Gt data audit

GT Data Audit

A 10-dimension audit framework for B2B contact data quality. Produces a 0 to 10 score across coverage, accuracy, freshness, cost, governance, and 5 other dimensions. Built for RevOps leaders, GTM Engineers, and Sales Operations at B2B SaaS past $10M ARR.

The full audit takes 3 to 5 hours of hands-on work for the person running it. The skill covers all 10 dimensions in order, scores each one against a pass benchmark, and returns a single number plus a prioritized fix list.

Two operating modes

Client-facing mode (default) For lead magnet downloads, prospect deliverables, and external sharing. Anyone can run this audit on their own stack. Voice is operator-real and jargon-free with no GT-internal references. Output formats: Notion page, 1-page PDF, scorecard.

GT-internal mode For the GT team running an audit on a live client portal. Adds internal context: client tier, engagement type, link to the GTM Alpha Playbook, deliverable templates, and a recommended next-step engagement based on the score.

Default to client-facing mode. Switch to GT-internal mode ONLY when the prompt contains an explicit marker:

• "for [client name]"
• "Velocity engagement" / "GTM Alpha engagement" / "GTM Alpha"
• "before GTM Alpha delivery"
• "internal audit" / "team mode"

If it is unclear which mode applies, ask once: "Is this for external sharing as a lead magnet, or internal use by the GT team running an audit on a client?"

When to fire

Fire when the user's message contains any of these intents:

• "audit our data" / "data quality check" / "data audit"
• "is our data the bottleneck" / "data layer for AI agents" / "AI agent data foundation"
• "email accuracy benchmark" / "what is a good bounce rate" / "verified email rate"
• "cost per usable contact" / "cost per credit" / "data provider pricing"
• "mobile connect rate" / "phone data accuracy" / "cold call connect benchmark"
• "data refresh cycle" / "how often refresh data" / "data decay"
• "CRM data audit" / "CRM hygiene" / "data governance for GTM"
• "validate our data provider" / "is Apollo enough" / "ZoomInfo vs Cognism"
• "waterfall enrichment vs single source" / "AI routing for enrichment"
• "RevOps audit" / "data scorecard" / "data foundation"
• "BetterContact audit" / "BetterContact partnership"

Also fire automatically when:

• A GTM Alpha Playbook is being prepped for delivery (audit is a precondition)
• A Clay implementation is being scoped (validate data before build)
• A prospect downloads the data audit lead magnet and books a call

Do NOT fire when:

• The user wants to BUILD a list from scratch ↳ defer to gt-list-building
• The user wants Clay workflow help unrelated to validation ↳ defer to gt-clay
• The user wants HubSpot portal admin work unrelated to data quality ↳ defer to gt-hubspot-admin
• The user wants cold email copy or sequences ↳ defer to gt-cold-email

Supporting files

Load these as needed. Do not paste a number into output unless it traces to resources/benchmarks.md.

File	Use it for
resources/benchmarks.md	Every stat cited in any output. Single source of truth. Refresh every 6 months
resources/scoring.md	Full rubric, interpretation bands, what to do at each score level
resources/crm-routing.md	Light per-CRM "where to start" guidance (HubSpot, Salesforce, Pipedrive, Close, Attio)
resources/providers.md	Tool catalog and Q2 2026 pricing across data providers, verifiers, dialers, hygiene tools
resources/audit-template.md	The 3 output formats (Notion full report, 1-page PDF, Slack snippet)
resources/sources.md	Full citation list with URLs for the quarterly refresh

The 10 dimensions

Every audit covers all 10 in order. Each dimension has a question, why it matters, audit steps, a pass benchmark, and a fix. Score 1 point per pass benchmark hit. Maximum 10.

Dimension 1: Coverage

Question: Can you actually find the contacts you need on your target list?

Why it matters: The average provider delivers only 50% accuracy on independent tests (Landbase 2026). Single-source providers return 400 to 600 usable matches per 1,000 queried on a tight ICP slice. The rest are misattributed titles, generic info@ inboxes, or contacts with no email.

Audit steps:

1. Pick one tight ICP slice (one role, one company tier, one geo). Example: "VP of Sales at B2B SaaS $25M to $100M ARR in North America"
2. Run the same exact query through 4 providers (Apollo, ZoomInfo, Cognism, Clay waterfall)
3. Count how many contacts each provider returns for the same query
4. Count how many have valid corporate email vs catch-all vs missing
5. Count how many have mobile direct dial vs office line vs missing

Pass benchmark:

• ✅ Strong: 70%+ match rate on a tight ICP slice
• 🟡 Acceptable: 50% to 69% (waterfall enrichment needed to fill gaps)
• ❌ Fail: Under 50% (single provider is leaving half your TAM on the table)

Fix: Switch to multi-provider waterfall. Waterfall enrichment hits 85% to 95% find rates versus 50% to 60% for single-source (DealSignal 2026).

