The Direct Marketer Digs Into Multichannel Analytics
The direct marketer has a specific mercenary goal in mind: he reaches out to customers, whether the entire public, a targeted segment, or even a single individual, with the goal of generating a sale.
Set a Strategic Communications Plan
At the highest level, the DM starts with a strategic plan, generally for a year at a time, that delineates which messages he plans to send to what audiences over time, including repeated attempts to get his message to the same audience.
Based on his previous experience marketing in this fashion, he uses a predictive model to calculate the potential revenue generated, and weights that against the cost of generating that revenue.
The direct marketer also conducts a number of experiments and ad-hoc tests where he lacks the information to predict an outcome, using a sample drawn from his database of leads.
As for scheduling, marketing may be sent out on a schedule (monthly, quarterly, etc.); it may be rules-based (if someone buys a specific product, trigger a direct message to sell them another); or it may be based on an event (a person gets a message on their birthday)
Customer retention may also be part of the direct marketer's strategy: the direct marketer may find that it is necessary to send an offer to a customer who won't respond, simply because his research indicates that the customer may respond better to a future offer if he's been on the list in the past.
Predict Individual Response and Value
EN: The author does a soft of convoluted dance at statistical analysis - he chases it all over the field and doesn't really get in a good kick - but fundamentally, what he's trying to describe is:
- Use data of various kinds (demographics, socioeconomic, attitudinal, etc.) to classify customers and define segments
- Correlate these factors with the likelihood of achieving a desired outcome
- Run a regression analysis to predict the likelihood of future success with several target audiences
- Derive the likely outcome, in terms of its effects on the bottom line, to justify the expense.
Execute Campaigns
The author belabors some of the basics of executing a direct marketing campaign. However, this discussion belabors some of the heads-down procedural stuff, such as cleaning the data and determine whether to batch-process or set rules for conditional execution, etc.
Old hat to anyone who's studied or done DM, so I'm eliding the details.
Measure and Attribute Responses
Measuring responses is key to determining the success of the campaign.
In some instances, the effectiveness of a communication can be measured by counting the number of sales that are flagged with a source code from a coupon, or looking at a time frame following the publication of an ad.
Where additional data is collected when a sale is made (phone number, delivery address, etc.), this can also be compared to the distribution list to determine whether the buyer was reached by an advert.
There are a number of arguments as to whether receiving an ad caused a person to buy something they wouldn't have bought anyway, whether a person might be drawn to a retailer to buy a different item than was promoted, whether they will share the advert with others, etc. - but none of this is measurable, and it's generally assumed that the difference in sales between the time before and the time after an ad was sent is attributable to the ad itself.
A droop-off report is used in some instances to model the number of individuals who follow through various steps ion a process: of all the people who received an ad, how many came to the showroom, home many test-drove a vehicle, how many negotiated about the price, how many qualified for financing, and how many actually bought. When compared to similar campaigns, it may be possible to identify steps along the way that are obstructing the flow.
Metrics for Direct Marketing
Some of the data associated with DM campaigns are:
- Reach: number of promos received (not returned as undeliverable)
- Number of responses: This may be broken down by types of response, also expressed as a percentage of reach
- Number of Conversions: The number of actual sales
- Lift Over Control: Compares performance of an ad against a control group used to test it
- Revenue: gross revenue, actual or predicted
- Cost per Response: Divides the cost of the campaign by the number of responses
Other data associated with DM customers are:
- Promotion history: sets a flag indicating a customer has received a particular promo
- Response history: a separate flag indicating when a customer responds to a promo. It may be more than a flag, and indicate the response type (inquire or buy), whether they responded as expected, and if there was any halo (purchase of non-promoted items)
- Transaction History - A list of purchases by a customer regardless of promotion
- Segmentation membership history - Indicates the customer's status with the company over time (generally by frequency or amount of purchases over time)
- Lifetime Value - A figure that indicates how much the customer has spent with the firm, and how much they are likely to spend in future