This reference page lays out most of the formulas we will use to plan your local Direct Response, Retail or Brand Support Radio tests, track results, and roll out into other stations or Network buys. 

All use simple arithmetic and audience data that is readily available from most local stations and all networks.

Each section is bookmarked. Use the back-button on your browser to return to a referring page.

Media Cost Per Sale = X
Media Cost Per Customer

GM/C=X
  Set Target G = CX/M
  Set Target M = CX/G
  Set Target C = MG/X
Cash In : Cash Out

Target Audience Demographics
Test Market Station Selection
UPC Flighting Tool

OEF = Optimum Effective Frequency
Quota = INQs / CUME
Track EF & G in Test
OEF Flighting
  Steam Speed
  Drag Response

Network Rollout

This page also serves as a reference guide to several sections of my Free Planning for ROI Excel Workbook. Bookmark this page, then send for the planner if you haven't already done so.

Media Cost Per Sale = X

Media Cost per Sale is a dynamic variable that changes constantly.  Target Media Cost per Sale is pre-set budget assigned to an incremental sale, customer, or lead. 

MCPS must be a reasonable percentage of incremental revenues less variable costs, overhead and profit. It must also be large enough to fund adequate media in Direct Response Radio. 

MCPS is roughly equivalent to the retail mark-up on a similar product sold at a comparable price.  Here are some ranges of MCPS for various categories.

Free offer or loss leader with significant initial up sell: 75% to 100%.

$19.95 1-800 AMA-ZING widget: 50% to 55%. 

Default: 38%.  (We set Target MCPS at 75% of breakeven to allow for a vigorous reinvestment in additional media each month, which in turn results in very fast growth. See below.) 

On-line sales augmented by PPC, SEOPS & re-orders: 20% to 25%. 

Retail traffic-driver promotions: 6%-12% mfg + 6%-12% retailer.

High End Cost per Lead (e.g. financial counselors, lawyers, real estate developers, etc.): 3% to 6%.

Media Cost per Sale does not include production costs, telemarketing training, website development and other like line-items.  (See below.) 

The tables below show various ways of setting Target MCPS.

Breakeven MCPS is Gross Revenues less Cost, Overhead and Profit. Target MCPS should be set at about 75% of Breakeven or Maximum Media budget. In the chart below for a $50.00 unit sale: C|O|P is 50%, Breakeven MCPS is 50%, and Target MCPS is about 38%:

Target MCPS can yield exponential growth. Start with an initial Month 1 media budget of $10,000.  At Breakeven (50%) MCPS, the same budget churns itself from month to month, yielding flat revenues and unit sales.  At Target (38%) MCPS, each month's residual cash can be reinvested in the next month's media.

Compare Months 1,2,3...11 & 12 and Year I Totals in the chart below:

Caution: Available dollars must still be sufficient to achieve Optimum Effective Frequency against your Target Audience.  We do not set Target MCPS at, say, 10% of Breakeven and expect actual revenues and profits to soar accordingly.

In practice, Actual MCPS varies from month to month with Response Rate, Conversion Rate and Media Cost per Thousand. Therefore it is always a good idea to set Target MCPS below Breakeven, whether or not you plan to reinvest the monthly residual in rapid growth.

Media Cost Per Customer

Direct Response advertisers and Retailers can define MCPC based on the downstream value of the average first time buyer.  Some percentage of trial buyers reorder once, some of them twice, and so on.

Brand Managers of consumer products do not know the identity of their customers and have to guesstimate the value of a single gained or lost customer.

The method below uses simple spreadsheet logic to estimate Customer Value per $1.00 of retail or wholesale revenue for a New and an Established Brand. 

1) Set Category Purchase Cycle in Days.  In the chart below P.C. is set at (18).  Therefore, the average category customer has (20)  Buying Opportunities per year.

2) Set Four Customer Categories: Exclusive, Loyal, Often, and Trial Customers who each give some defined percentage of their annual business to the Brand.  Values shown below are arbitrary.

For a New Brand:  Most sales come from Trial and "Fairly Often" customers who like the brand well enough to give it (10%) and (40%) of their annual Buy Ops. On average 100 customers buy 550 units per year.  The average customer buys 5.5 units per year and is worth $5.50 per $1.00 of retail or wholesale revenue.

