Marina Havlis

Data & Analytics Consultant

AI Operator

Product & Project Manager

I build tools and automations, run prescriptive analytics, and coach junior analysts. Over a decade across e-commerce, med-tech and FMCG.

AI is omnipresent and cherry picking the right bits to integrate into your work takes judgment. I test new features weekly to keep my setup sharp.

Project management, scrum, product ownership, development, UX/UI, QA: on tiny projects I cover all the ground, on larger ones I coordinate and collaborate.

Open to short- and medium-term projects. Prague and remote.

Bring me in for

I follow a problem from the first number to what to do about it. The deliverable is a dashboard when it answers the question, otherwise the action itself: a decision made, an email sent, an Asana task filed, a Salesforce case assigned, a rep training deck delivered.

I ran agents in Cursor via the Keboola CLI before MCPs got the spotlight, then moved to Claude Code for knowledge work. My setup scans, triages and reports, mostly on demand rather than always-on. Tokens cost money, so I keep only the processes that earn their keep.

Give me a few weeks on the vocabulary and I'm productive. I've shipped across most business functions and multiple industries. Underneath it's much the same work whatever the domain: break big things into digestible pieces, keep people in the loop, stay focused till completion.

How I work

Nothing revolutionary, just the order I tend to go in and the bit that's stuck around through whichever tool's in fashion. Hover or tap a step for the detail.

01
Signal
Something looks off in the numbers. I come up with a hypothesis before I run any queries, then treat it as wrong until the numbers back it.
02
People
I find whoever actually owns the process and ask how it really works. Most of the answer is in their heads anyway. The warehouse only gets you halfway.
03
Map
I trace the process end to end and map how it all fits together.
04
Diagnosis
I work out what is missing or wrong and what it leads to downstream.
05
Fix
I split the fix into digestible tasks and assign them to the owners. More and more I just build the tools that do those tasks, with the owners still in the loop.
06
Measure
I build the before-and-after so we know if it worked. With AI generating fixes fast, this is the part I spend the most time on now.

Use cases

A few things I've uncovered.

A few things I've built.

A few things I've delivered.

Regional supply optimization

A chain of Claude Code routines that scans a region's listings, finds the ones bleeding margin or under-converting, hands a fix to the account owner or content team and checks the result once the change is live. Two of the first five deals I piloted signed within three weeks.

how it ran

My first real category-management project, built end to end. I chose the relevant KPIs, then ran the pilot region's numbers. Each underperforming listing got diagnosed against its peers, and each fix went to the account owner as something they could take to the partner. The under-converters were sent for the content team to work on.

The change audit closes the loop: it waits for clean before-and-after data, then tells me whether the change stuck. A 30-day read on the margin was built into the design.

Merchandising process evaluation

Mapped how merchandising and impressions actually get decided (and where the gaps were) by talking to the people who run them.

how it ran

The signal showed up in the refund and impression data, but how it all fit together was in people's heads, so I spent days talking to the market managers and merchandising owners and drew the process out. Refunds weren't yet a factor in deciding which listings deserved a spotlight. I designed a fix so high-refund listings stop getting promoted.

Refund-driver deep dive

Refund growth looked like a volume problem until I split it by reason. The biggest piece was availability: people who couldn't get a booking.

how it ran

Refund rate on its own says nothing, so I split refunds by root cause. The fastest-growing slice was availability: people refunding because they couldn't get a booking. That's what pointed at booking tools, and turned into the booking-related work under Product & Project.

AI setup for the role

A setup that runs the recurring category-manager work for me: a set of specialist routines, scheduled or fired on demand, with results pushed to chat or Asana. Built solo, meant to be copyable by someone who doesn't code.

what's in it

What I kept on schedule is the daily email-and-chat digest with a voiceover I listen to on the move, and the weekly business review that goes to the whole vertical. The rest either became plain Python or runs on demand to save tokens. A quick demo and a running check on how mature it is are written up here: Video recap · AI Maturity Check.

Bulk Salesforce case filer

A Python script that files Salesforce improvement cases in bulk, so an analysis turns into things reps can action.

Weekly business review automation

A routine that drafts the weekly review, pacing, promos, refunds, inventory, and drops it in chat before the Monday meeting.

A mix of reporting solutions

Led a team of developers delivering reporting solutions for finance, sales ops and product.

how it ran

I owned the backlog and priorities, hosted the daily stand-ups as scrum master, shipped updates and managed stakeholders till the final task's sign-off in Asana.

Product and checkout teams' power tools

Partnered with product & checkout teams on A/B experiments and checkout-funnel analytics.

how it ran

The point of the A/B experiment evaluation dashboard was making test results readable for PMs and POs who aren't statisticians, without losing the rigor: a clear "this variant is ahead" read instead of a p-value nobody trusts. It's also where I learned how the product side of the platform works.

Checkout funnel dashboard was my step into the payments and fraud functions. I led the build of a daily funnel tracker plus an alert that caught a broken country-and-platform combination before anyone filed a ticket. I did the front end and ran the project, coaching a backend analyst who was building the data layer.

Booking-platform analysis

Built the case for which booking platforms to partner on, and handed Product a pilot recommendation.

how it ran

I scanned the listing inventory and compared performance by booking tool, which gave me a shortlist of the strongest candidates. Then I scraped the web to check which of our partners already use them, so the pilot recommendation came with real partners attached. The account owners got Salesforce cases to start the partner conversations.

Experience

Where I've built and led the analysis.

AI's only been usable for me since early 2025. The urge to automate the boring parts goes back further.

