| Step 1. Pick a name for your new list of prospects |
|
We suggest a connotative term such as Upsell candidates. Add a date suffix if you intend to reuse the name – Upsellers Jun06. Use this name if you want to recall the associated settings. |
| Step 2. Choose a type of service |
|
The possible choices are
- Upsell– customers likely to make a repeat purchase of a previously purchased product; use this choice for campaigns to sell a particular set of products
- Likely buyers – customers who are likely to buy something in the upcoming period; this is a good choice for a general, not product-specific campaign; now skip to step 5
- At risk – customers whose rate of purchasing has slowed enough so that they are in danger of defecting; improve retention by making attractive offers to these customers before they defect; now skip to step 4
- Cross-sell – customers likely to buy a product they have not previously purchased; use this choice to broaden the range of products purchased by selected customers; now skip to step 4
Unless you chose Upsell, skip to step 4. |
| Step 3. Chose product groups you want to sell |
|
You can pick one or many. Only customers who have previously purchased these products will be selected. |
| Step 4. Set a lower bound on the predictive score |
|
For the upsell service, the score is the purchase probability. For cross-sell, the predictive metric is New Item Score. For At risk, the metric is Early Warning Score. |
| Step 5. Narrow your selection by choosing customers within a specified range of Loyalty Scores |
|
Scores are translated into percentage rankings (Loyalty Rank, or just Lrank), and the ranks are grouped into deciles. One temptation is to select all the deciles (all the customers). If the communications cost is low (for example, email), this is not a bad choice. For a more efficient campaign, restrict the list to a smaller set of deciles. Avoid the second temptation of just marketing to your top deciles customers; there are many likely buyers in the lower ranks. Look at the Decile Distribution report. If a disproportionate percent of the revenue comes from top deciles customers, campaign more vigorously to the middle deciles. |
| Step 6. Set campaign calendar [only applies to Upsell service] |
|
Probabilities are perishable and degrade over time, so you need to pick the window of time during which the campaign will be active. First pick the calendar start date [mm/dd/yyyy] and then the duration of the campaign. |
| Step 7. Set advanced filters [optional] |
|
The Advanced button opens access to several filters on several other metrics. |
| Step 8. Calculate predictions |
|
After all the filters are set, Longbow reports the number of records that meet the selected criteria, the expected response rate and the expected revenue. If these numbers are satisfactory, proceed to the Download step. Otherwise, return to the step where you want to modify the selections. |
| Step 9. Download the records that meet your criteria |
|
Press the Download button and you will be transferred to the Download Center, where you can transfer the selected list to your local system. When you press Download, your account will be charged the indicated amount. |
Step 10. Send target list to campaign launch module |
|
Press the Campaign button and you will be transfered to the Campaign tab, where you can configure your email or print marketing campaign. |
