Most companies don’t lack data. They lack numbers that agree.
I build the data foundations, signals and reporting that B2B marketing and sales teams use to make decisions. I turn what the data says into a story people can act on.
A variety of tools that disagree, one source of truth, four groups of people who decide. The rest of this page is how that happens.
One trusted source of truth: built the marketing and business data warehouse (BigQuery)
AI and LLM agents in real workflows
Automating manual or repetitive work with workflows and integrations
Turning vague stakeholder asks into scoped, decision-ready analysis
Brought intent data and champion job-change alerts for sales outreach
Owned HubSpot administration, workflows and data standards
Set pipeline attribution and measurement of what marketing actually caused
Dashboards and recurring reporting that leadership actually acts on
2022
2015 – 2022
Klear (acquired by Meltwater)
Marketing Manager → Business Analyst
Jun 2015 – Aug 2022 · 7 yrs 2 mos
↑ Business Analyst · 2017
Analyzed product and business performance
Built recurring dashboards
Ran CRM and lifecycle campaigns
02
Selected stories
CV bullets, told as what was wrong, what I did and what changed.
01 · Data foundations
Many tools, one answer
Skills: CRM data models & governance · SQL & analytical problem-solving · BI & dashboard development · Connecting marketing, sales & revenue
Situation
Marketing and sales data lived in separate tools that didn’t tell what happened and why.
Why it was hard
Each tool had its own idea of what a lead, an opportunity or an account was. [Add what made this hard: volume, politics, definitions, legacy data…]
Before: each team reported from whichever tools it used, so the same question got different answers.After: every tool feeds one warehouse, and every team reads from it.
What I did
I built the company’s marketing and business data warehouse from scratch in BigQuery.
What changed
One trusted source of truth.
The skill underneath
Designing data models that make a company’s numbers agree.
02 · Reporting & story
Two weeks back, every quarter
Skills: SQL & analytical problem-solving · BI & dashboard development · Data visualization & storytelling · Data into decisions for non-technical people
~2 weeksof manual work, every quarter
Before
Pulling numbers
Drafting slides
Cutting
After
Generated
Shaping the quarter’s story
12345678910
Ten working days per quarter. Illustrative:how the days split between steps, and how long generation takes, are guesses.
Situation
The quarterly board deck took about two weeks of manual pulling, drafting and cutting.
What I did
I automated slide generation straight from the database.
What changed
The team now spends that time shaping the quarter’s story instead.
The skill underneath
Automating the routine so people can do the thinking.
03 · Measurement
The better question about renewals
Skills: Marketing & business analytics · Attribution & measurement · SQL & analytical problem-solving · BI & dashboard development · Data into decisions for non-technical people
Situation
The renewal report compared customers marketing had touched with those it hadn’t. That mostly showed that engaged customers renew anyway.
Why it was hard
A real effect and a selection effect look the same in a simple comparison.
What I did
I split the analysis by when marketing touched each customer.
Renewal-rate lift over customers marketing never touched, in percentage points. Illustrative numbers: they show the shape of the finding, not real results. Asked this way, marketing looks effective everywhere.Split by timing, almost all of the lift comes from the final year.
What changed
Only touchpoints in the year before renewal made a real difference. That finding became a live report for the customer marketing team, segmented by ARR, location and industry, which they now use to plan events and flag accounts at risk of not renewing.
The skill underneath
Separating correlation from impact, then making the answer usable.
04 · Reporting & story
A weekly review nobody has to compile
Skills: Marketing & business analytics · Connecting marketing, sales & revenue · Data into decisions for non-technical people · Data visualization & storytelling
BigQuery
Salesforce
HubSpot
Google Analytics
Search Console
AI agent
To C-level, VPs
EveryMonday
Weekly business review
Pipeline this week
Website & search
What changed, and why
Worth a look
A mock of the email's shape. Sections and send day are placeholders.
Situation
Leadership needed a weekly view of the business without anyone compiling it by hand.
What I did
I built an AI agent that pulls from BigQuery, Salesforce, HubSpot, Google Analytics and Search Console.
What changed
It sends a weekly review to the C-level and VPs.
The skill underneath
Using new tools to put the right numbers in front of the right people.
05 · Go-to-market signals
Who is ready to talk?
