advertisement

B2B Prospect Scoring…Getting to a More Targeted Prospecting List

56 %
44 %
advertisement
Information about B2B Prospect Scoring…Getting to a More Targeted Prospecting List
Marketing

Published on March 14, 2014

Author: KDDanalytics

Source: slideshare.net

Description

Spend fewer marketing dollars to find the same number of high potential prospects; or find more high potential prospects with the same marketing budget...either way predictive prospect scoring of B2B marketing lists can help you come out ahead.
advertisement

www.kddanalytics.com v1.0 Click to edit Master subtitle style B2B Prospect Scoring Getting to a More Targeted Prospecting List

www.kddanalytics.com Problem: How do I find the best prospects to market to? •Sales volume •Revenue ($) •Repeat buyers •Multiproduct buyers •Etc •How do I determine which ones are similar to my customers? •How do I prioritize a list so that marketing/sales reach out to the best prospects first? But which ones should I give to marketing??? I have a list of 19m business sites. I know who are my best customers.

www.kddanalytics.com Problem: How do I match best customers to a marketing list? But which ones should I give to marketing??? I have a list of 19m business sites. I know who are my best customers. “Match” customer attributes to those in your marketing list: 1. Firmagraphics: NAICS/SIC employee size, site sales, site status (branch, HQ, standalone), enterprise employees, etc. 2. Sector-Specific Fields: IT spend, PCs, DP spend, health insurance spend, MDs, fabrication machines, etc. How to Match: 1. Append data to customer list; 2. Select modeling sample of customer and non-customer sites; 3. Statistically derive model that “explains” a best customer in terms of firmagraphics and sector-specific fields.

www.kddanalytics.com Problem: Which prospects should I give to Marketing? But which ones should I give to marketing??? I have a list of 19m business sites. I know who are my best customers. Prioritize matched sites: 1. Apply model (“score”) to entire marketing list; 2. Sort by highest to lowest likelihood that a site matches the attributes of your best customers; 3. Give to marketing the top X% of list (e.g. start with top 10%, then work your way down the sorted list) Bus Site ID Rank Index 366043194 2 5.206 893096294 7 2.537 309612334 15 0.215 155334662 16 0.149 8024609 18 0.070 879091777 19 0.069 588339068 20 0.064 112495115 22 0.037 82260450 26 0.017 948960617 27 0.015 48952922 28 0.008 406550483 29 0.007 355478638 34 0.003 347713194 36 0.003 693130478 37 0.003 343722416 41 0.003 8865965 43 0.003 779265997 46 0.002

www.kddanalytics.com Problem: But how good is my matching? But which ones should I give to marketing??? I have a list of 19m business sites. I know who are my best customers. Assess Model “Lift” 1. As part of the modeling/matching process, compare what the model predicts with what is actually true in the customer sample (“internal” validation); 2. After sorting your sample from highest to lowest predicted likelihood of a match, you should find that actual customers tend to skew towards the top; 3. The degree to which this is true is called “lift”; 4. Secondary assessment makes use of actual marketing campaign data, a campaign that made use of the model’s scores (“external” validation).

www.kddanalytics.com Case Study A client was using an “off the shelf” product to identify prospects in a B2B marketing list; This product matched customer sites using a limited set of firmagraphics; Client wanted to see if this matching could be improved; Improvement metric used = lift It should be noted that there is a huge difference between matching customer attributes in a marketing list and converting these prospects into customers. The later depends on the skills of the marketing and sales departments and, hence, is beyond the control of the modeler. It should be noted that there is a huge difference between matching best customer attributes to a marketing list and converting these prospects into paying customers. The later depends on the skills of the marketing and sales departments and, hence, is beyond the control of the data scientist.

www.kddanalytics.com Case Study Client provided sample of customers and non- customers; The baseline was a simple model using 5 firmagraphic fields in their raw form: Enterprise employment and revenue; Site employment, type (HQ, branch, standalone), NAICS6, Metropolitan Statistical Area (MSA) The study looked at: Whether the addition of sector-specific fields (in this case, IT fields such as spend, PCs, servers, etc) could appreciably increase lift; And whether more careful attention to data preparation and modeling techniques could further incrementally improve lift.