Tools: Single-source: Apollo, ZoomInfo, Cognism, Lusha. Waterfall: BetterContact, Clay (with provider stack), Cleanlist, Datagma.

Dimension 2: Email Accuracy

Question: Are the emails in your list actually deliverable?

Why it matters: A "verified" list can sit at 38% bounce on the first campaign send (SparkDBI 2026). Validity's 2025 benchmark shows senders with bounce rates above 5% see inbox placement drop by 20 percentage points. Once your sending domain is flagged, every email you send is at risk, including replies to existing customers.

Audit steps:

1. Take 200 contacts from your CRM at random
2. Run them through 3 independent verifiers (ZeroBounce, Million Verifier, Prospeo)
3. Count how many come back valid vs catch-all vs risky vs invalid
4. Compare each verifier's score against the others (catches verifier disagreement)
5. Send a small test sequence (50 contacts) and measure actual bounce rate

Pass benchmark:

• ✅ Strong: 90%+ valid rate on a well-defined ICP list
• 🟡 Acceptable: 80% to 89% (requires active list hygiene)
• ❌ Fail: Under 80% (disqualify the data source)
• Real-world bounce rate target: under 1% (Validity 2025)
• Hard bounce above 2% triggers spam filters at Google and Microsoft

Fix: Add real-time verification at point of capture (form, CSV upload). Verify any list older than 30 days before send. Set up bounce monitoring as a daily metric.

Tools: ZeroBounce, Million Verifier, NeverBounce, BriteVerify, Prospeo, Mailfloss.

Dimension 3: Phone and Mobile Connect Quality

Question: What percentage of your phone numbers actually reach a live human?

Why it matters: US B2B SDRs lose 35% of their day to manual dialing and dead-ring time (Bridge Group 2026). The average cold connect rate sits at 3% to 10%. Top teams with verified mobile direct-dials hit 18% to 22%, which is 2x the baseline (Cognism State of Cold Calling 2026, 200,000 calls analyzed).

Audit steps:

1. Pull 100 phone numbers from your CRM
2. Categorize each: mobile direct, office line, switchboard, missing
3. Dial all 100 and log: live connect, voicemail, dead ring, disconnected
4. Calculate connect rate by source (which providers gave you mobile vs office)
5. Check whether your data provider distinguishes mobile from office at point of enrichment

Pass benchmark:

• ✅ Strong: 60%+ mobile direct dial coverage on a US list
• 🟡 Acceptable: 30% to 59% (this gap is your highest-leverage fix)
• ❌ Fail: Under 30% (you are paying for numbers that do not connect)
• Connect rate target: 18% to 22% on verified mobile, 8% to 12% on generic

Fix: Switch to mobile-first phone providers. Verified mobile is the single biggest connect rate lever (Prospeo 2026). Mobile numbers connect at 45% higher rates than non-mobile.

Tools: Mobile-first: Cognism, Lusha, Nooks. Mixed: ZoomInfo, Apollo.

Dimension 4: Data Freshness and Recency

Question: How old is the data sitting in your CRM right now?

Why it matters: B2B contact data decays at 2.1% per month, compounding to 22% to 30% annually (Dun & Bradstreet). A list that was 95% accurate 12 months ago is likely sitting at 65% to 70% accuracy today. 23% to 30% of emails go stale per year. 18% of phone numbers change. 15% to 20% of professionals change employers.

Audit steps:

1. Look at the "last enriched" or "last updated" date on 100 CRM contacts
2. Calculate what percentage was last touched more than 6 months ago, and more than 12 months ago
3. Run those stale contacts through fresh enrichment
4. Count how many are now outdated (job change, email change, company change)
5. Look at your provider's claimed refresh cycle: weekly, monthly, quarterly, or "as needed"

Pass benchmark:

• ✅ Strong: 90%+ of CRM contacts refreshed in the last 90 days
• 🟡 Acceptable: 60% to 89% (set up automated refresh)
• ❌ Fail: Under 60% (data layer decaying faster than you refresh)
• Provider refresh cadence: weekly (top tier) or monthly minimum

Fix: Set up automated re-enrichment on a 30-day cycle for active sequences and a 90-day cycle for the full CRM. Trigger re-verification on any contact that bounces or gets a "no longer with company" reply.