Skills: CRM data models & governance · Connecting marketing, sales & revenue · Marketing & business analytics
Situation
Reps targeting Enterprise accounts had no way to know who was in market, and past champions were walking into new companies unnoticed.
14 of 96 accounts worth a call this week. Illustrative accounts: the counts are made up.
What I did
I streamed intent and buying-stage data into the CRM with rep alerts, and set up job-change tracking for contacts from churned and closed-lost accounts, flagged in the CRM as warm outreach targets.
What changed
Outreach started with accounts already showing interest.
The skill underneath
Turning third-party signals into a short, actionable list for sales.
06 · Reporting & story
Three rhythms, three audiences
Skills: BI & dashboard development · Data visualization & storytelling · Data into decisions for non-technical people
One year of reporting. Each mark is one report; the larger the mark, the deeper the report.
Situation
The board, management and campaign teams each needed different numbers, at different speeds.
What I did
I ran three reporting rhythms: quarterly board reporting, monthly management reporting, and weekly campaign and website dashboards.
What changed
Each stakeholder got the right numbers at the right time, and acted on them.
The skill underneath
Matching the depth and timing of reporting to its audience.
07 · Alignment
One set of definitions
Skills: CRM data models & governance · Connecting marketing, sales & revenue · Data into decisions for non-technical people
Sales≠=RevOps≠=Channel≠=Marketing
I became the person they aligned around.
Situation
Sales, RevOps, Channel and Marketing often worked from different numbers.
What I did
I kept definitions consistent across teams.
The skill underneath
Being the neutral, trusted owner of the numbers.
08 · Data foundations
AI and LLM agents in real workflows
Skills: Marketing & business analytics · CRM data models & governance
Situation
AI earns its place when it runs inside a real workflow, with checks around it.
What I did
I build LLM agents into reporting and operations, with a person reviewing every output before it ships.
What changed
One example: a weekly business review that used to be assembled by hand now drafts itself for C-level and VPs.
The skill underneath
Designing agentic workflows with Codex, from data pull to reviewed output.
After every offline event, the lead list went into the CRM by hand: cleaned, checked for duplicates, uploaded, and assigned to an SDR and a sales rep. Leads waited about a day before anyone could work them.
Why it was hard
No single step was hard. Together they took a day, and every hour of it was time the leads sat untouched.
~1 day10 minfrom event list to a rep working the lead
▶ Automated workflow · Runs once triggered
01Receive the event’s lead fileBy hand▶ Trigger
02Clean and standardise the dataBy hand✓ Automatic
03Remove duplicates against the CRMBy hand✓ Automatic
04Upload the leads to the CRMBy hand✓ Automatic
05Assign an SDR and a sales repBy hand✓ Automatic
5 steps done by hand after every event, about a day from list to rep.The same 5 steps, run by one workflow: the event’s file arrives, it’s triggered once, and the rest happens on its own.
What I did
I built a workflow that does all of it. Once the event’s file arrives, one trigger cleans the list, removes duplicates, uploads the leads to the CRM and assigns each one to an SDR and a sales rep, in about 10 minutes.
What changed
Event leads now reach the right reps within 10 minutes instead of about a day, and nobody has to do it by hand.
The skill underneath
Turning a manual routine into a reliable, one-trigger workflow.
Select a skill to see every story it shows up in, or a dot for one example.
CRM data models & governance
Many tools, one answer I built the company’s marketing and business data warehouse from scratch in BigQuery.
Who is ready to talk? I streamed intent and buying-stage data into the CRM with rep alerts, and set up job-change tracking for contacts from churned and closed-lost accounts, flagged in the CRM as warm outreach targets.
From a day to ten minutes I built a workflow that does all of it. Once the event’s file arrives, one trigger cleans the list, removes duplicates, uploads the leads to the CRM and assigns each one to an SDR and a sales rep, in about 10 minutes.
SQL & analytical problem-solving
Many tools, one answer I built the company’s marketing and business data warehouse from scratch in BigQuery.
Three rhythms, three audiences I ran three reporting rhythms: quarterly board reporting, monthly management reporting, and weekly campaign and website dashboards.
Who is ready to talk? I streamed intent and buying-stage data into the CRM with rep alerts, and set up job-change tracking for contacts from churned and closed-lost accounts, flagged in the CRM as warm outreach targets.