For an Established Brand: Most sales come from Exclusive and Loyal customers who like the brand so much they give it (100%) or (80%) of their annual Buy Ops.  On average  100 Customers buy 1,710 units per year.  The average customer buys 17.1 units per year and is worth $17.10 per $1.00 of retail or wholesale revenue.

Media Cost Per Customer should be set as a percentage of the annual or longer term value of an average customer.  This principle applies to Direct Response, Retail and Brand advertising.

Brand Introductions usually set a high MCPC (20-30%) in order to fund the broad reach and high Effective Frequency (3.0 to 5.0) per purchase cycle normally required to induce Trial.

A typical grocery store product invests heavily in advertising for two or three years, then recoups that investment from the revenue streams of Exclusive and Loyal customers who habitually buy the product.

Brand Maintenance & Share Defense advertisers can set a more modest MCPC (3% to 5%).  Reminder advertising requires less frequency (1.0 to 1.8) per purchase cycle at any target reach. 

Note that if an established brand loses one Exclusive Customer or Loyal to another brand or a lower priced store generic, today's loss of a single sale predicts the future loss of all (20) or (16) annual units.

Media Budgets = MCPC x Target Customers.  Given the Customer Values above and a target of (10,000) customers, the media budgets per $1.00 of retail or wholesale revenue for a New or Established brand might be:

Advertisers who use television to introduce a brand should use radio to defend it.  In spot or national buys, Radio :60s cost half as much as TV :30s.  But you need only half the Effective Frequency.  You can defend a brand in radio at 25% of the cost to introduce it in television.

BACK TO TOP

GM/C = X

In Direct Response Radio MCPS is always a function of three dynamic variables: Response Rate, Media Cost Per Thousand, and Conversion Rate - whether or not you choose to track them.  GM/C=X tracks all three variables simultaneously.

G = Target Audience Gross Impressions (in 000's) per Inquiry.
M = Media Cost per Thousand TA GIMPS.
C = Conversion Rate of Inquiry to Sale.
X = Media Cost Per Sale.

Example:  2,500 Gross Impressions yield one call.  Media costs $5.00 per thousand.  Therefore, each call costs 2.500 x $5.00 = $12.50.

You convert 40% of calls to sales.  Each sale costs $12.50 / .40 = $31.25.

GM/C = (2.500 x $5.00) / .40 = $31.25

G, M and C are dynamic variables, so Actual MCPS (X) will vary from day to day and station to station.  Of the three, Conversion Rate (C) tends to be the steadiest over time, but it exerts more direct leverage over MCPS, and hence profitability, than either Response Rate (G) or Media Cost per Thousand (M).

The chart below shows the three variables in a single week, single station buy using multiple insertions in a variety of day parts.  We obtain Target Audience Gross Impressions from media reps in advance, and place test commercials in key day parts, often at different costs per :60.  In test markets we monitor G and C only, since M can always be negotiated later on (see below).

Note that neither the dollar cost per :60 or the number of commercials bought are necessary to determine MCPS.  Similar flighting on stations of similar format will normally produce similar G and C. 

GM/C=X should be used to summarize weekly results across a broad swath of network buys. Given long-term contracts at fixed CPM and consistent Conversion, the main variable over time will be Response Rate (G). When G shows any sign of turning up, I insert fresh creative or discontinue running on that vehicle.  Proper OEF Flighting can avoid that necessity.

GM/C + TM + I + P... = X

Strictly speaking, Marketing Cost per Sale, Customer or Lead should include the variable costs of in-bound telemarketing, web design & hosting, and even the prorated out-of-pocket costs of creative and production.

Infrequent expenses such as commercial production, web hosting, telemarketing training & set up, etc. should be treated as Overhead Costs and amortized over time.  Variable telemarketing Unit Costs, such as average phone time, average TM time, and commissions can be grouped into an overall "TM" so GM/C + TM = X.

You can also choose to include Shipping & Handling charges in gross revenues (X +  S&H), or treat them as a separate profit center.

Since my job is to drive Inquiry, I use the simpler formula GM/C=X to track progress and leave the more intricate details of cost analysis to your accountant.

Set Target G = CX/M

Given stable C, M and Target X, I set Target G for new Direct Response Radio creative or for a new station or destination program using the formula G=CX/M.