Tools: Clay (auto-refresh workflows), HubSpot Operations Hub, Salesforce Data Cloud, RingLead, Cloudingo.

Dimension 5: Cost Per Usable Contact

Question: What does each contact actually cost, after you strip out the unusable ones?

Why it matters: Listed cost is not real cost. If a provider charges $0.50 per record but 40% bounce, your real cost per usable contact is $0.83. Credit-based providers (Cleanlist, People Data Labs) typically save 40% to 60% versus seat-based annual contracts (ZoomInfo, Cognism) for the same volume (Cleanlist 1,000-record benchmark 2026). Pricing across 15 tested providers ranges from $0.01 to $2.50 per record.

Audit steps:

1. Pull last quarter spend on each data tool (annual contract divided by 4, plus any usage fees)
2. Pull the total contacts enriched or pulled in that quarter
3. Calculate raw cost per record (spend divided by total contacts)
4. Apply your verified-valid rate from Dimension 2 to get real cost per usable contact
5. Compare across providers. Look for hidden overage charges, seat caps, export limits

Pass benchmark:

• ✅ Strong: Under $0.30 per usable contact on a high-volume waterfall
• 🟡 Acceptable: $0.30 to $0.80 per usable contact for mid-volume
• ❌ Fail: Above $0.80 per usable contact (audit your provider mix)
• Credit-based is typically 40% to 60% cheaper than seat-based for the same volume

Fix: Audit which providers carry the load on which lead types. Drop overlap. Move from seat-based to credit-based where volume justifies.

Tools: Credit-based: Cleanlist, People Data Labs, BetterContact, Datagma. Seat-based: ZoomInfo, Cognism, Apollo (annual).

Dimension 6: Enrichment Method and Validation Layer

Question: Is your enrichment single-source, waterfall, or AI-routed?

Why it matters: Single-source databases deliver 50% to 60% find rates. Multi-source waterfalls hit 85% to 95% (DealSignal 2026). AI-routed waterfalls (dynamic provider sequencing based on lead type) outperform static waterfalls by another 5% to 15%, depending on ICP fit.

Audit steps:

1. Map how your enrichment currently flows (one provider in CRM, chained waterfall in Clay/n8n, or AI-routed system)
2. For a sample of 100 contacts, record which provider returned the winning data point per field (email, phone, title)
3. Calculate which provider is doing the heaviest lifting, and which fire often but rarely win
4. Check whether the order of providers is static or changes based on region, seniority, or industry
5. Check whether your enrichment layer supports fallback chains, retry logic, per-field verification

Pass benchmark:

• ✅ Strong: Multi-source waterfall + AI routing (best in class)
• 🟡 Acceptable: Multi-source waterfall (static order)
• ❌ Fail: Single-source only (leaving 30 to 45 percentage points of find rate on the table)

Fix: Move to a multi-source waterfall. Add dynamic routing logic. Some leads need US-mobile providers first, others need EU-compliant providers first. Track which providers win and adjust the order quarterly.

Tools: Waterfall builders: Clay, n8n, Make, BetterContact. AI routing: BetterContact, Findymail, Datagma.

Dimension 7: CRM Hygiene and Format

Question: Is the data in your CRM structured so AI agents and workflows can actually use it?

Why it matters: 70% of CRM systems contain outdated, incomplete, or inaccurate information (Landbase 2026). Even if source data is clean, the way it sits in the CRM determines whether downstream automations work. Duplicate companies create fragmented account views. Inconsistent country fields ("US" vs "USA" vs "United States") break segmentation. Wrong title casing breaks lead scoring.

Audit steps:

1. Run a deduplication scan on companies and contacts. Count duplicates
2. Check title formatting. Are titles standardized or free-text? Example: "VP Sales" vs "VP, Sales" vs "Vice President of Sales"
3. Check country, state, and city fields for inconsistencies
4. Check email and phone number formatting normalization
5. Check which fields are required at point of creation, and which are optional

Pass benchmark:

• ✅ Strong: Under 2% duplicate rate on companies
• ✅ Strong: Standardized title taxonomy in use
• ✅ Strong: Country, state, city normalized in one format
• ❌ Fail: Any of the above missing (downstream automation will misfire)

Fix: Run Salesforce or HubSpot native deduplication first. For larger databases, use RingLead or Cloudingo. Set up validation rules at point of capture. Apply a title-normalization workflow on import.

