How to Get Your CRM Data Ready for AI
Webinar summary
In this NachoTuesday episode, Andy Karuza speaks with Matt McKinnon, Co-Founder and COO of YourICP, about the critical role of data hygiene in the era of AI. McKinnon explains that while AI offers immense scale, "bad data" only exacerbates existing problems, potentially leading to costly GDPR violations or ineffective outreach. The discussion centers on moving from a "spaghetti-on-the-wall" approach to a "closed-loop" CRM system where standardized activity data—like call dispositions and verified roles—continuously refines AI models to improve connect rates and personalization.
5 Key Takeaways
The 85% Rule for AI: You don't need "perfect" data to start using AI effectively. Once your CRM reaches about 85% accuracy on "Class A" fields—such as LinkedIn handles, verified emails, and phone numbers—it becomes highly operational for AI-driven outreach.
The High Cost of Poor Consent: Beyond bad leads, the greatest risk of poor data hygiene is legal. A single GDPR violation due to lack of proper consent data can cost a company up to 3% of its global revenue.
Close the Feedback Loop: A truly effective CRM acts as a "system of account learning." Teams must feed every outbound outcome (positive and negative replies, meeting bookings) back into the system to shorten the feedback loop and help AI learn which personas and messages convert best.
Standardization Over Complexity: To operationalize "messy" notes at scale, teams should prioritize standardizing fields and simplifying dropdowns (e.g., distinguishing between "not a fit" vs. "not the right time"). This turns unstructured language into "labels" that AI agents can actually use for routing and workflows.
Context-Driven Personalization: True personalization is not about "creepy" details like where someone lived in college; it’s about aligning your message with the specific pain points of a persona (e.g., CTO vs. Sales Manager) and finding the common thread that connects the entire buying committee.
Full video transcript
0:10
Good afternoon, everybody, and welcome to another Nacho Tuesday! Today I have Matt here with Your ICP. Without further ado, we’d love for you to introduce yourself and your company.
0:16
Yeah, absolutely. I’m Matt McKinnon, COO and co-founder of Your ICP. Think of Your ICP as the trust layer between go-to-market platforms and data sources.
When you look at go-to-market teams—whether it’s SDRs, BDRs building account lists, or marketing teams running nurture campaigns—you want to ensure two things:
The data is accurate
Read the rest of the transcript →Hide the rest of the transcript ←
You’re reaching people through their preferred channels
Our biggest value proposition is helping maximize connect rates across every engagement channel.
1:12
What are some of the most common data quality issues you see in CRMs that quietly break AI performance?
1:20
Great question. As companies scale AI, they often encounter the same issues that existed before—but amplified.
Stale data: Contacts change jobs frequently—within weeks or months
False positives: You think you’re targeting the right person, but context is missing
Missing behavioral data: You don’t know if someone is actually a decision-maker
Duplicate records: The same person gets multiple entries and ends up in conflicting campaigns
And one of the most critical issues: privacy and consent.
You need to know not just who to contact—but who not to contact.
3:06
How clean does CRM data need to be before it’s useful for AI?
3:13
It doesn’t need to be perfect.
At ~60% quality, you’ll generate a lot of noise
At 75–85% on key fields, you’re in a strong operational position
Focus on “Class A fields”:
Phone number
LinkedIn profile
Consent status
Once those are reliable, you’re in good shape.
4:26
What are the highest-impact fields teams should prioritize?
4:31
For B2B:
Email address
Phone number
Consent/opt-in status
Full name and company
Activity and engagement data
Data decays about 30% per year, so ongoing hygiene is critical.
5:18
What happens if you feed bad data into AI?
5:29
Worst-case scenario: compliance violations.
For example, violating regulations like GDPR can cost up to 3% of global revenue.
Beyond that:
Wasted outreach
Damaged brand reputation
Poor AI outputs
AI is powerful—but only if used correctly.
6:32
What does a true closed-loop CRM system look like?
6:38
It’s about continuous feedback and learning.
Capture every interaction
Feed outcomes back into the system
Update data in real time
Use insights to trigger next actions
You’re constantly testing:
Messaging
Personas
Channels
The goal is faster feedback loops and continuous improvement.
8:46
How can teams improve AI effectiveness over time?
8:57
AI has changed the buyer journey completely.
Buyers now do most research independently
Many already have a shortlist before talking to sales
So teams must:
Track engagement across all channels
Break down silos between sales and marketing
Feed activity data back into systems
The more connected your data, the smarter your AI becomes.
11:07
How can teams capture feedback without heavy resources?
11:13
Focus on standardization:
Simplify forms (1–3 fields instead of 10+)
Standardize call outcomes
Use consistent dropdowns
Connect systems via integrations
You don’t need heavy engineering—just clean, connected workflows.
13:04
Is personalization more important than ever?
13:16
Yes—but there’s a balance.
Bad personalization feels creepy (like referencing overly specific personal details).
Good personalization focuses on:
The problem being solved
The persona’s role
Their impact within the organization
Different roles need different messaging—but aligned around a common problem.
16:03
What signals are most valuable for improving AI?
16:11
Key signals include:
Connect vs. no-connect rates
Positive vs. negative replies
Meeting bookings
Disqualification reasons
Role validation
Also include offline interactions—events, meetings, conversations.
17:55
Do AI systems struggle with unstructured CRM notes?
18:07
Yes—because they’re subjective.
AI works best when you:
Convert notes into structured data
Capture tone and intent where possible
Combine human judgment with AI processing
20:03
What’s the right balance between structured and unstructured data?
20:09
Structured data: Drives decisions
Unstructured data: Provides context
If patterns emerge in unstructured notes, convert them into structured fields.
21:35
How can teams turn messy notes into usable data?
21:40
Think of it as:
Turning language into labels
Example:
“Not interested” vs. “Not right now” → different outcomes
Map these insights into structured fields to guide workflows.
22:34
Where is CRM heading in the AI era?
22:42
CRMs will become:
More structured at the core
Surrounded by flexible AI layers
Integrated with product and behavioral data
Instead of one system, you’ll have a connected data ecosystem.
24:26
Who is Your ICP best suited for?
24:33
Companies that:
Need to clean existing CRM data
Want ongoing data hygiene
Are testing new data sources
Want to scale outreach without corrupting their CRM
Think of it as both cleanup + continuous filtration.
26:20
What’s next for the company?
26:27
Focus areas include:
Identifying the true ICP based on real data
Expanding total addressable market insights
Leveraging behavioral and technographic signals
Enabling hyper-personalized engagement sequences
But at the core:
👉 People still trust people. Human connection remains essential.
29:21
Where can people learn more?
29:28
Visit youricp.com or try the self-service platform at app.youricp.com.
You can:
Upload CRM data
Run a free hygiene analysis
See what’s working and what needs fixing
30:26
Great—thanks again, Matt, for joining us today!
30:45
Thanks again, Andy. Appreciate it!
Discussed in this episode
YourICP helps you keep your contact data clean and stay that way over time.
Related NachoTuesday interviews
NachoTuesday: C-Suite Insights
Exclusive live 1:1 interviews with SaaS executives, investors, media, and founders.
Learn tips, growth hacks, and strategies from industry leaders
Hear behind-the-scenes stories you won’t get anywhere else
Engage live with our guests on LinkedIn Live, every Tuesday