Gaurav Kumar Singh
About

Digital marketing taught me the business. Then I went and built the systems.

Gaurav Kumar Singh working at a desk

My career started in digital acquisition — SEO, performance marketing, demand generation. Over ten years it moved steadily down the stack: into the analytics that measure it, the CRM that manages what it produces, the automation that removes the manual handling, the platforms it all runs on, and now the AI applications on top.

That path was not a plan. It was the consequence of repeatedly hitting the same wall. You cannot fix lead quality if the attribution is wrong. You cannot fix attribution if the measurement layer was never designed. You cannot fix the measurement layer if the website was built as a brochure. Each problem sat one level below the last, and solving it meant learning that level properly.

Today I lead digital transformation initiatives across four B2B industrial businesses — Bio & Healthcare, Materials & Semiconductor, Energy & Environment, and Mobility. In practice that means the corporate web platform, a centralised lead management workflow spanning Power Automate, SharePoint and Salesforce, the GA4 and Looker Studio measurement architecture, employee experience digitisation on Microsoft 365, and a first AI-enabled customer-facing application.

What I am, and what I am not

I am not a data scientist and I do not claim to be. I have not trained a model and I would not position myself against someone who has.

What I do is sit between business leadership and technical teams. Define the problem precisely. Design the solution. Choose the technology on fit and total cost rather than novelty. Drive it through delivery and the stakeholders who have to adopt it. Then instrument it, so the business can see what actually changed.

Increasingly that includes building directly — prototyping, specifying and shipping with AI coding agents under version control. Not because I am a software engineer, but because the distance between “we should try this” and working, governed software has become short enough that one person who understands the business problem can now cross it. Knowing where that line sits, and where it does not, is a large part of the job now.

How I work

Use what you own

The question is rarely “what should we buy?” It is “what does the organisation already have a licence for, and what could it do if someone designed a proper experience on top of it?”

Vocabulary before automation

Automating an ambiguous process just produces ambiguity faster, with a dashboard on top lending it false precision. Agreement is harder than the technology and matters more.

Measure the before-state

The baseline is free exactly once, before you start. After that it is gone permanently — and “it improved” is a much weaker claim than a number.

Design the failure mode first

Especially with AI. Decide what the system is structurally incapable of doing before deciding what it should do. That is what gets it approved.

Career

Education

Tools I work with

Analytics

Google Analytics 4 · Google Tag Manager · Looker Studio · Search Console · Conversions API · Attribution modelling

CRM & MarTech

Salesforce · HubSpot · Mailchimp · Lead lifecycle design · Marketing automation

Automation

Microsoft Power Automate · Microsoft 365 · SharePoint · Microsoft Forms · Workflow design

Web & product

React · Next.js · Tailwind · HTML/CSS · PHP platforms · Headless CMS · Git · Technical SEO

AI

Claude · ChatGPT · Codex · AI coding agents · Retrieval-augmented assistants · Evaluation & guardrail design