Case Study: Growth Kinetics & DKSH – Turning Data into Actionable Insights

DKSH is a Swiss multinational company known for its Market Expansion Services providing essential distribution, sales, and marketing support to help other firms grow in new markets.

In practice, DKSH partners with companies (often in pharmaceuticals and consumer goods) to deliver products across diverse Southeast Asian markets.

This role demands navigating a patchwork of languages, regulations, and supply chains in countries from Thailand to Malaysia.

With a 40,000+ person workforce spread across 36 markets, DKSH’s operations are immensely complex. Ensuring consistent market penetration for clients’ products means managing everything from regulatory approvals and logistics to on-the-ground sales efforts.

In short, DKSH sits at the intersection of global producers and local Asian markets – a position that generates vast amounts of data but also significant operational complexity.

Initial Engagement

DKSH initially engaged Growth Kinetics in a modest way: the company invited Growth Kinetics’ team to deliver a talk on Artificial Intelligence and data analytics. This guest lecture was intended to educate DKSH’s leadership on emerging AI trends.

However, the presentation quickly sparked deeper interest. Growth Kinetics’ CEO, Alex Mrvaljevich, demonstrated AI examples that struck a chord with DKSH’s challenges. “We thought it was just a seminar, but it became clear DKSH had an appetite for practical AI solutions,” Mrvaljevich recalls.

The successful talk led to ongoing conversations – DKSH’s executives realised Growth Kinetics brought strong capabilities that could be applied to their business. What began as a one-off presentation evolved into a broader collaboration.

Impressed by the team’s expertise and pragmatic approach, DKSH invited Growth Kinetics to examine some of its pressing operational pain points where AI might make a difference.

Business Challenges

As a large market expansion enterprise, DKSH faced several major challenges undermining efficiency and insight:

Fragmented Organisation & Data Silos: DKSH’s regional divisions and business units (from Healthcare to Consumer Goods) operated on different systems. Information was siloed – sales data, supply chain metrics, and HR records lived in disparate databases.

This fragmentation made it difficult to unify business intelligence or get a single source of truth for decision-making. C-level leaders struggled to glean cohesive insights across countries and divisions, hampering strategic planning.

Legacy Processes & Inefficiencies: Many internal processes had not kept pace with DKSH’s growth. Reporting and analysis often involved manual work (like exporting spreadsheets from multiple systems).

Such labour-intensive processes were slow and prone to error, delaying critical insights. In the fast-paced markets of Southeast Asia, these delays translated into missed opportunities and reactive (rather than proactive) management.

Massive Recruitment Workloads: A particularly acute problem was in talent acquisition. With thousands of roles ranging from sales representatives and supply chain specialists to pharmacists and regulatory experts, DKSH’s HR department was inundated with massive volumes of CVs.

Each market posted jobs in multiple languages, attracting candidates of widely varying backgrounds.

Manually sifting through résumés to find the right talent was time-consuming and inconsistent. Recruiters had to read each CV, interpret qualifications against job requirements, and do this across a spectrum of job functions – an almost herculean task at the scale DKSH operates.

Together, these challenges painted a picture of an organisation with immense data resources but difficulty harnessing them. DKSH’s leaders knew that if they could unify their data and apply intelligent automation, they could unlock significant value – from efficiency gains to more strategic use of information. This realisation set the stage for engaging Growth Kinetics in a hands-on project.

Internal Efforts Before Engagement

Before partnering with Growth Kinetics, DKSH had tried to address some issues internally. In the HR realm, the talent acquisition team made valiant efforts to cope with the CV deluge using manual triage and basic tools.

They set up standardised screening checklists and keyword searches within their Applicant Tracking System to filter candidates. However, these measures had limited impact.

Human recruiters could only process so many applications per day, and important nuances often slipped through the cracks. For example, a candidate might have excellent leadership experience in a related industry but be overlooked because they lacked an exact keyword match to the job description.

