We design and build

Marketing databases
Single customer view
Prospect pools
Data analysis
Reporting automation tools

We support and maintain

Database hosting
Database maintainence
Campaign planning
Data processing
Business intelligence reporting

We advise and consult

Project management
Data sourcing
Data product feasibility
Loyalty/CRM strategy and implementation
Supplier evaluation
Data planning



Approved Apteco FastStats Partner since 2007

Pell & Bales

Harvest – Design and build of a call centre lead generation system

Introduction

Pell & Bales is the leading telemarketing provider servicing the charity sector in the UK. It is no stranger to data services – since 2006 it has created three products designed to provide high quality leads to the charity client base: Harvest , E-Recruit and Street, each reflecting the various channels used for recruiting  Regular Donors.

Harvest is a prospect lead generation system, identifying prospects from the UK population most likely to become a Regular Donor, via outbound calls.  The system works on a basis of self-selection: individuals are contacted and asked a short battery of questions to identify their disposition towards charitable giving – recipients of the calls opt in or out of further contact.

The next stage involves re-contacting the opted-in recipients and attempting to recruit them as Regular Donors for a named charity.  The last stage involves the completion of a Paperless Direct Debit to initiate the donations.

The brief

The Harvest system had been designed by and outsourced to a database marketing specialist; Pell & Bales asked us to redevelop the product so that it could be operated in-house by the Supporter Insight team.

Of our own volition we took the opportunity to streamline and future-proof the product, as it had already been in operation for 18 months and the majority of best prospects using the incumbent’s methodology had been extracted and contacted.

The challenge

Aszent needed to develop a method of identifying best Regular Giving prospects suitable for telemarketing amidst a backdrop of an ever-increasing TPS rate, which was impacting upon the profitability of the product.   The cost of developing and running this new process had to be less expensive than the incumbent’s fees.  We also had to consider and develop a data base process that could be managed in-house, without disruption to the intensive daily operations that existed within Pell & Bales.

Our approach

Two workstreams were created, both working concurrently to minimise expended time.  Workstream 1 developed and delivered the technical solution, while Workstream 2 developed the analytical models required to deliver the commercial requirements.

1.The database and supporting systems

a)Design of the database

In essence, the Harvest system is a prospect pool containing 39 million opted-in households using the expanded ER from Experian (known as National Canvasse).
The database currently holds a minimum of 20 variables, all of which were found to be highly predictive in identifying prospects for outbound calling for charities.

The Harvest database design uses Service Oriented Architecture (SOA) calling upon 10 business processes with which to create the resultant database:

  • File verification
  • Load routines
  • Bulk  load
  • Validation of data type
  • Validation of look ups
  • PAF matching
  • Fixing the feed if errors occur
  • Matching the datafeed against the called PAF IDs
  • Validating the matching
  • Populating the Harvest database

b)The feeds into Harvest

The database takes in feeds from the various call centres under Pell & Bales, these records have been called and updated with new information, so it is vital that we can track these changes and ensure that we maintain the new information as well.  To accommodate this we built a fully versioned database structure.

The feeds/returns are posted to a secure FTP site and automatically extracted at specific times during the day to undergo the Harvest process.

c)Reporting services

A critical element of the Harvest system was the ability to monitor the performance of the calling records used by the Call Centres, e.g. how quickly were they calling the extract of prospects, what was the success/hit rate per batch/extract, provide the leads for the next calling stage…etc.  We built an automated reporting service that analysed the data feeds received against the extract that was sent.  This allowed our client to monitor the call burn rate and then extract additional prospect calling records, if needed.

The benefits to the client

Faster turnaround times are now achieved, with fewer errors in the data as there are fewer handoffs, annual or otherwise.  The process is very simple and streamlined enabling those with basic SQL Server skills capable of managing the system – indeed, the system requires little management, and has therefore reduced the amount of time expended on preparing extracts when compared with the previous version.

Pell & Bales now owns a fully automated telemarketing prospecting database system that enables them to reduce operational overheads and increase profitability and customer service, through being able to respond very quickly to increased demand for Regular Givers.

2.Lead generation – predictive modelling development

a)Analysis of past performance and method

We analysed all elements of the targeting: past selections and their respective performances, the targeting methodology, data sources, cost to service, etc.   From that we were able to identify and calculate which selections had worked, those which had not worked as well and the available universe.  Since previous selections had focused on females we were conscious that there was a potential audience in the form of male households.

b)Identification of new data sources

We identified a number of new sources of prospect data, analysed them as to their suitability for Harvest (when comparing the target audiences).   Experian’s National Canvasse was selected as the supplier of the prospect pool.  We obtained an extract of all variables available from National Canvasse and the using factor analysis, devised the subsequent 20 variables required to develop the series of predictive models.

c)Development of predictive models

The next stage involved developing a series of models to predict the likelihood of:

  • Telematching the prospects
  • Prospects becoming leads
  • Converting the leads  to a paperless direct debit (PDD) and thus becoming Regular Givers

This series of models was built for both female and male populations so that we had 6 models in total.  The models were combined and cost assumptions incorporated so that Pell & Bales could make selections based upon specific objectives, e.g. cost, volume, conversion, etc.

d)Test

The models were tested against a random sample from the National Canvasse database; they were then applied to the entire National Canvasse database (at that time numbering 39 million individuals) and each record scored.  The resultant records contained a range of scores reflecting the models that were applied.

These records were then grouped into quintiles and the results entered into a spreadsheet so that data and selection planning could take place.  We were able to make complex combinations to meet to the objectives and targets of our client.  It also allowed us to include a few test cells to gain learning about the performance of un-targeted populations such as Single Males.

e)Implement

We identified high performing cells on the basis of the predictive model scores, the combination of which would have generated Pell & Bales’ target requirements.  These were extracted from the National Canvasse database and then telematched.

f)Results

The Telematch models proved to be very accurate (for the summer extract selections the overall estimated telematch rate was 23.6% against the actual of 24.1%).  The Lead and PDD models applied to National Canvasse data resulted in a 30% improvement in ‘Combined Lead PDD Rate’ for the summer extract over previous incumbent’s selections.

Using this predictive model technique we were able to find an additional 1.1 million good prospects after telematching had been applied.

The benefits

Pell & Bales was able to increase the target number of high quality conversions for their client base, without having to worry about finding the extra volume; the charity clients were able to secure more high quality Regular Givers, without spending over budget.

The automated nature of the system meant that Pell & Bales could take over the operation of the system with only a few days’ handover.

The outcome

Harvest continues to be a successful service for Pell & Bales and their clients.

Contact:
Paul de Gregorio, GAD, Pell & Bales
pdgregorio@pellandbales.co.uk

We wouldn’t work with anyone else as we really don’t feel that there are any others within the industry who can deliver what we need. — Abi Thornton, Head of Listener Insight, Global Radio


Call Aszent on 07887 653595