If Your CRM Data Really is Garbage, Clean It Up!

Garbage can graphicI recently completed another project to address our CRM data quality issues for our account records in Salesforce.com.  Specific firmographic data (annual revenue, employee count and industry) determine our market segmentation and account assignments so having accurate data is critical to keeping the sales teams focused on selling and not on who should own which accounts.  The source of our original account data has varied over the twelve years we’ve been using SFDC.  Sometimes it is from purchased lists, sometimes trade shows and sometimes directly created by Sales or Marketing.  Over time the quality of this data (any data for that matter) decays so a long term solution needs to allow us to update records with current data on a periodic basis.

When I surveyed the market in late 2012 for possible solutions that integrated with Salesforce.com I found several software packages that all claimed to quickly and accurately solve the vast majority of the data quality problem.  As a rational person I didn’t believe it.  They were expensive and made claims too good to be true.  At that time I decided to expand the use of Data.com Clean to include the Data.com Premium package which, once installed and configured as an object in SFDC, systematically attempts to identify each SFDC account and match it to a DUNS number.  Once matched you can copy over to the SFDC account record any of the nearly 100 fields available in the Data.com object.  Data.com sources firmographic data from D&B.  If you use Microsoft, Oracle or SAP CRM systems you can similarly integrate the D&B 360 object in your system directly sourced from D&B.  I should mention here that in my opinion D&B data is a good choice as the primary source of corporate firmographic data but it’s not perfect.  You need to allow for a systematic approach for exceptions where their data is obviously deficient. More on that later.

If your account record doesn’t have an accurate physical address, phone number, website URL and, of course, an accurately formed account name then Data.com will not find a match with companies in the D&B database.  The pre-sales test of our database performed by the Data.com sales team indicated that about 50% of our records had sufficient data to make a match.  The folks at Data.com explained that 50% was slightly higher than typical Data.com customers for first time matches but my expectations were significantly higher.  My goal was to match something just under 100% of the accounts!  Thus, Data.com was only going to be part of the solution.

We obviously needed someone with the expertise to advise me on best practices regarding CRM data cleansing or normalization and to perform mass de-duplication based on policies rather than one at a time.  Enter Theresa DeRycke of Data Therapy, LLC.  She uses CRM Fusion’s Demand Tools so, if you go this route, you will need a CRM Fusion license before engaging her.  Theresa was an immense help in developing clever methodologies to achieve a quality account database that included the HQ corporate entity and all of its significant subsidiaries.  That’s an important distinction for two reasons.  First, the HQ entity is where annual revenue and employee count is retained in the D&B data structure.  If accounts are to be assigned to a sales team based on either of these metrics then you need to reference the HQ entity in your database, regardless of where the buying actually occurs.  Further, if sales rep ownership of a parent account is deemed to include subsidiaries (based on the assumption that a single buying group influences, if not controls, buying at the subsidiary level) then you need to document the corporate hierarchy in SFDC as well.  Theresa successfully guided the necessary revisions to our data to accomplish all of this.

The first approach I tried with Data.com was too slow because it was based on trial and error in finding the proper combination of name, address, phone and URL.  In Q2 of 2013 Data.com added a feature that permitted us to match on the DUNS number alone, disregarding any address or account name.  With this additional feature I found it was much faster to lookup the account in Hoovers (a D&B subsidiary), find the HQ entity, copy the DUNS number to the SFDC record, and then use Data.com to update the relevant firmographic data.  The Hoovers search and presentation functionality allows users to identify the best DUNS number faster than the Data.com search functionality.  This remained true even after Data.com released “Hoovers like” search functionality late that same quarter.

Data.com may have a valid role in your company if you receive a large number of leads or if your Sales Development Reps create a large number of new accounts on a regular basis but it is an expensive option if you’re only using it to periodically update account firmographics.  If I had it do all over again I would use Hoovers exclusively.

Finally, once the data is accurate you have a short time frame to implement policies and procedures to insure that the data doesn’t degrade over time.  This could include limiting the user’s ability to create new accounts or to monitor the creation of new accounts while enhancing their data quality as you go.  Along these same lines you may want to lock fields containing certain critical data within the account record so that accounts won’t be “accidently” reassigned based on territory rules and SFDC triggers.

Creating a new field in the account record for the DUNS number will allow you to update all of the firmographic data in a mass extraction, transformation and load (ETL) process using any source of data that includes the DUNS number.  As I mentioned previously, you also need to create a process that allows you to update those records which didn’t have accurate D&B data when last checked.  For this purpose I created a field on the account record which indicated the source of the non-D&B data so I could avoid over-writing it with what may still be inaccurate data from D&B.

There are a number of alternative, specialty data sources (some free) which I use to enhance some of the firmographic data.  Here are a few:

American Hospital Directory    http://www.ahd.com/disclaimer.php
iBank                                              http://www.ibanknet.com/
Charity Navigator                        http://www.charitynavigator.org/
PrivCo (private entities)             http://www.privco.com/login”>

There are also firmographic data sources which are not specific to market segments:

InsideView                                    https://my.insideview.com/iv/welcome.do
Discover.org                                 https://www.discoverydb.com/EXEC/
OneSource                                    http://custom.onesource.com/
Zoom Info                                    http://www.zoominfo.com/

If you’re planning a project of this nature, my experience is that it can take one person, one month to manually match or enhance about 1,000 accounts.  In quantifying your needs remember that a DUNS number match may still be inaccurate if the D&B account it represents isn’t at the HQ level.  Using the approaches here and over a period of six months I achieved a greater than 99% match rate for all accounts.  More importantly, the real test of accurate account data quality is when the territories are segmented and accounts assigned, none of the selling resources complained that their territory was inequitable compared with their peers; too small, yes; but inequitable, no.  This is a “first” in my career!

If your experience is different from mine, please share your perspectives or advice to further improve this laborious process.  Sooner or later everyone who is responsible for a CRM system has to deal with it.

Cheers!

Bob Bacon

About Bob Bacon

I work with global B2B high tech Sales leaders to help them enable and optimize the effectiveness of their organization Find out more about Bob here: http://bobbacon.net/blog/about/
This entry was posted in Sales Ops and tagged , , , , , , , , , , , , , , , . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *


This site uses Akismet to reduce spam. Learn how your comment data is processed.