Who is ready to talk? I streamed intent and buying-stage data into the CRM with rep alerts, and set up job-change tracking for contacts from churned and closed-lost accounts, flagged in the CRM as warm outreach targets.
From a day to ten minutes I built a workflow that does all of it. Once the event’s file arrives, one trigger cleans the list, removes duplicates, uploads the leads to the CRM and assigns each one to an SDR and a sales rep, in about 10 minutes.
Three rhythms, three audiences I ran three reporting rhythms: quarterly board reporting, monthly management reporting, and weekly campaign and website dashboards.
Three rhythms, three audiences I ran three reporting rhythms: quarterly board reporting, monthly management reporting, and weekly campaign and website dashboards.
I split the analysis by when marketing touched each customer. Only touchpoints in the year before renewal made a real difference. That finding became a live report for the customer marketing team, segmented by ARR, location and industry, which they now use to plan events and flag accounts at risk of not renewing.
Separating correlation from impact, then making the answer usable.
I built an AI agent that pulls from BigQuery, Salesforce, HubSpot, Google Analytics and Search Console. It sends a weekly review to the C-level and VPs.
Using new tools to put the right numbers in front of the right people.
I streamed intent and buying-stage data into the CRM with rep alerts, and set up job-change tracking for contacts from churned and closed-lost accounts, flagged in the CRM as warm outreach targets. Outreach started with accounts already showing interest.
Turning third-party signals into a short, actionable list for sales.
I ran three reporting rhythms: quarterly board reporting, monthly management reporting, and weekly campaign and website dashboards. Each stakeholder got the right numbers at the right time, and acted on them.
Matching the depth and timing of reporting to its audience.
I owned its administration, building workflows, integrations, custom objects and field standards, for clean, governed CRM data. GTM motions and reports could be trusted.
Governance as the ground everything else stands on.
I build LLM agents into reporting and operations, with a person reviewing every output before it ships. One example: a weekly business review that used to be assembled by hand now drafts itself for C-level and VPs.
Designing agentic workflows with Codex, from data pull to reviewed output.
I built a workflow that does all of it. Once the event’s file arrives, one trigger cleans the list, removes duplicates, uploads the leads to the CRM and assigns each one to an SDR and a sales rep, in about 10 minutes. Event leads now reach the right reps within 10 minutes instead of about a day, and nobody has to do it by hand.
Turning a manual routine into a reliable, one-trigger workflow.
04
The toolkit
Tools are easy to pick up; understanding the underlying business problem is what matters. Once you master the logic, picking up any tool is second nature.
01 Data warehouses
Google BigQuery
Snowflake
02 Data modelling
dbtdbt
DfGoogle Dataform
03 LLMs
GPTChatGPT
Claude
04 CRMs
SFSalesforce
HubSpot
PdPipedrive
05 Programming & scripting
Python
JavaScript
06 Automation
Zapier
n8n
Make
Google Apps Script
HubSpot Workflows
07 Enrichment & GTM orchestration
ClClay
6s6sense
CRCommon Room
OFOnFire
ZIZoomInfo
08 Web analytics
Google Analytics
Google Search Console
Google Tag Manager
09 Spreadsheets
Google Sheets
XlExcel
10 BI & visualization
DSGoogle Data Studio
TbTableau
Logos from Simple Icons, where available; monograms where a brand doesn’t publish an open logo.
05
How I think
Six steps from a question to a decision, followed through one real example: the renewal analysis.
Step 1 of 6: QuestionThe dots are a diagram, not data.
1 · Question
Start with a business question
Not “what does the data say?” but “what are we trying to decide?”
In the renewal analysis Does marketing help customers renew?
2 · Data
Find the data that can answer it
Usually it lives in three tools that don’t agree.
In the renewal analysis CRM, marketing automation and web analytics.
3 · Model
Build a model you can trust
One row per customer, one definition per metric, checked before anyone sees it.
In the renewal analysis Every account, every touchpoint, every renewal date.
4 · Insight
Find what actually makes a difference
Look for the comparison that separates cause from coincidence.
In the renewal analysis Only touches in the year before renewal moved the needle.
5 · Story
Tell it so it lands
One chart, one sentence, for the people who will act on it.
In the renewal analysis A live report the customer marketing team reads.
6 · Decision
Enable a decision
The work is done when someone does something differently.
In the renewal analysis Plan events around renewal dates, flag accounts at risk.