Belkis Castillo Customer Success Manager

Customer Success Manager · Building with AI

Turning enterprise SaaS customers into long-term partners. Now building with AI.

Belkis Castillo
$1.5M–$3.8M Portfolios managed
91%+ Gross retention achieved
110%+ Renewal rates achieved

Four stories behind the numbers.

Click any story to read the full version.

When I joined as the first dedicated CSM for an emerging region, the company had customers there but no playbook for serving them. I built that playbook end to end: segmentation, onboarding cadence, QBR structure, escalation paths, stakeholder mapping templates.

Portfolio engagement on the region grew from 36 percent in August 2021 to 55 percent two months later, and reached 73 percent within a year. That meant moving the customer base from infrequent ad-hoc check-ins to monthly strategic conversations across the entire portfolio.

The accounts I managed included a major regional telecom operator, a regional grocery delivery platform, a major media company, a regional transport authority, and several financial institutions. Each market had its own cultural and operational rhythm, and the playbook had to flex for that.

What I take from it: the difference between a region that exists on a map and a region that exists as a real CS practice is somebody willing to build the unglamorous infrastructure before the wins show up.

I was named the first internal coordinator of EMEA product certifications. The role started simple: track who was completing the five mandatory technical certifications that helped CSMs have sharper conversations with customers.

What happened next was unexpected. Once completion rates were visible, CSMs started volunteering as proctors. Others designed new certifications. Managers began competing on team completion rates. The mechanic shifted from compliance to engagement.

Twelve months later, completion rates told the story:

Cloud Platform: 98 to 100 percent. Analytics: 91 to 98 percent. Data Visualization Tool: 83 to 89 percent. Search: 74 to 81 percent.

The pattern I learned: making something visible is often a stronger intervention than adding incentives on top. People show up when they can see what good looks like.

The hackathon was the final step of a six-month technical training arc with a global security services company. Across that period I orchestrated four functions to deliver it.

Solutions Architecture covered technical readiness. Partners arranged sponsorship, catering, prizes, and partner-side technical support. Marketing handled communications and participant follow-up. The customer's own teams managed room logistics, attendance, and a session led by them showcasing their use cases.

The day produced two strategic workloads I had not previously seen: an internal-compliance repository and a billing operations platform. It also opened a second cadence call with the customer's Architecture Operations team, breaking a previously single-threaded relationship. They went on to give us direct access to their team leads.

Hackathons sound like marketing. Run well, they are a way to turn passive customers into champions and surface workloads that would never come up in a standard QBR.

Early in that tenure I inherited a major regional telecom operator on a five-year on-premise contract. When I took the account, only one internal stakeholder had real context on it. The customer was running their workloads entirely on legacy infrastructure, their cloud entitlements sat unused, and a significant portion of the Professional Services hours in their contract had never been consumed.

I partnered with a Professional Services consultant to map the stakeholder landscape inside the customer and inventory their active projects on the legacy stack. From there I built a two-year churn risk assessment and flagged it internally.

The diagnosis was the easy part. The hard part was getting buy-in. Several people felt that flagging risk two years out was premature. I persisted, documented everything, mapped the customer's internal decision-makers, and built a mitigation plan around surfacing the unused entitlements back to them.

When the account was reassigned, I handed over a full playbook: contacts, risk map, PS hours roadmap, and the relationship groundwork. The CSM who took over executed the plan. The renewal closed successfully.

What I take from this: churn risk in legacy accounts usually is not mysterious. It is what the account team has been too close to see. It takes someone willing to name it early to save it.

Took sole ownership voluntarily of the certification program and brought structure and focus to ICs, moderators, and managers.

Written performance review

Credentials.

Formal Education

Master of Business Administration

London Metropolitan University

Bachelor of International Relations

Universidad Central de Venezuela

Courses & Certifications

Certified Customer Success Manager (CCSM) Level 1

SuccessCOACHING · CPD Certified

Domina la IA

Universidad Central de Venezuela · Ernesto Fuenmayor

Database Design with SQL Server

City, St George's University of London

What I'm building.

I learn AI by building things that serve a real use case. Here is what is currently in my workflow.

Built in my own time, for friends

Built an invoice and gift card automation for a friend who freelances, using Claude. Not paid work, just something I built in my own time. Also built WhatsApp and Instagram auto-reply workflows for e-commerce, using ManyChat and Claude. None of this was requested at work.

What's next

Currently learning to build more agents and workflows aimed at small businesses and freelancers, the segment with the least access to this kind of tooling and the most to gain from it.

Tools I work with

Claude, Gemini, ManyChat, Gamma, and Manus.

AI and small business impact

Studying how AI tools reshape operations, customer experience, and growth for small and medium-sized businesses. The interesting question is not whether AI helps — it is which interventions compound over time.

Say hello.