I've mostly worked in projects, hopping teams and domains on purpose to keep things interesting.

2023–PRESENT

Independent Data & Analytics Consultant

US E-com Platform

Came in as product owner and scrum master, project-managing a team of developers and owning the data products for finance, product and sales ops: dashboards, datasets, pipelines. Then pivoted to advanced concierge-style analytics for a VP of Growth who, on taking charge of the beauty and wellness vertical, tasked me with managing revenue performance. Later took on more business-focused work: category management of massage and fitness.

Ran the analysis on supply churn and retention, the conversion funnel, revenue, refunds, supply gaps and price elasticity, partnering with sales, sales ops, marketing, market managers, product and finance up to C-suite. Mentored junior analysts supporting the work.

Built the AI setup for the role: it monitors performance, prescribes actions for sales reps, then checks whether they worked. Started with Cursor in early 2025, moved to Claude Code.

Owned the backlog, sprint planning, ceremonies, release and QA for the data products, between Commercial Finance and Revenue Management teams and IT. Then took on one new area after another: checkout, payments, fraud, product, sales ops, category management.

UK Startup

Built an e-commerce, omnichannel marketing data model and a set of pilot dashboards for a startup weighing a subscription model.

Modeled the data from scratch and built the dashboards: sales by channel, customer segmentation, acquisition, retention, engagement. Designed the app activation funnel: footfall, impressions, gamification, downloads, sign-ups, purchases.

Generated the test data and the data flows where the real data wasn't there yet, so the dashboards could be built ahead of go-live.

A self-contained engagement: scope it, model it, ship the dashboards, hand over.

2022–2023

Senior Data & Analytics Consultant

Merkle · Salesforce Implementation Partner

Led a team of two developers on a Salesforce data migration for a luxury brand. Served customers of Salesforce Data Cloud and CRM Analytics.

Built the data model, segments and calculated insights in Data Cloud, and CRM Analytics recipes and dashboards for validation.

Built reusable CRM-Analytics audit recipes so the same checks ran the same way every time.

Switched from building to overseeing the work as project manager for data migration: responsible for budget, time estimates, backlog management and prioritization.

2013–2022

Analyst → Senior → Principal Analyst

Medtronic / Covidien · Med-tech

Grew from analyst to principal on the individual-contributor track, and stood in for team leads and managers when they were out.

Built the back and front end of reporting solutions for senior stakeholders across customer service, sales, finance, commercial ops, compensation and master data. Ran Tableau and new-hire training sessions.

Automated the data flows so the dashboards refreshed themselves without me rebuilding them every month.

Kept moving across functions, solving reporting problems across customer service, sales, finance, commercial ops, compensation and master data.

2006–2010

Credit Controller → Financial Analyst → Project Manager

Procter & Gamble · FMCG

Credit control, logistics and trade-terms budgeting, and a migration of logistics operations from Poland to Czechia. Where the finance fluency and the FMCG background come from.

2005–2006

Junior Tax Advisor

Ernst & Young · Consulting

Expat payroll and income-tax compliance for EY's clients in Almaty.

Skills

Tools
SQLBigQuerySnowflakeTeradataPythonHTMLTableauKeboolaAlteryxClaude CodeCursorNotebookLMMCPSalesforceAsanaJiraExcel / Sheets
Analytics
Data analysisData visualizationKPI / metric definitionCohort / funnel analysisA/B testingData modeling
Data & delivery
ETL & data pipelinesData governance & qualityProcess automationStakeholder managementAgile / ScrumProduct owner / PMCoaching & team leadership
Domains
ConsultingMed-techE-commerceMarketplacesSubscriptionsBeauty & wellnessFMCGLuxury goods

Education, languages, certifications

Education
Ing. (MSc equivalent), International Business
University of Economics, Prague
Languages
English C1 · Czech C1 · Russian C2 · Spanish B1 · French A2
Certifications
Salesforce Data Cloud Accredited Professional
Salesforce Tableau CRM & Einstein Discovery Consultant
Operational Excellence, Covidien (Lean Daily Management)

In others' words

"Marina is a Godsend. Her unwavering curiosity, thorough discipline and superb technical skills, married with one of the best moral compasses I've ever seen, made her the right partner in crime for the task we had at hand. A stagnant, bureaucratic, slightly messy marketplace in search of steer, determined action and data driven growth is no minor arena to jump into. She always went headfirst, no task was too big or too small, no stakeholder was far away for her to reach, convince and convert.
If you really mean business, you need her by your side."
VP & GM
"I have been struggling pursuing Tableau-related subjects, including projects and no one ever gave me a clear presentation or hints like you did yesterday - it sounded simple and genuine, instead of 'miraculous, commercial and ambiguous' like the previous experiences I have had. The visuals you presented were a lot better looking than what I have seen demonstrated to many customers."
Andre, Data Engineer, Merkle (training participant)
"Thanks for your detailed, insightful and creative work modeling out changes. I initially shared a very broad policy question, and you were able to gather, synthesize, and analyze data making recommendations which will have a significant positive impact."
David Greene, Senior Director, Commercial Operations & Customer Care, Medtronic

Get in touch

Drop me a note.

Marina Havlis

Independent consultant, open to short- and medium-term projects.

CET time zone, working with US-based customers since 2013.

Easiest way to reach me is email.

marina.l.havlis@gmail.com
linkedin.com/in/marina-havlis
Printable resume:
marinahavlis.pages.dev/resume
Case study:
marinahavlis.pages.dev/case-study