SAN FRANCISCO – Salespeople are notorious for dumping incomplete contact information into Salesforce.
After returning from a trade show, for example, they may dump names and titles of prospects into Salesforce, a customer relationship management system -- with no email address or phone number -- then pat themselves on the back for feeding the system with hundreds of new leads. Other business units are culprits as well; they may buy lists with poor-quality data and hurl them into Salesforce without vetting them for authenticity, completeness, or duplicates. The result, of course, is a contact database that's bloated with bad information.
Experts say this approach to stuffing the system without regard for whether the leads are of good quality is like building a castle on sand. Increasingly, though, companies are coming to realize that poor-quality data is detracting from bottom-line sales and productivity -- miring the business in chasing down better information rather than cultivating customers.
Houston-based PROS Inc., which helps companies create pricing strategies for their customers through data analysis, was in a rut with its data. It had duplicate, dirty data. But it also had a lot of "white space," or incomplete data.
Rob Zelinka, senior director of IT at PROS, discussed some of the challenges during the session "Data You Can Trust" at Dreamforce 2014. "We needed a single source of the truth, Zelinka said. "Records [were] stale, inaccurate, incomplete. So people tend not to trust the integrity of data. Then they go back and seek clarification on the data, which affects agility."
Zelinka and his team decided that they needed third-party help to address the dirty data problems and brought in Endiem, a Salesforce consulting partner in Houston. Geraldine Gray, a consultant at Endiem, noted that the first step is to eliminate the mistrust of data among business units.
"People shouldn't have to worry about the data not being good," Gray said. "Your job is to make the data clean."
Best practices for cleaning dirty data
Gray walked attendees through tips for Salesforce admins and other data stewards to keep Salesforce information clean.
Identify the problem and who is creating it. Gray said it's critical to prevent users from dumping, say, 50,000 leads into Salesforce. "How many of those leads are real? It’s junk," Gray said. "It’s better to get 3,000 really good leads," Gray said, that are qualified and complete.
Create policies and guard rails for data. Gray said that users need to understand the process and the policies that surround well-manicured data. "Explain to them, 'If you're going to dump leads into the system, you’re going to score all of them, run them through Marketo, Pardot or something before you can dump them into Salesforce,'" Gray said. "Then you have a warehouse or holding room where you put that data, then run it through a marketing program [for quality checks]."
Validate and monitor. Gray also suggested using other tools, such as Data.com, to validate the data before dumping incomplete records into Salesforce. "Before you dump a spreadsheet of 3,000 leads into Salesforce, make sure you have all the fields that are required; email address, industry -- the information that’s important to your company," Gray said. "Think about how you can help people get their data to a good place."
Eliminate duplications. Gray explained that wading through duplicate data is a massive waste of human effort and money. Gray noted that Salesforce has just released a new de-duping tool that's free and "does cross-name matching, fuzzy matching -- all kinds of fancy things. Think about how to use that new tool to stop dupes [from getting into your system] in the first place," she said.
Improve Web to lead. If your company generates leads from your site, partition the data away in Marketo, Pardot or whatever tool you use until you can qualify these leads and match them on your email application. "There’s also a tool on the Salesforce AppExchange called RingLead that will stop dupes from getting into your system," Gray said.
Educate users about contact uploads. New releases, such as the iPhone, can wreak havoc on the Salesforce database. Make sure that users understand how to bring business contacts and not personal contacts into Salesforce after a new release.
List imports, integrations. You need to find a way to de-dupe before you import records into Salesforce. Gray used to work for a company that dumped 100,000 Salesforce records every night. "They couldn’t understand why dumping data from ERP without cleaning it up is a problem," she said. "There are Gmail and Outlook paid tools that indicate whether a new lead exists before you bring it into the system. I used to work for Appirio, so I use Appirio Cloud Sync, which is $4 per user, per month."
Defend your system from dupes. Duplicates travel in packs, said Gray. Identify some of the low-hanging fruit of dupes that travel together. For example, organize data so that you know "Bob Smith" and "Robert Smith" are the same.
Train users to Search before Create. It's easy for users to create new contacts without checking whether those contacts already exist in Salesforce. Gray found a workaround by writing code that overrides the New Contact features and diverts users to a search bar to look for the contact in Salesforce by email.
Turn off Quick Create. The Quick Create feature doesn't run validation rules, so that is the "easiest way to get dupes into your system," Gray said.
Work on data completeness. In the case of PROS, the company "was missing a lot of data. We ran Field Trip, which is on the AppExchange," Gray said. Field Trip shows which objects are missing data and indicates a percentage of completeness on data. It will indicate places that have unused fields that someone may have thought "was a good idea at the time but has since been abandoned," Gray said.
Flip the record type. If you go onto Success communities for Salesforce, if you have Enterprise-level or above, you can use record types to manage which fields are required on the page. You can progress the data that's required as your sales cycle progresses. If you're in oil and gas -- with a long sales cycle, it's important not to ask for too much information up front. At the beginning, you may ask for five fields, then over time you can add the number of fields required as opportunity progresses.
Use free data dashboards. These dashboards can help you see where people are creating records, then not completing them. "It may show you where salespeople are sandbagging pipelines," Gray said.
Create a custom text field. Gray recalled a data import she had done at a previous job. "They didn't have access to the [application programming interface]. I screwed up. I spent the entire weekend deleting by hand 5,000 records. So I always have a custom field called Migration ID: Month, month; year, year and my initials and version. Then when I load the data, I use that as one of the field values in my imports. And if I screw up, that is my backup to ensure that I don't have to delete all your records."
Turn off workflow rules. If you don't turn off workflow rules and you have Marketo- or Pardot-type apps, you're going to spam all of your customers.
@mention users. If users create records that are duplicates or incomplete, shame them on Chatter, the Salesforce social network. "You can do an @Rob mention and say, 'You didn’t complete your record,' and put them on the wall of shame," Gray said.
According to Gray, @mentioning users makes sloppy data practices public knowledge throughout a company. It's almost like writing a note on the fridge that says, "Who stole my lunch?"
For more on the Salesforce conference, check out our DF14 guide here. Share or retweet this story at #DF14.