The manual approach was simply not scalable: as DKSH’s hiring needs grew, bottlenecks became apparent – qualified candidates were being missed and time-to-hire was longer than desired.

DKSH’s digital transformation team had also experimented with off-the-shelf software for resume screening. They trialled a couple of Software-as-a-Service (SaaS) solutions promising AI-driven recruitment filtering. But these generic tools struggled with DKSH’s diversity of roles and languages.

The accuracy was inconsistent, and the one-size-fits-all algorithms did not adapt well to niche positions or multilingual resumes. After these internal efforts plateaued, it was clear a new approach was needed. The organisation recognised that simply adding more staff to brute-force the problem was not sustainable.

Instead, they sought a smarter way ideally an AI-driven solution that could act as a tireless, unbiased “first pass” reader of every CV, flagging the best candidates for human follow-up. This vision set the scene for Growth Kinetics’ involvement.

Growth Kinetics’ Approach

Growth Kinetics began the engagement with a discovery-led approach. Rather than rushing to deploy a pre-packaged tool, the team first spent time understanding DKSH’s data, workflows, and pain points in detail.

Mrvaljevich emphasised that context is key, he explains.

In practice, Growth Kinetics started with a data-intensive phase – gathering samples of DKSH’s resumes, job descriptions, and recruitment outcomes from past hiring cycles.

By analysing this data, they aimed to identify patterns: What skills and experiences did successful hires have? What keywords or phrases signalled a good match (and in which language)? Where were human recruiters spending the most time? A major insight from this phase was that traditional automation wouldn’t suffice.

DKSH didn’t just need a faster way to keyword-match CVs; they needed a way to truly “read” and understand each CV in context. This is where Growth Kinetics’ philosophy of using Large Language Models (LLMs) as high-value readers came into play. Conventional resume screening systems rely on rigid filters, but LLMs can interpret context and meaning in unstructured text.

In other words, the AI could be taught to evaluate a candidate much like a human would – considering the nuances of their experience, even if the wording differed from the job posting. “We saw an opportunity to use AI not just to generate content, but to digest it,” says Mrvaljevich. “LLMs allow us to read thousands of CVs in minutes, extracting insights that a recruiter might take weeks to uncover.”

This ability to “read at scale” and pick up on context meant the system could spot qualified candidates that a keyword search might miss – for example, recognising that “managed client portfolios in Ho Chi Minh City” is relevant to a sales role in Vietnam even if the CV never explicitly says “sales executive.” As a recent industry analysis notes, LLM-based parsers can go beyond keywords by analysing the meaning behind terms, offering a more holistic evaluation of each candidate.

With this vision, Growth Kinetics adopted a prototyping strategy best summarised as “make it real before making it right.” Rather than spending months in design, the team rapidly built a proof-of-concept AI model early in the project. In a matter of weeks, they had a prototype that could process DKSH’s CVs and output a ranked shortlist of candidates for a given job description.

This prototype was not perfect – its purpose was to spark feedback and refinement. It was demonstrated to DKSH’s HR managers and the Digital Transformation team in live sessions. Seeing their own data being analysed in real-time by the AI made the benefits tangible for DKSH’s stakeholders.

It also allowed those stakeholders to clarify their requirements: they could say “That ranking makes sense” or “No, in this market we’d value X skill more.” Through iterative collaboration – frequent workshops and review sessions Growth Kinetics fine-tuned the system. HR recruiters provided input on nuances of various job functions, the IT team ensured integration requirements were met, and executive sponsors aligned the project with strategic objectives.

This tight feedback loop ensured the evolving solution stayed practical and user-centric. Mrvaljevich reflects on the process: “In a complex enterprise environment, a prototype can speak louder than a 100-page proposal. By making the solution real early on, we got invaluable input that guided us to the right outcome.”

Technical Solution

The solution that emerged was a hybrid architecture combining Machine Learning and LLM components. In simple terms, Growth Kinetics engineered a system where traditional algorithms worked hand-in-hand with advanced language models.

