Use GM/.C = X to Test and Roll Out

The Basics of Direct Response Radio / Part VI

  Local Radio Tests Pave
The Way to Network.

       The main purpose of a local test is not to make instant money.  It is to gather useful information.  How well does your commercial pull in different formats and day parts?  Does frequency build response? How well do your telemarketers or web designers convert Impulse Shoppers to first time customers? 

     Always test under Ideal Conditions - in your best markets, on your best stations, and on your best business days. If you're trying to beat a Control, run a low cost ROS & Streaming buy long enough to hit your OEF. If you're trying out a totally new campaign, run enough spots in your best day parts to reach a 3 frequency against your broad Target Audience Demo.

   A good local test runs for a week to 10 days, covering one weekend.  It spreads your message out well enough to track response to individual spots, and to compare response to frequency build, drag, even OEF.  Costs vary with market size and station rates.  I prefer to run in one of the top 25 markets on one or two strong stations.  Budgets range from $6,500 to $25,000, plus creative.

   I like to lead off with a 1:00am and 2:00am :60 on a big major market blowtorch, just to see if anyone calls or clicks!  A couple of Monday morning Drive Time spots seed your brand name.  Most spots should run 10a-3p Mon-Fri.  Heavy up at the beginning of the week.  Run 20 to 30 Streaming Spots a week against the unrated 3% to 12% of the cume who listen in on their computers.  Run something on Saturday and Sunday if your TMs are available. 

   Look at where TA G and QA G end up, not where they begin. If frequency does build response, you should see lower G at the beginning of the second week.  If the copy is weak, G will start high and then rise in the middle of the first week.  Kill that spot asap.

The grid below plots a hypothetical buy in which QAs are 40% of TAs.

   Track any inquiries you get after the flight ends.  Drag can roll in for two hours or three weeks! Track Conversion along the way to see if it changes, and to compute running MCPS.  In the sample above, if your QA CPM was $29.50 and you made 57 sales:

QA G    =   .903
QA M   =  $29.50
C         =  57/144 = 39.5%
X         =  (.903 x $29.50) / .395 = $67.43

   If your conversion rate was steady, you would be most interested in your last few days of G.  Look at QA G for Wednesday:

QA G    =   .667
QA M   =  $29.50
C         =   39.5%
X         =  (.667 x $29.50) / .395 = $49.81

   Don't try to do this in your head or with a hand calculator.  That's what spreadsheets are for.

 

 LOCAL ROLLOUT 

   That final range of QA G (1.000 to .667) is likely to apply to your creative until you reach burnout. You can continue to run on the same station, perhaps using lower ROS rates. You might get a QA M of $21.65 on a 4-week flight of ROS spots Mon-Fri 6a-6p.  You can project MCPS as:

QA G    =   .667
QA M   =  $21.65
C         =   39.5%
X         =  (.667 x $21.65) / .395 = $35.56

   You will likely get different TA G and QA G from different format stations.  So test News/Talk first, then Sports (if your TA is male) and then perhaps Country or Jazz.

   If your New Ideas work, keep running in that test market, station or program until OEF maxes out, then drop back to Purchase Cycle maintenance. Later on you can add new markets or run in network radio at much lower costs per thousand.

   In the short run, you can air one Control commercial on a single station or set of stations to plot your range of Optimum Effective Frequency.  Thereafter consider running a number of commercials equal to the middle of that range.  (See Multiple Selling Propositions.) 

NETWORK ROLLOUT.

If your offer has or should have national appeal, you can use local test results to predict response on national destination programs such as Sean, Rush or ESPN Mid-Days, or on big blasternets, such as ABC Prime Reach and CBS-News Net.

Similar formats tend to attract similar audiences everywhere.  So network QA G is highly likely to be about the same or similar to spot market QA G, especially if you test on a local station in a network program.  However, network CPMS are much lower than local stations'!  You pay more per spot, of course, because you're reaching hundreds of thousands, even millions of QAs.

    It is hard to get national Scarborough data on detailed definitions of Qualified Audiences. You might want to use other sources such as the US Census to estimate what percentage of P25+, for example, own a car and have graduated from college, and earn $50K+.  If your QA is about 40% of your TA, divide the network CPM by .40

    Use QA G and derived QA M:

QA G    =   .903
TA M   =  $2.89
QA M  =  $7.23
C         =   39.5%
X         =  (.903 x $7.23) / .395 = $16.53

   If your local streaming test pulls well, consider a national streaming network such as Clear Channel's CCOMR. If you have a medley of good testimonial or demo spots, consider XM or Sirius.  Each network reaches about 7MM subscribers, plus passengers!, and offers very competitive CPMS for P25+.  You can get high frequency against small but widely dispersed Loyal Cumes for not much cash.

   You usually get better response from :60s than from :30s, unless you have a massive national budget ($10MM+/year).  It takes people a few seconds to tune in to your copy.  I like to introduce a brand on a new destination program with two :60s that run about an hour apart. BAM! (Who was that?) BAM!  F1 & F2 Drag can roll in for weeks.

   You can also use network results to identify good spot markets for heavy-up buys. Most networks report RADAR gimps by DMA.  You can tote up all the impressions that run in each market and compute weekly G by market:

    I use a National Time Zone Analysis spreadsheet to monitor calls, clicks and sales before, during and after a network run.  Those states or markets that perform significantly better than expected may be worthy of concentrated effort.  You might test TV on a few CTVOs in your best markets.  Add markets, add national cablenets... one day "Today!"

   It helps to index each state's or DMA's percentage of national sales against percentage of national GIMPS.  Some states, like CA will normally index quite high at first.  Others, like MA, RI and NE will need a little prodding.

Most of what we've covered so far pertains to 1-800 Direct Response Radio.  How do the same principles apply to building on-line sales? Let's compare PPC to Click-to DR Radio...

 

Intro   DR Radio v Print    How Frequency Builds Response  How To Select Stations

GM/C = X versus CICO   How To Test & Roll Out   Click-to & 1-800 DR

ADVANCED CONCEPTS OF RADIO

RADIO GLOSSARY