Tools: Native: Salesforce Duplicate Management, HubSpot Operations Hub. Dedicated: RingLead, Cloudingo, Insycle, DemandTools.

Dimension 8: Data Governance and Ownership

Question: Who owns this and how often is it audited?

Why it matters: 44% of teams report more than 10% annual revenue impact from CRM data decay (Landbase 2026). The decay is not the surprising part. Most teams have no named owner for data quality. Without a single accountable person, the data layer drifts back to broken within 90 days of any cleanup.

Audit steps:

1. Name the single accountable person for data quality (title and email)
2. Map who has write access to the CRM. Are vendors, partners, or anyone outside the company on this list
3. Document the data audit cadence (daily, weekly, monthly, quarterly, never)
4. Document the data sources allowed. Which providers are approved. Which are banned
5. Document field ownership. Who decides what "lead status" means. Who owns lifecycle stage definitions

Pass benchmark:

• ✅ Strong: One named accountable person
• ✅ Strong: Monthly audit cadence
• ✅ Strong: Written rules for data sources, field definitions, write access
• ❌ Fail: Any of the above missing (the layer will drift back within 90 days)

Fix: Pick one person (RevOps lead or Head of GTM Engineering is the typical fit). Set a 30-minute monthly audit on the calendar. Write a one-page rules doc. Pin it in the team workspace.

Tools: HubSpot Operations Hub, Salesforce Flow, n8n (governance triggers).

Dimension 9: Compliance and Opt-Out Tracking

Question: Are you allowed to contact each person in your database?

Why it matters: GDPR (EU), CCPA (California), CAN-SPAM (US), CASL (Canada), and PECR (UK) all have different opt-out and consent rules. A single opt-out that fails to propagate across tools is a compliance risk. Spam complaint rate benchmark for 2025 is 0.04% to 0.06% (B2B average). Above 0.3% is the critical threshold that triggers Gmail and Outlook filters.

Audit steps:

1. Pull every contact that has unsubscribed in the last 12 months. Verify they are suppressed in every outbound tool (Lemlist, EmailBison, HubSpot, Salesforce)
2. Check whether you have a single source of truth for unsubscribes, or whether it is tracked per tool
3. For EU contacts, check whether you have a documented legal basis (legitimate interest, consent, or contract) for cold outreach
4. Check whether you honor opt-out within 10 business days (CAN-SPAM) and immediately for EU
5. Check your spam complaint rate. Is it tracked. What is the rolling 30-day number

Pass benchmark:

• ✅ Strong: 100% suppression propagation across tools (required)
• ✅ Strong: Spam complaint rate under 0.1%
• ✅ Strong: Documented legal basis per region (required)
• ✅ Strong: Single source of truth for opt-outs
• ❌ Fail: Any of the above missing (you are one audit away from a compliance event)

Fix: Centralize opt-out tracking in one tool, typically the CRM. Use webhooks or native sync to push opt-outs to every outbound tool within minutes. Document legal basis per region. Set a 0.1% spam complaint alert threshold.

Tools: Suppression tracking: HubSpot, Salesforce Marketing Cloud, Customer.io. Spam monitoring: Glock Apps, Mailgun, Litmus.

Dimension 10: Measurement and the 3 Numbers That Matter

Question: What 3 numbers should you track after every audit?

Why it matters: You can audit data quality once and feel good about it. Without ongoing measurement, the layer decays 22% to 30% per year and you are back where you started. The 3 numbers below are the minimum viable scorecard for any GTM data layer.

The 3 numbers:

Number 1: Verified email rate ↳ Target: 90%+ on active lists ↳ Pull from your verifier. Calculate weekly ↳ Trigger re-verification on any list below 90%

Number 2: Mobile connect rate ↳ Target: 18% to 22% on verified mobile data ↳ Pull from your dialer. Calculate weekly ↳ Compare against the SDR team actual connect rate. If the gap is larger than 5 points, data is the bottleneck

Number 3: Cost per usable contact ↳ Target depends on volume and ICP, but track quarter over quarter ↳ Total spend across all providers, divided by total verified-and-deliverable contacts added to active sequences ↳ Drop providers that drift above your threshold

Pass benchmark:

• ✅ Strong: All 3 numbers tracked weekly or monthly
• ✅ Strong: Numbers visible to RevOps, GTM Engineering, and Sales leadership
• ✅ Best in class: Action thresholds defined (example: below 85% verified email = pause campaign)
• ❌ Fail: No ongoing measurement (the layer will decay back within 6 months)

Fix: Build a one-tab dashboard. HubSpot reports, Salesforce dashboards, or a simple Looker or Metabase board. The point is visibility, not perfection.