The ML components handled structured data and specific business rules (for example, ensuring certain mandatory qualifications or certifications were noted), while the LLM dealt with unstructured text reading the free-form content of CVs and job descriptions to gauge candidate suitability.

This hybrid approach provided the best of both worlds: the consistency and speed of machine learning plus the contextual understanding of a human-like reader through the LLM. Crucially, the entire system was designed to plug seamlessly into DKSH’s existing IT ecosystem. DKSH’s recruitment platform was built on Salesforce, and Growth Kinetics leveraged this for integration.

When a new CV was submitted or a job posting created in Salesforce, the data would be securely sent to a Growth Kinetics-hosted API endpoint. Behind the scenes, DKSH’s data was transmitted over an encrypted channel to Growth Kinetics’ cloud environment, where the AI models did their work.

The analysis process went something like this: the LLM would parse the candidate’s resume and the job description, extract key features (skills, years of experience, education, etc.), and form a semantic understanding of how well the candidate matched the role. Meanwhile, the ML components might add quantitative scores (e.g. a score for experience length, a score for skill overlap).

The system would then combine these signals into an overall match score and generate a brief summary of the candidate’s strengths and any potential gaps relative to the job requirements. The results of this analysis were then pushed back to DKSH’s Salesforce via a webhook. From the end-user perspective – the DKSH recruiter – it all happened in a matter of seconds after they uploaded a CV.

They could remain in the familiar Salesforce interface and see an AI-generated “candidate insight”: a match score, highlighted keywords or phrases showing why the candidate was recommended, and even a short textual summary (generated by the LLM) explaining the candidate’s fit.

This tight integration meant minimal change management for DKSH’s staff. As one HR manager put it during testing,

Beyond its intelligence, the solution was built with enterprise-class robustness in mind.

Because the heavy computation runs on Growth Kinetics’ cloud, DKSH did not need to maintain new hardware or complex software – infrastructure overhead was kept low. The system scaled on-demand: whether it was 10 CVs or 1,000 CVs in a day, the API could handle the load by elastically scaling resources in the cloud.

Accuracy of the model remained high because Growth Kinetics continually refined the LLM using DKSH’s own data (with proper data privacy safeguards). Importantly, the design was generalisable across job functions.

The same architecture could assess a medical sales representative or a supply chain analyst role, without needing a completely new model for each – the LLM’s broad knowledge and the custom DKSH data made it flexible. And whenever DKSH introduced a new category of job, the model’s training data could be updated to include a few examples, allowing it to adapt quickly.

Compliance and security were non-negotiable aspects of the technical solution. CVs contain personal data, and DKSH, dealing with healthcare, also had to consider confidentiality and data regulations in each country.

Growth Kinetics worked closely with DKSH’s IT security team to ensure the solution met all corporate and regional security requirements. Data was encrypted end-to-end, no resumes were stored longer than necessary, and the AI models were hosted in a secured cloud environment with access control.

By designing the API as an external service, an extra layer of isolation was provided – DKSH’s core systems never had to directly expose data beyond the agreed interface, and they could monitor and log every transaction for compliance. In short, the technical solution delivered a scalable, accurate, and secure AI service that embedded seamlessly into DKSH’s workflows.

Results and Impact

Once deployed, the AI-driven screening solution made an immediate impact on DKSH’s talent operations. Recruiters found that what used to take hours now took only moments. On average, the system could process several thousand CVs per month – essentially every application that came in – ensuring no candidate was overlooked.

This was a dramatic change: previously, HR staff could only thoroughly review a fraction of applications and often had to rely on guesswork or first-impressions to prioritise.

Now, every CV got attention, with the AI highlighting those worth a closer look. As a result, DKSH reported that recruiters were saving significant time (hours each week) and could reinvest that time into engaging with top candidates in person, rather than wading through paperwork. The quality of hiring decisions also improved.

