SRDS but I am not that comfortable with it since I was told the data is compiled from warranties people send in. Please help.
- The Media Guru Answers(Monday, March 29, 1999 ):
Why are you troubled by data from warranty cards? Of course, this data will not be from as random a sample as Simmons, MRI or The Mendelsohn Media Research Affluent Study. However, its skews are probably based on income more than anything else and data should be fairly useful in terms of finding the geographic dispersion of behavior like vet visits.
Another approach is this: the survey data listed above does report separately on several major DMA's. Beyond that, the mapping/modeling software offered by suppliers like CLARITAS can extrapolate the survey data into regional or market by market data. Of course, this may be as removed from validity as the warranty data.
Why not contact the associations like American Animal Hospital
Association (AAHA) and American Veterinary Medical
Association (AVMA) to inquire about visits per market?
- Tuesday, March 23, 1999 #2403
I have been researching these questions for a number
of days now and have been unsatisfied with the
answers I have been receiving. I am a new member
and new to this field, any direction would be most
helpful. Thank you in advance...
1) What is the difference between Rate Base (a number
guaranteed by publishers and audited by ABC) and
Readership (a number provided by, say, MRI) levels
for magazine publications?
2) Which number (above) is most often used to
calculate CPM (I believe this calculation is
3) Is 'readership' really a composite number (perhaps
a result of some other formula)? If so, does
Page Exposure Rates factor into 'readership'?
- The Media Guru Answers(Tuesday, March 23, 1999 ):
If you went to AMIC's Rates, Dates and Data area and clicked the link
"Audience data from MRI is available for
Fall 1998 for Total Audience, Circulation and Readers Per Copy " you would see the table from which this image is taken:
The following discussion will use this table as a visual aid.
"Rate base" refers to circulation, the actual number of copies of a publication printed and sold for the average issue over a specified period of time. In the table, "Circulation" is the middle column of data.
"Readership" is the number of readers of the average issue. It includes "passalong" readers, who may not be the buyers / subscribers but read some else's copy. In almost every case, total readership will be greater than circulation. The first three columns of the MRI table we are looking at are readership numbers.
CPM can be calculated based on either circulation or readership. The circulation CPM (Cost Per Thousand) calculation is: divide ad cost by the number of copies in circulation.
The readership cpm calculation is: Divide ad cost by number of readers of an average issue. Often readers within a specified demographic the advertiser is targeting are the divisor in this second calculation. As a planning tool, the readership CPM is more common than the circulation CPM, especially for categories of print that use readership research, such as MRI.
Many people misinterpret the common reporting of "readers per copy." The last three columns of the MRI data are readers per copy figures. What audience research actually measures is readership. A random sample of consumers is interviewed and asked about their magazine reading to determine how many readers there are for an average issue of a magazine. Readers per copy is a calculation done after the fact, dividing the readers measured by the circulation. It is a handy factor used to compare magazine pass-alongs or to calculate other audience elements.
- Wednesday, December 02, 1998 #2192
Dear Guru. It is not still clear to me how to
measure or calculate Reach of the ad campaign
using media mix. For example, my ads on TV provided
90% reach, and ads in print reached 25% of the
target audience. What is the total reach, frequency
of the campaign? What other indexes can we find for
And my second question is about outdoor advertising.
It is essential to measure the effectiveness of the
ad campaign comparing awereness and sales before and
after the ads placing. But that is somehow the post-
campaign analisys and my client would like to see some
feagures before the campaign starts (pre-campaign).
What indexes (like reach, frequency, GRPs, OTS) can
we provide to the discription of the outdoor
ad. campaign? Thank You very much.
- The Media Guru Answers(Wednesday, December 02, 1998 ):
Reach of a medium in a plan is simply a statistical probability. Further, it is generally thought that each medium overlaps each other medium randomly.
So, in your example, if you consider the reach of each medium as a decimal, the probability of not being exposed to TV is 0.10 and of not being exposed to print is 0.75.
The probability of not being exposed to either one, is therefore 0.10 times 0.75 = 0.075.
Therefore, total reach of the mix is 92.5 (if 0.075 or 7.5% don't see it then 92.5% do see it).
Other basic "counts" for a campaign are impressions (OTS), cost per rating point and cost per thousand impressions.
All of these counts; reach, frequency, GRP, OTS, etc are possible for outdoor, if the research has been done, in your country, to count the audience of the locations used.
- Friday, July 17, 1998 #1957
I am curious to find out how user-centric Web research firms
maintain "freshness" of their samples. How often and how many
samples are being replaced on what intervals by companies
like Media Metrix, Nielsen, Net Ratings and Relevant Knowledge?
I also would like to know how Media Metrix and RK drawn business
samples from what sources.
Thanks for your help.
- The Media Guru Answers(Friday, July 17, 1998 ):
"Freshness" is not necessarily a desirable element in sampling. Panel data, such as is provided by some services mentioned is explicitly not based on fresh samples.
