Case studies

    What the work actually looks like

    Real engagements, shown the same way each time: the starting problem, what we found, what we changed, and what happened as a result. Where a figure is quoted, the methodology behind it is shown alongside it.

    Case study 1#caspian-one

    Caspian One: Vincere data enrichment and deduplication

    Client context

    Caspian One, a recruitment specialist working in financial services, FinTech and investment banking.

    Starting problem

    Caspian One held years of accumulated CRM data in Vincere that needed cleaning up to become a genuine business asset. Thousands of records had incomplete information, duplicate entries, invalid contact details and outdated employment data, limiting the sales team's ability to target effectively and holding back marketing ROI.

    Tech / CRM / AI stack involved

    Vincere (CRM platform), Clay, and large language models including the ChatGPT API and Claude, used for AI-driven data validation, classification and enrichment.

    Diagnostic process

    The engagement began with a full data audit across Vincere's Companies, Contacts and Candidates records, working with Caspian One's project lead to establish priorities and clear success criteria before any changes were made.

    Interventions made

    • Deployed Clay alongside LLMs (ChatGPT API, Claude) to validate, classify and filter data with AI reasoning rather than static rules
    • Enriched missing website details, LinkedIn URLs, phone numbers, mobile contacts, email addresses and industry classifications
    • Applied AI-driven deduplication, including name-matching logic to catch duplicates that traditional field-matching would miss, and to preserve relationship history for contacts who had changed employer
    • Used AI-powered quality control to surface high-confidence enriched data and filter out low-confidence flags, while identifying systemic patterns behind data quality issues

    Before/after results

    Figures below reflect internal measurements taken by Syun Consulting before and after the engagement, based on audit data within Caspian One's Vincere platform, not an independently verified external metric.

    • Contact database: Meaningful increase in functional expertise and industry tags, email bounce rate reduced to negligible levels, LinkedIn profile coverage expanded to close to half the database, a substantial number of mobile numbers added, and around a quarter of previously unreachable candidates given valid personal emails.
    • Company database: Thousands of high-quality company records retained after enrichment, close to a quarter of duplicate companies identified and consolidated, and comprehensive website and LinkedIn coverage for precise targeting.
    • Candidate database: Around half of all candidates enriched with current LinkedIn details, a double-digit percentage increase in functional expertise tags, and a strong LinkedIn enrichment rate delivering current employment intelligence.

    Challenges or trade-offs faced

    The primary challenge was scale. The volume of records across Companies, Contacts and Candidates in Caspian One's Vincere database meant enrichment and deduplication had to hold up consistently and accurately across a large dataset, not just a small, easily managed one.

    Client feedback

    “Our database is now in a considerably better state which has led to greater company-wide trust in our CRM. We're no longer held back by unusable data, with our contacts and candidates enriched beyond what we believed possible. This data is already being leveraged to bring in new business, make more deals, and reduce our dependence on third-party tools. The project delivery was excellent. The team demonstrated a clear understanding of our objectives from the beginning. Everything proceeded smoothly, they were flexible with respect to our other initiatives, and all work was completed on schedule. It was a pleasure collaborating with Syun Consulting.”