To validate the model, DKSH’s HR leadership conducted a blind study. They took a set of real job openings and candidate CVs and had three separate evaluations done: one by their experienced recruiters, one by a leading third-party resume screening software, and one by Growth Kinetics’ AI (without knowing which was which).

The outcomes were then compared against the candidates who actually performed well or were hired in those roles. The result was eye-opening – Growth Kinetics’ model outperformed both the human-only process and the off-the-shelf software in identifying high-potential candidates.

In particular, the AI surfaced several candidates who the recruiters initially ranked lower but turned out to be successful hires, and it correctly filtered out some applicants who looked good on paper to humans but lacked substantive relevance. “When we saw the model consistently pinpoint great candidates our teams had missed, it was a proud moment,” Mrvaljevich noted.

This evidence helped win over any remaining sceptics. DKSH’s HR team grew confident in the AI as a trusted second pair of eyes, rather than a black box. Notably, the AI’s recommendations were accompanied by explanations (the highlighted skills/experience matches), which provided transparency and helped HR understand the rationale – a critical factor for user acceptance.

For DKSH’s Digital Transformation department, the project became a flagship success. It demonstrated that adopting a bespoke AI solution could yield superior results to generic approaches.

The win had multiple dimensions such as a clear efficiency gain, better talent outcomes, and a positive story of man and machine working together. This boosted the Digital team’s credibility internally, showing that their investments in innovative technology could pay off in strategic areas of the business.

One executive remarked that the project “turned a necessary but tedious process into a competitive advantage” – now DKSH can staff up rapidly and smartly as it expands in Asia, armed with data-driven hiring insights. Given the success in the initial deployment (which was focused on a few key markets and job functions), DKSH’s leadership is now considering a broader regional rollout.

Plans are underway to extend the AI screening system to all business units and countries where DKSH operates, effectively making it a standard tool in the recruitment process company-wide. Moreover, the collaboration with Growth Kinetics has sparked ideas for other applications.

Seeing how LLMs can distill actionable insight from text data, DKSH is exploring whether similar AI approaches could help analyse market research reports, sales call transcripts, or supply chain logs to uncover trends.

In other words, this project not only solved the immediate CV screening problem but also catalysed a mindset change – DKSH is embracing the notion of turning its abundant data into actionable intelligence. Strategically, that positions them to be more proactive and agile in market expansion, which is exactly the outcome a C-level audience cares about.

Key Takeaways

This case study offers several key insights for enterprise leaders managing large data environments.

Bespoke Solutions Outperform Off-the-Shelf: DKSH’s experience showed that a tailored AI solution, crafted for the organisation’s specific context and data, delivered far better results than generic third-party tools.

In complex environments (multiple countries, languages, and business lines), one-size-fits-all software often falls short. Investing in a bespoke approach ensured the nuances of DKSH’s business were captured, yielding higher accuracy and a competitive edge.

LLMs as High-Value Readers (Not Just Writers): A common misconception is that Large Language Models are only useful for generating text or chatbots. In reality, their ability to read and understand vast amounts of unstructured data may be even more powerful.

Growth Kinetics leveraged LLMs as tireless readers that could comprehend resumes and documents at scale, extracting insights that humans might miss.

This case highlights how AI can augment human decision-making by sifting through information and presenting the right insights, rather than simply churning out content.

Scalable, Flexible AI for Enterprise Data: Growth Kinetics delivered a solution that was both scalable (able to handle DKSH’s high volumes) and flexible (adapting to different roles, markets and evolving criteria).

The architecture’s integration into existing systems (like Salesforce) exemplifies how AI can be woven into enterprise workflows with minimal disruption. It’s a reminder that AI projects should not be tech experiments in isolation, but practical tools aligned to business processes and security standards.

When done right, as with DKSH, an AI initiative can start in one domain (HR) and scale up to drive transformation across the organisation.