Scoring

Score 1 point per pass benchmark hit. Maximum 10. Full interpretation and per-band next steps are in resources/scoring.md.

Score	Interpretation	Priority next step
0 to 3	Data layer is the bottleneck. Every AI agent and workflow built on top of it is degraded	Pick one dimension. Fix it this quarter. Do not try to fix all 10 at once
4 to 6	Some hygiene but no systematic governance. Decay catches you within 6 months	Audit your provider stack. You are probably paying for overlap
7 to 9	Strong foundation. Focus on 1 or 2 missing dimensions	Layer AI routing and predictive quality scoring on top
10	Top 5% data layer. AI agents built on top will actually perform	Maintain monthly audit cadence. Share your playbook with your team

CRM-specific routing

The 10 dimensions and benchmarks apply identically regardless of CRM. resources/crm-routing.md has light "where to start" guidance for HubSpot, Salesforce, Pipedrive, Close, and Attio. For deep portal work, defer to the dedicated admin skill (gt-hubspot-admin, future gt-salesforce-admin).

Output format

After scoring, deliver the result using one of three templates from resources/audit-template.md:

• Notion full report (client-facing) ↳ the canonical version lives at https://app.notion.com/p/38499b4b2619817f83abeed4b3ebbc37. Re-deliver the content programmatically or point the user to the page
• 1-page PDF executive summary (client-facing, condensed)
• Slack snippet (GT-internal, 3 numbers plus next steps)

In client-facing mode, end with a soft CTA: "If you want help running this audit on your stack, book a call at growthtoday.co/contact."

In GT-internal mode, end with the recommended GT engagement based on score:

• 0 to 3 ↳ Velocity rebuild (data foundation engagement)
• 4 to 6 ↳ GTM Engineering audit plus provider rationalization
• 7 to 9 ↳ AI routing layer plus signal-based plays
• 10 ↳ no rebuild needed, recommend Plays implementation

Voice rules

Apply to every output, both modes:

• No em dashes. Use "to", "and", or periods
• No contractions. "You are" not "you're". "It is" not "it's"
• Banned vocabulary: leverage, unlock, synergy, drive, empower, disrupt, transformative, robust, holistic, paradigm, game-changer
• ↳ arrows only for sub-points. Never → arrows
• Operator-real language, 5th-grade vocabulary where possible
• Specific numbers, never abstractions ("X% lift" not "significantly improve")
• No marketing transitions ("the key insight is...", "what we learned was...")
• No prescriptive "you should" / "you need to". State the principle directly
• Cite sources briefly when using benchmarks ("Validity 2025", "Cognism 2026")

References to other GT skills

• gt-gtm-alpha-playbook ↳ Data audit is a precondition. Run this skill first
• gt-clay ↳ Use the audit result to scope the Clay table build
• gt-hubspot-admin ↳ Dimensions 7 and 8 deeper dive for HubSpot users
• gt-cold-email ↳ Dimensions 2 and 9 are preconditions for cold email infra
• gt-list-building ↳ Dimension 1 (Coverage) is what list building feeds
• gt-signal-sourcer ↳ Signal-based selling assumes clean data
• gt-services ↳ This skill maps to the Velocity Framework

Benchmark refresh schedule

Benchmarks degrade as the industry shifts. Refresh every 6 months:

• Q1 and Q3 each year: check the Validity benchmark report, Cognism State of Cold Calling, Cleanlist provider rankings, DealSignal annual report
• Update resources/benchmarks.md with new figures
• Update resources/providers.md pricing and feature set
• Note changes in the version history below

Version history

V1. June 2026

• Initial release
• 10-dimension framework, CRM-agnostic
• Light CRM routing for 5 platforms (HubSpot, Salesforce, Pipedrive, Close, Attio)
• Provider catalog Q2 2026
• Lead magnet Notion page live at https://app.notion.com/p/38499b4b2619817f83abeed4b3ebbc37

TODO before V1.1

• Live HubSpot portal verification of audit shortcuts (optional, lead-driven)
• Live Salesforce portal verification (optional, lead-driven)
• Add an interactive scorecard widget (Notion or standalone HTML) as a 4th output format
• Add an audit history template (track scores quarter over quarter per client)
• Web-search benchmark refresh (Q3 2026)

---

Created by Growth Today — AI-native GTM engineering firm. Maintained and updated by Brigitta Ruha. More open Claude Skills for go-to-market teams: https://www.growthtoday.co/claude-skills