Frank Kavenik “5 Generations of Signal Development Models”

Extraordinary success in Signal Development strongly supports the growth of the listening audience and donor revenue.  This article can help you decide which of the five generations of signal development models would best meet your signal expansion needs.

When a radio network analyzes the acquisition of a new station or translator, it must accurately forecast future donor revenue.   Revenue forecasting models help determine the purchase price offered for an acquisition.

There are five generations of revenue forecasting models ranging from:

1G – The oldest and least accurate model, up to   

5G – The newest and most accurate emerging revenue forecasting model. 

The largest and fastest-growing radio networks use higher-level revenue forecasting models.  These networks gain a significant competitive advantage from the improved accuracy of the newer and more complex models. 

The table below summarizes the five generations of revenue forecasting models:

1G – Population-based, Single-contour Revenue Forecasting Model

Historically, the most widely used revenue forecasting model is 1^st Generation.    It is easy to understand, quick to develop, and inexpensive to build and maintain.  Unfortunately, the 1G model also produces the least accurate revenue forecasts.   

The term  “Dollars-per-person” is only utilized in a 1G model.   The four more complex and more accurate models do not calculate or use the simplified dollars-per-person concept.  In considering the acquisition of a low-cost signal, the inaccuracies of a 1G model might be acceptable.   

When considering the purchase of more expensive stations, the improved accuracy of a higher-level model may be more appropriate. 

A 1G model will sometimes overestimate the future donor revenue of a new signal.   An erroneous 1G revenue overestimate can lead to overpaying for a station and financial challenges due to lower-than-expected donor revenue.

2G – Population-based, Multiple-contour Revenue Forecasting Model

A 2G revenue forecasting model improves accuracy by compensating for the concentricity of the signal to the market population.    The three alternatives for a signal’s location are city-center, suburban reach-in, and rural reach-in.     

• A signal has high concentricity when the antenna site is near the population center.
• A reach-in signal from the suburbs has lower concentricity.
• A reach-in signal from a rural location to a population center has the lowest concentricity.       

By compensating for the concentricity of the signal coverage, a 2G model produces greater revenue forecasting accuracy than a 1G model.

3G – Demographic-based Revenue Forecasting Model

A 3G model uses detailed donor demographic data instead of general population data to forecast future revenue.   Demographic data is analyzed throughout the coverage area on a signal strength-weighted basis.  Using donor demographics for revenue forecasting improves the 3G estimates over the population-based 1G and 2G revenue forecasting models.

4G – Psychographic-based Revenue Forecasting Model

A 4G revenue forecasting model cross-tabulates demographic data to produce clusters of potential donors.      When donor demographic data is analyzed in a matrix, it forms natural clusters of potential donors.  

A 4G psychographic model typically produces several dozen naturally occurring donor clusters.   A subset of those donor clusters produces a majority of donor revenue.   By cross-tabulating donor demographics, a 4G model produces more accurate revenue forecasts than 3G, 2G, and 1G models.

5G – Donor-based Revenue Forecasting Model

The 5G Donor-based revenue forecasting model is an emerging new model that analyzes a large number of donor-specific characteristics.  Because these characteristics extend far beyond a few dozen psychographic clusters, 5G models promise to be more accurate than 4G models. 

There are two levels of the developing 5G models: 

• 5G models based on Data Science that may analyze as many as 100 donor-specific characteristics.
• 5G models based on Artificial Intelligence that may analyze as many as 1,000 donor-specific characteristics.

Because of the more complex and detailed donor segmentation characteristics that 5G models analyze, the emerging 5G revenue forecasts will be the most accurate of the five generations of revenue forecasting models.

Further Increasing the Accuracy of Revenue Forecasts with Signal Measurements

All five generations of revenue forecasting models are based on computer forecasts of signal strength and coverage patterns.  However, donor revenue is generated based on actual signal strength rather than the forecast signal strength.

The actual measured signal strength may be weaker than forecast over certain portions of a coverage area.  A station will produce less revenue than initially forecasted from areas where the measured signal strength is lower than the forecast signal strength.  

Measuring the signal strength throughout the coverage area helps forecast future donor revenue for all five models more accurately.   In measuring signal strength across a market, thousands of automated signal strength measurements are recorded.  A detailed analysis of this recorded data is then used to improve revenue forecasting accuracy further.

For example, the map below shows the forecast signal strength as the background colors on the map.  The colored dots on top of the background colors represent the measured signal strength. 

The areas where the colored measurement dots differ from the background computer-forecasted color indicate that the measured signal strength varies from the computer-forecasted signal strength.

Map of computer forecast as background colors and actual measurements as dots.

SUMMARY

There are five generations of models for use in forecasting revenue for purchasing new signals: 

1G – Population-based, Single-contour (Least accurate)

2G – Population-based, Multiple-contour

3G – Demographic-based

4G –  Psychographic-based

5G – Donor-based (Most accurate)

When considering the acquisition of a low-cost signal, the inaccuracy of a lower-level revenue forecasting model may be acceptable.

When considering the acquisition of higher-cost signals, the improved accuracy of a higher-level model may be worth the additional investment of time and money.

In addition, recording and analyzing actual signal strength readings throughout the coverage area helps further improve revenue forecasting accuracy for all five models.

—–

Frank Kavenik is a Christian Radio Data Scientist who invented the BroadcastMetrics Data Science System.  As a strategy consultant, Frank uses Data Science and BroadcastMetrics to support Christian radio leaders as they increase signal coverage, listening audience, and donor revenue.

Frank has over 20 years of Data Science experience in Christian radio.  He has completed over 900 Data Science projects serving leaders of major Christian radio networks.   Frank holds degrees in management, law, and engineering. 

Frank’s HisAir.Net articles are HERE

His LinkedIn profile is HERE

Frank’s e-mail is: FrankK@data-sci.biz