If budget permits, I G-test new creative in local markets, before adding it to the national campaign. I always set Target G when negotiating buys on different formats (e.g. Country or Jazz) whose response rate is unknown. 

If (G) is within normal limits, we proceed. If (G) is impossibly low, I either try for a lower spot cost, yielding lower CPM, or pass on the buy. The table below shows an impossibly low G for a proposed Network Program renegotiated to a lower price.

Set Target Retail G = X/M

In any local retail or brand support test, track Gross Impressions (in 000's) per incremental up, purchase, or unit sale. 

Clearly Media Cost per Sale (X) = GM. 

If G = .125 (125 TA GIMPS per + unit) and local CPM is $9.50 then
X =  .125 x $9.50 = $1.18.

If target X is $.75 and network CPMs are $2.50, then
rollout Retail G = $.75/$2.50 = .300.

Set Target M = CX/G

Given stable C, G and Target X, I set Target M for a new Direct Response Radio station or program using the formula M=CX/G. This method is useful in three distinct settings.

NETWORK ROLLOUTS.

M=CX/G is the preferred method for translating local spot radio test results to network rollouts. 

In test, reliable values for G & C may be ascertained, but the practical cost of local radio is too high to achieve Target X.  The chart below demonstrates how acceptable G, C, and X can be achieved in network vehicles at a lower M.

PARITY ROLLOUTS.

This is also the preferred method for bidding on remnant bourses or opening discussions with media reps of stations with formats similar to those of our benchmark stations (e.g. News/Talk and Sports).

If (M) is within normal limits, we proceed. If (M) is impossibly high, I either try for a lower spot cost at or about Target M, or pass on the buy. The table below shows an unacceptably high (M) for for a proposed Network Program renegotiated to a lower price.  Note that G may be somewhat lower than normal, but possible in a known format.

CHANGE MCPS.

If the Target Media Cost per Sale (X) changes and Response & Conversion rates are stable, we can quickly reset Target (M).

MCPS might increase if we switch from a percentage of initial sale to percentage of lifetime customer value. That yields a higher Target (M), which can open up new media opportunities.

MCPS might decrease if the product needs to sell at a lower price. That yields lower Target (M), which can close out all radio outlets other than network or satellite.

Set Target C = MG/X

Given stable G, M and Target X, I set the target Conversion rate (C) for a new Direct Response Radio initiative using the formula C=MG/X.

This is the preferred method for negotiating a buy where the absolute price per :60 is less flexible.  The table below shows an impossibly high (C) of 26% for for a proposed Network Program host-endorsement campaign. 

In the first round of negotiation, I set Target G at the 12-Week average and Target C of 18%.  That yields a bid price of $225 per :60.  In the second round, the media rep requires a somewhat higher spot cost, which requires a lower (G). That in turn requires an Executive Decision.

CASH IN : CASH OUT

We discourage the use of traditional CICO tracking metrics such as CPK, CPO and MER in Direct Response Radio.

CICO metrics only measure the recent sales performance of an individual station or program.  They are notoriously poor at predicting or comparing the long-term performance of the total marketing effort across a variety of stations & formats.  In a sense, CICO is like a meteorologist who only reports yesterday's weather.

The chart below summarizes a five-week CICO remnant buy on five stations.  Each station was given a modest cash stipend for remnant avails.  If results were promising Week 1, the budget was doubled.  When results flagged, the funding was reduced or cut.

The actual results were hit-or-miss.  Some stations did fairly well, others poorly. CICO tracks the past but gives the advertiser no clear direction as to what to do next.  How much money should be spent on a 6th or 10th station?  How much on a network rollout?

CICO values, such as CPL (Cost per Lead)  CPK (Calls per $1,000 in media) tend to be random across a selection of stations*.   Random results are statistically unrelated.  They should not be used to predict future results on any other station.

CICO metrics mask the nature of Remnant Radio Buys.  CICO Buyers tend to buy blocks of remnant or ROS spots from friendly station reps at a low price per spot.  The common denominator is the number of dollars spent.  It should be the number of gross impressions bought.

The buyers do not ask for AQH, CUME or CPM, nor do they track G or EF.  Each week's buy is different, since remnant spots only exist when no other advertiser in the market is willing to buy them at any price. 

When  the creative finally burns out - unexpectedly, of course! - the cash committed to the last block of low price commercials is irretrievably lost.

