|
|
|
|
|
|
How to Improve Incremental Results
This entry was posted on 12/24/2007 1:17 PM and is filed under Measurement,Direct Mail.
“If
you’re not being asked to report on—and be measured by—incremental gains, you
will be,” predicted Harte Hanks Channel Optimization Strategist David
Funsten.
Funsten
joined Harte Hanks’ Executive Creative Director, David Nichols, for a December 20,
2007 Webcast entitled, Optimizing
Incremental Results for Deposit/Loan Acquisition and Cross-Sell Programs,
hosted by OnsiteConference, Inc. a privately held research marketing firm
located in Tampa, Florida.
An
incremental result, according to Funsten, is the difference in purchase activity between a group that was targeted with
some type of marketing (advertising, direct mail, e-mail or telemarketing)
compared to a control group (not treated).
For example, if your mailing generated a .45% response rate, and your
control group yielded a gain of .19%, your incremental result would be the
difference, .26%.
“You then use your incremental result to show the
true effect of your ROI and marketing,” commented Funsten. “If you mailed 200,000 pieces at a cost of
$120,000, and as a result generated 520 incremental equity lines/loans (200,000
x .26%), you would have an incremental cost of $231 per account.” He added, “if
you just used your overall response rate, and concluded your cost per account
was (200,000 x .45% = 900) $133, you would not gain an accurate picture of your
efforts.”
If you segment your
responses, you may further use your incremental gains to understand such
factors as:
“What customer or prospect
segments are accountable for the majority of the results?”
“Which media (DM, EM or
Phone) is most effective for this promotion?”
“Are we spending enough on
the promotion?”
“If
your goal is to maximize your marketing investment, traditional response
modeling can be inefficient because it includes customer segments that would have
responded without treatment,” Funsten noted.
“Incremental modeling considers only the population that reacts
positively to marketing treatment, thereby increasing marketing efficiency.”
This is done by identifying
the ‘right’ audiences as well as developing a fully integrated direct marketing
program.
Use of control groups are an
essential element of an incremental result calculation. Most direct marketing programs today,
according to Funsten, include control groups.
Incremental results are measured
and evaluated by the difference between the buying activities in a household
who received some type of direct marketing treatment compared to those who
don’t.
For a proper control group, the
essential requirement is that you include exactly the same type of people as in
your test/treatment group (random nth select).
Ideally, the control group would be a static control group, which is a group that has not received a
similar marketing treatment within at least in the last 12 months. Finally, the control group needs to be of a
size sufficient large to yield valid results.
Funsten suggests a rule of
thumb to use when estimating an appropriate sized for the control group: The group should be sized to yield at least
100 purchases at a base level of expected response. For instance, if your expected base level of
response is 1%, then you would need a control group of at least 10,000 (10,000
x 1% = 100). Another way to calculate
the approximate size of the control using this rule of thumb would be:
100 purchases / expected baseline response rate =
minimum control group size.
According to Funston, when marketing
programs fail to produce incremental results it is often due to such factors
as:
- Poor targeting
- Undifferentiated
offers
- Not capitalizing
on multiple response channels
- Inadequate
contact frequency
- Improper
creative strategy and execution
Segmentation
and modeling are key elements to assign the proper offer. Funsten offered three segment examples regarding
successful checking campaigns. “For high
checking balances, bundle discounts on
deposit and investment services. Offer aggressive cash bonus and financial
planning. For medium checking balance
households, Bundle discounts on credit and insurance services and offer cash
bonus or premiums. To attract low
checking balances, emphasize free checking but impose fees for NSF, excessive
checking writing. Encourage use of
remote channels such as ATM and online access.
And offer premiums,” recommended Funsten.
Businesses
with large universes of actual and potential customers will tend to benefit
more by developing incremental models including the use of randomized and
static control groups to isolate benefits of different marketing treatments.
Incremental
models are particularly effective when many factors influence a customer or
prospect’s behavior, such as in highly competitive markets with multiple
channels and communications.
“In
my experience,” stated Funsten, “incremental results grow over time. Repeat
contacts to the top-response deciles of a targeting model improve incremental
results. This means that campaigns
should be conducted involving multiple contacts, and that results should be
measured over a series of mailings/campaigns.”
|
|
|
|
|
|
 |
|
