Lead List Quality Scorecard: Free CSV Template

A lead list is not good because it has many rows. It is good when the right records are complete, traceable, current and usable for a defined next step.

This scorecard gives you a repeatable QA process before a list reaches a CRM, dialer or research project. Download the free CSV template, replace the example row and keep the source columns alongside your business data.

The six dimensions of lead-list quality

Dimension	Weight	Pass question
Target fit	20	Does the record match the intended category, geography and customer profile?
Contact completeness	20	Is at least one appropriate public business channel available?
Uniqueness	20	Is this a distinct business or branch after normalization?
Freshness	15	Was it collected or verified inside the campaign's accepted window?
Provenance	15	Can you show the source URL and collection date?
Operational readiness	10	Are formats valid and suppression/compliance checks complete?

Score each dimension as pass, partial or fail. A simple implementation awards full points for pass, half for partial and zero for fail. Keep the component result; a total of 80 can hide either a missing phone or a missing source, and those are different risks.

1. Target fit before completeness

A complete record in the wrong market is still a bad lead. Start with the campaign definition:

Exact category or accepted category set.
City, service radius or territory.
Required business characteristics.
Exclusion rules such as chains, closed locations or existing customers.

Write those rules before inspecting the list. Otherwise reviewers unconsciously change the standard to fit whatever the source returned.

2. Contact completeness by channel

Do not reduce completeness to “has email.” In our 3,147-record snapshot, 93.6% had a phone, 73.0% had a website link, 71.3% had both and 4.7% had neither.

That suggests a four-way routing field:

phone_and_website
phone_only
website_only
manual_verification

The contactability benchmark explains the segments. If a channel is missing, do not guess personal data to make the spreadsheet look complete.

3. Deduplication needs more than a business name

The research snapshot had only one exact duplicate group under the conservative key country + city + business_name: two rows with the same name in Buenos Aires. That low number does not mean the dataset contained no other duplicates. Real entity resolution must handle:

Punctuation and capitalization differences.
Transliterated or localized names.
Shared phone numbers.
Tracking parameters and redirected websites.
Multiple branches with the same brand.
Slight coordinate differences for the same place.

Use a layered key. Start with normalized name and city, then compare phone, root domain and coordinates. Do not collapse legitimate branches just because the brand name matches.

4. Freshness must match the use case

There is no universal “fresh” threshold. A one-time market-density study may tolerate older observations; a calling campaign needs current channels; a compliance record may require a new check before each use.

The template includes collected_at and record_age_days. Define your maximum age in the campaign brief, then route older records to verification rather than deleting them silently. Keep the original source value when you normalize a phone or category so changes remain auditable.

5. Provenance is part of data quality

Every row should answer two questions: where did this come from, and when? Use source_url and collected_at. Provenance helps you:

Recheck a changed field.
Explain a record to a customer or reviewer.
Separate public business facts from later enrichment.
Honor correction, suppression or deletion requests.
Compare source quality over time.

A list with perfect formatting but unknown origin is not high quality.

6. Operational and compliance readiness

Before handoff, validate formatting and campaign rules:

Preserve international phone prefixes.
Keep URLs in a consistent canonical form.
Separate empty values from “not checked.”
Record duplicate decisions.
Apply existing customer, do-not-call and unsubscribe suppression lists.
Distinguish business-level fields from identifiable personal data.

Public availability does not automatically authorize every outreach use. The legal and privacy guide provides a general checklist; obtain qualified advice for your jurisdiction and channel.

A 30-minute QA workflow

Freeze the raw file and work on a copy.
Add the scorecard columns from the template.
Normalize names, phones, URLs and locations without overwriting raw values.
Run exact duplicates, then fuzzy candidates.
Calculate channel segments and record age.
Manually inspect a random sample plus every low score.
Export accepted, verify and rejected queues separately.
Save the rules, date and reviewer with the result.

For large files, sample at least across each country, category and source batch rather than inspecting only the first rows. Errors often cluster by query or region.

What a useful QA report contains

Report rates, not just row counts:

Target-fit rate.
Phone and website coverage.
Both/neither contactability share.
Exact and suspected duplicate rate.
Records outside the freshness window.
Records without provenance.
Accepted, verify and rejected shares.

Compare these with the 2026 digital presence snapshot only when category and geography are reasonably similar. The published sample is a benchmark, not a pass mark.

Download and adapt the template

The CSV scorecard is intentionally vendor-neutral. Add CRM IDs, owner fields or campaign-specific checks as needed, but keep the source and raw values. For a separate field-presence diagnostic, see the Google Maps profile completeness score.