The chart below demonstrates the enormous range of actual results on many different stations that can produce the same CICO metric "CPK=100" or "100 calls per $1,000 in media."

*Direct Response Radio advertisers can quickly measure the randomness of their CICO reports in les than a minute. Select an Excel column of weekly CPO or CPK figures, for example.  At the bottom enter AVG, and below that enter STD DEV. 

If the Standard Deviation is greater than about half the average, the values are more likely to be random, not statistically related, and hence unusable for future predictions.

TARGET AUDIENCE DEMOGRAPHICS

In Direct Response Radio, we always use stations and programs that deliver weekly cumes in which your TA (Target Audience) and QA (Qualified Audience) comprise a high percentage of total.

My Free Planning for ROI Excel Workbook asks you to plot your current and target customer profiles according to standard Media Demographics.

The first table below distributes 25,000 CURRENT and 65,000 TARGET customers across a typical TA demo break out. Left hand numbers show national population and percentages of each demo. The next two tables gather additional demographic criteria that define your QA.

BACK TO TOP

TEST MARKET STATION SELECTION

In Direct Response Radio I always try to test new creative under ideal circumstances.  AM News/Talk & Sports formats tend to pull better than FM Music formats. 

I assemble a preliminary buy-list of stations based on TA and QA Cume Rankings.  Then I select the station or stations that offer the lowest QA CPM and/or the highest QA density.  In the chart below, I'd test on WABC or WDEF or both.

As a general rule, about 70% of any station's weekly cume are P1's, that is they Prefer or give more listening hours to that station than to any other.  P1 GIMPS are the lion's share of all GIMPS.  P1's generate most of a station's response.

I add 70%-of-QA cume totals down a ranker list until I reach about 100% of the available QAs in a market. Those are all the stations you'll ever need.

Direct Response and Retail support radio should run during peak sales hours.

In Direct Response, mid days (Mon-Fri 10a-3p) are normally the best times for 1-800 or click-to commercials. In Retail you want to reach people when they are in a shopping mood - ideally on the road and headed for your store, restaurant or other facility. 

A local radio test for a consumer product should run for two consecutive or three alternating purchase cycles.  Give your own loyal customers and the occasional users of competitive brands time to run out of their current supply and enter a temporary Steam Heat mood. 

In a retail test market we use the Cume Ranker method above to select those stations whose P1 cumes add up to our target reach. You can derive peak unit sales hours from UPC data.  One method:

DOUBLE-AVERAGE SELECTION.  Plot Unit sales in some benchmark number of stores by day and hour. A 60-day or 180-day summary is an excellent baseline. Weekly reports will show any recent changes in same-store sales. Compute overall average unit sales per hour.

Group I: Select all above average hours. Compute Group I average. 

Group II: Select all above-average hours in Group I.

The graphic below shows total unit sales from 8am to 11pm, Monday-Sunday for one week.    Boldface hours are above average (106).  Highlighted blocks are above the double average (188).

Run Real People Radio during Group II hours and days.  

Use OEF (see next section) to determine the number of spots necessary to run on each station in order to achieve target Effective Frequency, normally 1.5 per purchase cycle.  Negotiate price and determine total Target Audience Gross Impressions per flight.

To the extent traffic managers will permit, distribute GIMPS by day and day part following the same pattern as unit sales.

The graphic below distributes 250,000 Gross Impressions across the days and day parts highlighted in Group II above.  On a practical basis, some days may not be worthy of support.  Tuesday and Wednesday are optional radio days.  The Thursday-Friday spots would run in afternoon drive.  The Saturday-Sunday spots would run in the 12N-6p daypart.

Track unit sales changes in the same stores and all day & hour blocks for two consecutive or three alternate purchase cycles. Compute overall G (000 Gimps per incremental unit sale).  Note significant changes in Group II blocks.

BENCHMARKET REFERENCE. If possible compare results to sales in one or two similar markets that get no radio support.  Plot baseline UPC data for the benchmarkets in advance, note changes during the test market, and assume the same changes would have occurred in the test market absent radio support.

BACK TO TOP

OEF = Optimum Effective Frequency

Effective Frequency is the approximate number of times that half or more of the cume audience of a program, station or day part have heard a  Direct Response Radio commercial during a flight.

