How to Cleanse Bad Sales Leads: 5 Step Data Cleansing Process

1. Run a CRM data audit to surface unqualified leads

A systematic audit of your CRM or lead database involves reviewing your entire database and focuses primarily on identifying those records that contain incorrect / bad information. The purpose of an audit is not to fix the problems; rather, it is used to identify which records need to be corrected/deleted. The findings of the initial audit will serve as the basis for the remaining data cleansing process.

Conduct the audit in three phases:

• Completeness check: Review your database for completeness by checking to see if any records are missing the required fields. Some examples of common incomplete fields are email address, job title, company name and company size. Records that lack complete information cannot be thoroughly evaluated for routing purposes.
• Validity check: Verify the validity of the email addresses contained in your database using email verification software. The software verifies SMTP response codes, MX record status and active domain status. Identify ‘hard’ bounce emails (those returned by ISPs) and ‘role’ based emails (such as info @ or support @) that indicate a lead is either not validated or should be assigned to another contact within the same organization.
• Duplicate detection: Run a duplicate detection scan that utilizes the email address and/or company domain as its primary matching criteria. Group suspected duplicates together and manually review them prior to merging into a single record.

Client Example

We were hired by a B2B data aggregator who was responsible for managing a database containing 50 million business records. Due to rapid degradation in their database quality, they needed help. We used automated rule-based validation and batch processing along with manual research to continually validate and correct their database.

Ultimately, our services produced a highly accurate database that could be used to support large-scale real-time prospect queries at the level of enterprises. Instead of maintaining a static list that continuously deteriorates between refresh cycles, they now maintain a continuously updated verified record base.

To read more about this case study visit: B2B data aggregator – database validation & enrichment

Pro Tip: Schedule data audits at least once every quarter. If you do not schedule a data audit during any given quarter, your data degrades at a rate of approximately 5–6%. Larger databases with constant lead flow (> 20k contacts) may benefit from adding real-time email validation on form submissions to minimize the amount of auditing necessary.

Auditing determines what you currently have. The remaining task is determining how much of it you will keep.

2. Create a lead scoring system to identify poor quality leads

Lead scoring is a number-based system of evaluating how well a contact matches your Ideal Customer Profile (ICP) and has interacted with your content. This system is what qualifies or filters out your leads before sending them to the sales queue.

A successful lead scoring model is based on two dimensions:

Behavioral Score

Award points for engagement behaviors

• Email Opens
• Link Clicks
• Content Downloads
• Website Page Visits
• Demo Requests
• Higher-intent actions receive higher point values

Demographic Score

Award points for firmographics

• Job Title
• Seniority Level
• Industry Vertical
• Company Revenue Range
• Employee Count

The contact will need to pass the minimum threshold in both behavioral and demographic to be sent into the Sales Sequence.

All major CRM systems and Marketing Automation Systems (including Salesforce, HubSpot & Marketo) have built-in Lead Scoring modules; however, it’s the calibration of the disqualifying threshold. The lowest score at which a lead is designated as “low-priority” and moved to nurture versus moving directly to a rep.

Use historical conversion data when setting this threshold; do not use instinctive decisions.

Pro Tip: Connect lead scoring to your CRM; if new engagement information becomes available (i.e.: previously cold lead re-activates), then lead scores should automatically update in real-time.

Static scores that don’t accurately represent a contact’s most recent activity will create false negatives such as a qualified lead being incorrectly placed into nurture instead of being passed to reps.

Lead Scoring identifies which leads are worthy of action. Lead Enrichment helps determine if you can act with enough relevant information.

Here’s something most teams discover too late: their lead scoring model is only as good as the data feeding it. A contact scored as “high priority” based on a job title that’s two roles out of date isn’t a good lead, it’s a confident mistake. That’s where the unglamorous work comes in. Automated lead cleansing tools handle the routine: invalid email removal, duplicate lead cleanup, outdated lead filtering.

But they need to run continuously, not just before a big campaign. Contact data verification and CRM data cleansing are what keep lead scoring accuracy honest over time. Without them, sales prospect database quality quietly erodes, and your reps spend their week chasing ghosts. Think of it less as a cleanup task and more as basic maintenance, the same way you wouldn’t run a campaign on a list you haven’t touched in eighteen months.

3. Use B2B data enrichment to complete incomplete lead profiles

Data enrichment for sales lead is the process by which you append new or missing fields (company size, annual revenue range, direct phone number, job title, LinkedIn profile URL, etc.) to an existing record. This cannot be confused with data cleansing. Data cleansing is the removal or correction of poor-quality data. Data enrichment adds new data to an existing record. Therefore, cleanse records first and then enrich. There is little point in enriching a record that will be deleted in the next step as it wastes enrichment dollars and provides false confidence in a record that has been identified for deletion.

Enrichment is most effective when applied to records that were qualified at a level greater than your minimum qualifications but lack one or more critical attribute elements. A lead with a confirmed email address, a job title matching your ideal customer profile (ICP), and 90 days of user engagement history but lacking company size and/or revenue information cannot be properly segmented or assigned to the proper representative. Once these additional fields are appended, this record becomes actionable.

For B2B databases, enrichment may come from third party data providers; LinkedIn; publicly available company registries; and web research.

At HabileData, our data enrichment services includes both automatic field addition via data appending combined with manual research to add fields that automated systems would not catch.