Results based on a sample of 10,000 are about 3 times more stable than results based on 10 consecutive samples of 1000, and equal to a result calculated from a consolidation of the 10 smaller samples, assuming both samples are equally random within the universes they are supposed to represent.
From a practical point of view, the cost of recruiting and installing a sample precludes major "freshening." Virtually all syndicated surveys with large samples (10,000+) and frequent reports (e.g. monthly) are based on panels, rather than new samples for each report. The natural churn of panels is probably 20-30% per year.
The services you inquire about are relatively open about methods. They've been written about in the trade press and the respective web sites are also informative.
- Tuesday, June 23, 1998 #1917
1)I have heard the concepts "awareness" and "response curve"
But I need more detailed explanations for them. E.g.
what kind of researches are needed, how to judge the
findings, how to use these results to improve/evaluate
a tv ad. schedule...
2) What do you think about "conversion factor" which
represents an index in terms of ratings for a target
based on another one. The point makes me unconvinied
with this concept is: Some targets are heavy wievers
and some are not. But there is nothing for these differences
in the "conversion" calculation.
- The Media Guru Answers(Thursday, June 25, 1998 ):
1) Awareness and response curve can both have more than one meaning. Advertising awareness is a result of quantitative, random sample survey research, where questions are asked to determine whether respondents can recall ads for a prodcut and/or what elements of the advertising they recall, e.g. copy points, which medium was seen, etc.
There is unaided awareness, i.e. "When you think of toothpaste, what brands' advertising have you seen?" and
Aided awareness, i.e. Which of these brands' advertisng have you seen?
There is also brand awareness, considered without regard to advertising.("What brands of toothpaste are you familiar with?")
Response curve too, can mean many things. The "curve" part just refers to plotting on a graph with one axis representing some form of behavior such as purchase, purchase intent, ad recall, brand awareness versus another axis representing some stimulus, such as advertising weight or promotional effort.
2)Conversion factor does explicitly account for differences in viewing behavior between one target and the next. For example, if a certain program or daypart has an average Household rating of 10.0 and an average women 18-49 rating of a 6.4, and an average Men 18-24 rating of 2.5 there is a conversion factor of .64 for the W18-49 and a factor of .25 for the M18-24.
This difference is because of the difference of heaviness of viewing of the specific programming by these demographics. Their general heaviness of viewing relating to any other dayparts is irrelevant here.
- Tuesday, June 09, 1998 #1886
how do i calculate reach of TV+PRESS, Is there a formula
- The Media Guru Answers(Tuesday, June 09, 1998 ):
As a rule, TV and press are thought to duplicate in a random pattern. That is, the random duplication formula is appropriate. The reach of each medium is treated as a decimal. To calculate net reach, we combine the probabilty of each medium's NOT reaching the target, to get the combined probability of neither reaching the target. The remaining people are the ones reached.
The formula works as follows when TV reach is 45 and press reach is 37.
People not reached by TV would be 0.55 of the target
People not reached by press would be 0.63 of target
Total people NOT reached are 0.55 x 0.63 or
0.35 of target.
The remainder of target is reached (1.0 - 0.35 = 0.65)
so reach is 65
- Friday, May 29, 1998 #1613
1.what is osto's model?
2.In case of an absence of duplication data for publications, how do l calculate
the effective reach using 2 or more media vehicles? in such a scenario, is it safe
to use the random theory even if multiple readership is negligible?
- The Media Guru Answers(Tuesday, June 02, 1998 ):
1) The Guru is not familiar with Osto's model. It may be specific to India, from where you are writing.
2) The random method is a starting point. If you can find two other similar publications with measured duplication, you can use the duplication ratio from those publications. If you literally mean "effective reach," that is, reach at or above a minimum exposure level, then you need a more complex formula or a computer program like Telmar's ADplus.
- Wednesday, March 25, 1998 #1553
How can be measurement error calculated? I would like to know is there any correlation between sample size and data validity? Thank you
- The Media Guru Answers(Monday, April 06, 1998 ):
Sample size and data reliability are in a "rule of squares" relationship:
A sample four times as large is twice as reliable. Note that "reliable" is the statistical term referring to the chances that a duplicate study with the same size random sample will get the same results, give or take a specified range of error.
"Validity" refers to correctness. It might have to do with whether a question asked can corectly produce a result that is desired. For example, a question like "What will you have for breakfast a week from Tuesday?" may not be a valid predictor of what people will actually eat on that day. But, with a proper sample, it will be reliable in predicting with a set degree of variability what people will eat.
The formula for calculation of error for a given sample is:
The Square root of (P x Q over N)
p = the percentage result to be tested (e.g. 10% of the people will have bacon)
q = the complement, or difference vs 100% (if p = 10% then q = 90%
n = the sample size
So, if a sample of 500 produces the result that 10% will have bacon, the sampling error for this result is
the square root of (10 x 90)÷500 or
so the answer of 10% should be read as "between 8.658 and 11.342" and
really means that 68% of the time the same study repeated would produce a percent of bacon eaters between 8.658 and 11.342.
If the sample is quadrupled, to 2000, then the error is halved, to 0.671.