EF is a function of two variables: Turnover (Δ) = Target Audience Cume/AQH and N = Number of Spots run in a flight. 

EF = (N / Δ) x .911

Optimum Effective Frequency Range opens at an EF sufficient to provoke a surge in response and closes at an EF where response from Active Shoppers begins to taper off. 

In test, and in sample rollout buys, I plot incremental inquiries for the opening commercial and for subsequent bundles of 5 spots against progressive EF.  In the example below, EF opens at about EF=1.13, peaks at 2.27 and closes at about EF=3.40.  (This looks a lot more complicated than it really is!)

Cume Quotas:  INQs/Cume  &  SALES/Cume

The total number of unique incremental Inquiries & Sales achieved during some length of time is always some percentage of a station's weekly cume. Cume Quotas indicate the percentage of Active Shoppers in the total market; they are likely to be the same in similar stations during similar time frames.

In the Optimum Effective Frequency Range chart above, the creative attracted 890 Inqs from Spot 1 through Spot 30, when we closed out OEF. 890/220,000 = .004.  During the entire test, 45 spots attracted 1,070 Inqs. 1,070/220,000 = .005.

We would expect, then, that for similar stations, we'd hope to attract Inquiry from 0.4% to 0.5% of cume.  We'd be surprised if a similar station format produced much larger Weekly Cume Quotas - no matter how many spots we ran!

In subsequent local ROS or network buys I will project Quota response from station cume, then schedule N number of commercials sufficient to achieve entry EF, but no more than that required to close out OEF.  This prevents Burnout in advance and can save thousands of dollars in media.

N varies with Δ (Cume/AQH) for all EF.  The highlighted section the table below straddles the same range of EF as in the WXYZ-AM test above.

The precise formula is EF = (N/Δ) x .911854, but "911" is easier to remember.

TRACK RESPONSE RATE & EFFECTIVE FREQUENCY

Opening and closing EF usually coincide with changes in G.  Tracking G requires a little extra work but is worth the effort.  I use Entry or Exit G to plot Target M in subsequent buys.

The seemingly complex tracking charts for OEF, Q, and G above collapse into one simple statement.  One station might yield skewed results.  Two or more stations, averaged, are highly likely to define expected results from future buys.

BACK TO TOP

OEF FLIGHTING

The objective of OEF Flighting is to achieve steady response from each Direct Response Radio commercial or series of commercials we run on any one station or program.

To do so, we run our OEF schedules just often enough to capture response from a fresh QUOTA each flight.  QUOTA is somewhat analogous to Purchase Cycle.

The recurring Purchase Cycle for Crest Toothpaste might be 3.5 weeks, a time frame derived from consumer research, census data, Nielsen Share data, and barcode scanning reports.  The weekly QUOTA for Crest among 1,000,000 HHs would be 3.5 weeks/52 weeks x 1MM ≈ 67,000.

The Re-order rate for an on-line or catalog brand can indicate the Purchase Cycle or Quota for a Direct Response Radio Campaign.

What is the Purchase Cycle for the first time purchase of a 1-800 AMA-ZING Widget, a personal injury attorney, or a canine nutritional supplement? Absent industrial data, we have to make an educated guess up front and refine our assumptions over time.

Our daily, weekly, annual or OEF Quota is that portion of a station's cume who Actively Shop a category, or are Actively Interested in solving a problem, AND are willing to buy direct. These Steam Heat Shoppers yield the vast majority of all Inquiries.  

In local market tests we can easily divide INQs by Cume to estimate short-term QUOTA.  In most retail consumer categories, the average weekly QUOTA is between .025% to 2.00% of the adult population.

To achieve proper OEF Flighting we need to know how often our QUOTA turns over or refreshes itself.

STEAM SPEED AXIOM:  The longer the time takes the average person to move from emotional Room Temperature to Steam Heat, the longer the time frame in which he or she might respond to a compelling commercial. The shorter, the shorter.

SLOW STEAM Category.

Our QUOTA of Active Shoppers takes longer to achieve Steam Heat.  They are in the market longer.  N commercials needed to achieve a target OEF Range (or EF ≈ 1.0+)can be spread out over several days.  The intervals between flights on the same station or in the same program can and should be several weeks, unless you have a huge MCPS and can afford saturation buys.

FAST STEAM Category.