Example

A California-based video communications firm wanted to enhance 120,000 customer profiles through data enrichment prior to launching an upsell campaign. HabileData found that the customer profiles lacked data regarding company size, geographic region and industry vertical. Through two months of enrichment activities, HabileData was able to update the 120,000 customer profiles.

Based upon HabileData’s enhanced profiles, the client was better able to create accurate audience segments and the client reported improved engagement rates during their subsequent marketing campaign based upon tighter targeting related to company size and geography.

Case Study: Customer Profiling and Enrichment for Video Communication Company

Pro Tip: Do not apply enrichment until all filtering has taken place. Spend enrichment money on only those leads that have been filtered above your minimum qualification criteria.

4. Remove dirty leads by using data deduplication and validation

This is an active removal process. In the first pass of deduplication, you identify and merge duplicate records, which prevent “split ownership” and “double outreach.” During this second pass of validation, you confirm that all the contact information you’re keeping in your database after merging any duplicates is correct and deliverable.

You will complete the three-step deduplication process in order:

• 1. Identify: You’ll use both the e-mail address and the company’s domain as the primary key to group suspected duplicates. If a person has multiple e-mail addresses across companies, they will generate false positive matches during the email-only identification.
• 2. Merge: When you find a group of duplicate contacts, you will merge them together into one contact record. You’ll want to keep the latest version of every field value from each duplicate. However, do not automatically merge edge cases such as two people having the same name but being employed at a different subsidiary. Since these individuals are probably not duplicates.
• 3. Flag for Review: All those records where there was ambiguity about whether they were duplicates (i.e., same name, different company email domains) should go to a manual review queue instead of either being automatically merged or removed.

Validation can occur at the same time as the above steps. Email validation will test three levels:

(1) Format/structure of the e-mail address (is it syntactically correct?)

(2) Domain activity (does a mail exchange record exist and does it resolve?)

(3) Mailbox existence (will SMTP respond to the contact?)

Also, phone validation tests if the phone number is currently active and properly assigned. Ideally, validations should run on a regular basis; i.e., not just upon creating a record. What was once true (the phone number was valid) can now be untrue (phone number was reassigned six months later).

HabileData’s data cleansing and data validation services will perform the validation layer with human review of those records that could not be cleaned up via automated processes.

Client example

A US-based lead generation company required an email list to be developed by web research. The challenges presented were multi-faceted: the target market experienced high rates of data decay; the format of collected data varied significantly among the various information-gathering sources; and there was a high percentage of outdated addresses within their legacy databases.

HabileData organized the data into standard formats; removed duplicate and incorrect entries; performed multi-layer validation; and completed additional missing field entry. As a result, HabileData provided its client with a deliverable mailing list that the client’s marketing department could use to distribute campaigns without requiring any additional review on their behalf.

Pro Tip: Cleanse/remove duplicates before ingesting new data source(s). If you add a new list to a database which contains pre-existing duplicates, it will multiply the amount of time you will need to spend cleaning up those duplicates. Remove (cleanse) the duplicate entries from the new list prior to importing that list into your existing database.

The last step cleanses the existing database as well as prevents future degradation.

5. Build a sales-marketing feedback loop to stop bad leads at the source

Bad leads recur when the process that generated them does not receive corrective information. A feedback loop is the mechanism that carries signal from your sales team back to the lead sources and ICP criteria that produced the problem.

The loop has three components:

• Reason-code flagging: Sales reps flag leads that fail in the pipeline with a standardized reason code: wrong ICP, undeliverable contact data, no budget, no authority. The flag is a data entry in the CRM, not a free-text note, so it can be aggregated and reported.
• Source-level reporting: Marketing reviews the aggregated flags weekly, broken down by lead source. A source producing a consistent high rate of ICP-mismatch flags is a structural problem, not a random quality variation. It requires a sourcing decision, not a cleansing run.
• ICP refinement: If a pattern of flags indicates that a segment of your ICP is consistently non-converting a specific job title tier, a company size band, an industry vertical the ICP definition needs adjustment. The loop feeds this back into lead scoring criteria and lead intake qualification rules.

Pro Tip: Track bad-lead volume by source on a weekly basis. A source that consistently produces flagged leads should be re-evaluated or removed from the lead intake mix – regardless of the raw volume it delivers. Volume from a low-quality source is a cost, not an asset.

With the five steps in place, the remaining question is how often to run them.

How often should you cleanse your lead database?

Sales lead cleansing is important for lead database. It should be conducted once a quarter. For databases, over 20,000 contacts with continuous inbound lead flow, move to a 60-day cycle.

When you wait for the annual cycle and let the bad data build up, you need to fix 20-25% of the database within the same window of time. That’s a reality given the average 22.5% rate of annual data decay recognized by the industry. Such a volume can disrupt your pipeline causing outbound to freeze while you absorb the load. To avoid the situation run the cleanse quarterly so each audit picks up only 5-6% of your database, only the portion that has gone stale since the last audit.

To reduce the bulk of bad leads reaching each audit, you need to set up automated email validation at the point of lead capture. This traps leads with invalid emails at the form submission layer and prevents them from entering the CRM. For large scale databases, HabileData’s data validation services cover both point-of-entry validation and scheduled batch validation ensuring bad leads are filtered out.