www.kddanalytics.com Case Study The base line model, using a simple logistic regression model, identified ~75% more customers (as compared to using no model) in the top decile. The “lift chart” should exhibit smoothly declining lift as one goes deeper into the sorted file (i.e. lower deciles) this one does not. Lift chart created by sorting the scored file from highest to lowest predicted likelihood of a match, dividing into deciles, finding the actual occurrence of customers in each decile and expressing this occurrence in terms of an index. Lift chart created by sorting the scored file from highest to lowest predicted likelihood of a match, dividing into deciles, finding the actual occurrence of customers in each decile and expressing this occurrence in terms of an index. Firmagraphics Model Performance (Deciles of Actual Customers) 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 1 2 3 4 5 6 7 8 9 10 Decile Lift Testing Model No Model

www.kddanalytics.com Case Study However, with the addition of sector-specific fields, the discriminatory power of the model improved considerably, with about a 15% gain in lift in the top decile. Note how the lift chart is now smoother and exhibits a more pronounced decline in lift from the highest to the lowest decile (1 to 10). Note how the lift chart is now smoother and exhibits a more pronounced decline in lift from the highest to the lowest decile (1 to 10). Firmagraphics + Sector-Specific Model Performance (Deciles of Actual Customers) 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 1 2 3 4 5 6 7 8 9 10 Decile Lift Testing Model No Model

www.kddanalytics.com Case Study Finally, using a more sophisticated data preparation and logistic modeling methodology, including the consideration of additional firmagraphics, added an additional ~5% to the model’s lift in the top decile. Methodology Model Performance (Deciles of Actual Customers) 0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 1 2 3 4 5 6 7 8 9 10 Decile Lift Testing Model No Model Using the model’s score in decile #1 => more than 2x the likelihood of identifying a site that is similar to a current customer (compared to using no model). Using the model’s score in decile #1 => more than 2x the likelihood of identifying a site that is similar to a current customer (compared to using no model).

www.kddanalytics.com Case Study In sum, both data and methodology matter. In this study: Using sector-specific fields in addition to a base set of firmagraphics adds ~15% to lift; More careful attention to data preparation and modeling methodology adds another ~5% to lift. The result is the potential identification of more prospects that match the attributes of your best customers: Customers by Decile 0 100 200 300 400 500 600 700 800 1 2 3 4 5 6 7 8 9 10 Decile Firmagraphics Firmagraphics + Sector Methodology Across top 3 deciles (30% of the file), methodology identifies 18% more actual customers than firmagraphic based model. Just adding sector-specific variables to firmagraphics identifies 12% more customers.

www.kddanalytics.com Who are we? A team of statisticians, economists, business and industry subject matter experts; Specialists in database marketing analytics (market sizing; market simulation; segmentation; predictive prospect, campaign and churn scoring; etc), with extensive experience in the B2B space; Data junkies who love to get their hands dirty integrating, cleaning, modeling and visualizing client and 3rd party databases; Experienced professionals with particularly deep experience in the telecom, IT and energy industries.

www.kddanalytics.com Who do we serve? Our clients range from large providers of B2B marketing data to small consulting groups; We typically wholesale our services but can and have worked directly with end users.

www.kddanalytics.com Contact Info Let us know how we can help you: info@kddanalytics.com www.kddanalytics.com

Add a comment

Related presentations

Brands are more invested today than ever before on curating and distributing paid,...

Marketers need to be creating and publishing original content across many differen...

As content marketing continues to increase in popularity in every industry, more m...

Il Direct Email Marketing (DEM) è una tipologia di marketing diretto che usa la po...

This presentation contains all 120 rules from Part 1 of the 2nd edition of "Email ...

Olá, somos o Paulo Bernardes e o Pedro Silvestre, temos uma ambição em comum de me...

Related pages

Data Preparation Model Validation Data Modeling

Who we are... ... To learn more about our capabilities, ... A Case Study in B2B Prospect Scoring: Getting to a More Targeted Prospecting List.
Read more

Media and Marketing - BrightTALK

Join our new marketing analytics webinar with b2b marketing expert ... list? •What should your first marketing ... more efficient and targeted ...
Read more

Terminator vs Terminator 2: Judgement Day - A case study ...

Share Terminator vs Terminator 2: Judgement Day ... • Write a pros vs cons list ... A Case Study in B2B Prospect Scoring: Getting to a More Targeted ...
Read more

Targeted Case Management - A Model in Progress ...

document.write(adsense.get_banner_code('200x90')); Slide 1 Targeted Case Management ... More Topics. Search; Home; Documents; Targeted Case Management ...
Read more

Targeted Case Management - A Model in Progress ...

document.write(adsense.get_banner_code('200x90')); Slide 1 Targeted Case Management - A Model in Progress Presentation to PAC October 16, ...
Read more

B2B Technology Marketing - BrightTALK

Integrated multi-channel content distribution tactics to reach both larger—and more qualified ... analytics to boost their B2B marketing ...
Read more

Resistance YourCoachingMatters.com 2013 Edition ...

Focus on FSBO Resistance YourCoachingMatters.com 2013 Edition ZERO Expired Expireds A Month Three How to List Prospecting Week 3 Homework Week 2 ... More ...
Read more