Our QUOTA of Active Shoppers takes little time to achieve Steam Heat.  They are in the market only briefly.  N co commercials needed to achieve a target OEF Range (or EF ≈ 1.0+) should be run in only one or two days.  The intervals between flights should be shorter - once a week even.

In a typical one to two week test market, the advertising captures response from only one QUOTA of Steam Heat Shoppers.  We can infer the Steam Speed of that QUOTA from DRAG. 

Drag is is the time required for Incremental Response to return to baseline or zero.

At some point in a local test we blitz the entire cume for one day - run up to ten :60s from early morning to midnight. Everyone listening to that station that day hears your commercial at least twice.  Then we go off the air entirely on that station.  What happens during the next few days gives us an idea as to how to space out OEF flights in future buys on similar stations or networks.

Long Drag implies Slow Steam Speed. The average prospect takes several days, weeks, or even months to move from disinterested Room Temperature to 100º in your category. 

The problem you solve is at or near top-of-mind when the commercials run.   Many prospects are in no hurry to solve it.  Single-day blitz response can take several days to return to baseline.  

Short Drag implies Fast Steam Speed. The average prospect enters your category quickly (e.g. people who need a Personal Injury Lawyer after an accident).  The problem you solve is at or near top-of-mind when your commercial runs, and the average prospect is in a hurry to solve it.  Single-day blitz response returns to baseline rapidly.

DRAG INQ RATIOS = Steam Speed.  If budget permits, we can run two single-station blitzes a week or two apart.  If there are significant differences in the total INQs attracted by two single-day blitzes, we can also infer that the available supply of Steam Shoppers has either diminished, refreshed itself, or increased.

BACK TO TOP

A successful rollout from local to national radio requires two actions.

First, we gross up the OEF Range, known response and conversion rates, Target CPM, and Target MCPS to a national destination program, news network, or satellite radio channel whose weekly TA cumes and AQH are also known. 

Second, we decide on a flighting schedule based on the imputed or known Steam Speed of the Target Audience. How quickly or slowly should we distribute our commercials over time? The two actions are demonstrated below.

Gross Up Test Results.  Although we prefer to use results from at least two local station tests, the charts below summarize a translation of a single-station test (on WXYZ-AM) to a nationally syndicated destination program (Sean Hannity). Actual prices and results are hypothetical.

Simply put, we assume that G and C will behave about the same way on Sean Hannity at each level of Effective Frequency as they did on WXYZ-AM. In this case, we selected the lowest MCPS in the test as our Target X. We took the G value from Entry EF and the C value from the total test. Target M = CX/G.

Exegesis:  In test, we entered our OEF Range (at EF .79) on Wednesday when G dropped from 1.209 to .934.  We exited our OEF Range (at EF 1.70) the following Tuesday when G rose precipitously from .761 to 1.768.  Conversion was a fairly steady 26.5%.  We attracted inquiry from .23% of the weekly cume of WXYZ and sales from .06%. The lowest MCPS in the test was $22.22.

Our target rollout values (see chart left) were applied to the Sean Hannity program.  We wanted to achieve a CPM of $6.29 or less, and buy a number of commercials sufficient to achieve Minimum EF of .79 but no more than 1.70.

Projected Calls, Sales and Quotas are shown for each N number of commercials.

OEF FLIGHTING

We want to we spread out the spots bought in an OEF flight so to attract maximum response from Active Shoppers, and then given the Cume Quota time to refresh itself.

The inferences drawn from Test Market results must be verified in all future roll-outs.  Over-flighting produces decreasing response per flight and wastes money.  Under-flighting misses available sales from Steam Shoppers, but is easily rectified.  OEF Flighting achieves steady response from one flight to the next.

OEF Flighting should coincide with re-order rates or other known purchase cycles, if possible. Obviously, competitive activity in the category may require tighter flighting if Target X permits. 

A unique brand with a slow steam speed and a low MCPS, such as Dinovite, can run profitably once every four or five weeks on an appropriate station, destination program or satellite channel.  A unique brand with a high steam speed and a high MCPS, such as a personal injury attorney, can afford to run more often.  A generic brand in a cluttered category may have to run much more often.  

In the chart below, the GREEN boxes represent the hour-long blocks of a  good network destination program such as Sean Hannity, Laura Ingraham or Rush Limbaugh. The RED bars are your commercial.