At a glance

• Most “bad MQL” complaints come from what gets in, not from how leads are scored.
• A validation gate checks every new record at the point of entry: format, domain, duplicates, role and fit.
• Use your platform’s native checks first. Add a third-party verification tool only for the gaps.
• Clean the backlog and stop new junk, but stop new junk first.
• Measure the gate: what share of leads is validated at entry, and why the rest were rejected.

When sales says “these MQLs are rubbish”, marketing teams usually reach for the scoring model. They add points here, take points away there, and raise the threshold.

Three months later, sales says the same thing.

That’s because scoring can’t fix a lead that should never have been in the system. If a fake email, a student researching a thesis, a competitor or the fourth duplicate of the same person gets into your CRM, every system after that point treats it as real. It gets scored, nurtured, routed and reported on. By the time a salesperson calls it, the damage is done.

The fix is to move the check upstream, to the door.

What junk actually costs you

Junk data rarely looks expensive, because the cost is spread across teams:

• SDR time. Every junk lead that reaches a rep costs a research session, a call or an email, and a disposition.
• Polluted scoring. Junk records that engage (open emails, revisit pages) inflate scores and teach your model the wrong patterns.
• Broken reporting. Conversion rates, cost per lead and pipeline forecasts all include records that were never real.
• Lost trust. Once sales stops trusting MQLs, they stop working them. Your real leads get ignored along with the junk.
• Compliance exposure. Records with no clear source or consent are a liability under laws such as GDPR and India’s DPDP Act.

Where junk comes from

Before building a gate, map every way a record enters your stack. In most B2B setups, it’s some mix of:

1. Web forms. Typos, fake details, bots and personal emails.
2. List uploads. Event lists, purchased lists and partner lists, often with inconsistent formats.
3. Integrations and syncs. One system pushing duplicates or half-empty records into another.
4. Enrichment overwrites. A tool replacing good data with worse data.
5. Manual entry. Reps creating records by hand, each in their own style.

Each entry point needs its own rule set. A gate that only covers web forms leaves four other doors open.

Building the gate

A validation gate is a set of checks every new record passes before it’s created or updated as a lead. Records that fail don’t disappear. They go to a review queue with a reason code.

Check 1: Format

Is the email syntactically valid? Is the phone number in a usable format? Are required fields filled with real values, not “asdf” or “test”?

Check 2: Domain

Is it a business domain or a free email provider? Is it a known disposable domain? Does the domain match the company name given?

For many B2B teams, a free-email lead isn’t rejected outright. It’s routed differently, for example to nurture instead of sales.

Check 3: Duplicates

Does this person or account already exist? Matching on email alone misses a lot. Use email, name plus company, and domain together, and decide which record wins when two match.

Check 4: Role and fit

Is this person in a role you sell to? Is the company in your ICP by size, industry and region? Students, job seekers, vendors and competitors are flagged here.

Check 5: Verification (only where needed)

For high-value entry points such as demo requests, add a real-time email verification service. Use it where native checks can’t tell you what you need to know, not everywhere by default. It’s another vendor that will see your data, so it needs the same approvals as any other.

Native first, third-party for the gaps

Salesforce, Account Engagement and HubSpot can handle most format, duplicate and required-field rules natively. Start there. Add tools only when a specific gap justifies the cost and the data-sharing approval.

Clean the backlog, but stop new junk first

Teams often start with a big cleansing project. That’s backwards. If the door is still open, the database fills up again while you’re cleaning it.

The order that works:

1. Turn on the gate for new records, even a basic version.
2. Quarantine obvious junk in the existing database (hard bounces, disposable domains, test records) so it stops affecting scoring and reports.
3. Cleanse and dedupe the backlog in batches, starting with the records closest to sales: open opportunities, recent MQLs and target accounts.
4. Enrich only what’s clean. Enriching junk just makes expensive junk.

One caution: cleansing and enrichment mean exporting records and often sharing them with a vendor. Get your InfoSec, privacy and data-processing approvals in place before you do. An NDA with the vendor isn’t the same thing.

How to measure the gate

You can’t improve what you don’t track. Add these to your marketing operations dashboard:

• Validated at entry: the share of new records that passed every check.
• Rejection reasons: a breakdown by check, so you know which entry point leaks most.
• Review queue age: how long flagged records wait for a decision.
• MQL-to-SAL acceptance rate: the number sales cares about, which should rise as the gate tightens.

Review the rejection reasons monthly. If one form produces most of the junk, fix the form, not just the filter.

Validation gate checklist

• ☐ Every entry point is mapped: forms, uploads, syncs, enrichment and manual entry
• ☐ Format checks on email, phone and required fields
• ☐ Free and disposable domain handling defined (reject, or route to nurture)
• ☐ Duplicate matching on more than email, with a “which record wins” rule
• ☐ Role and ICP fit flags for students, vendors, competitors and out-of-region leads
• ☐ Real-time verification only on high-value entry points
• ☐ A review queue with reason codes, not silent deletion
• ☐ Approvals in place before any data leaves your systems
• ☐ Gate metrics on the ops dashboard and reviewed monthly

Need a validation gate designed for your stack? It’s one of five workstreams in my MarTech Stack Development service. Also worth reading: Why is my marketing automation not working? 7 root causes.