Professional systems
Two tools from my work at LLYC, rebuilt independently with synthetic data.
Independent reconstructionSynthetic data
C2
Reporting close automation
A campaign close that used to be done by hand, now mostly run from a spreadsheet.
A flow in Google Sheets, co-developed with a colleague from Data Analytics, that automates part of the reporting close of a campaign. A person still reviews the result before it goes out.
- Inputs
- Preparation
- Charts
- Human review
Approximate
- Date
- August 2026
- Scope
- Share of the process, with human review.
- Source
- Professional estimate
- Date
- August 2026
- Scope
- Estimate per close.
- Source
- Professional estimate
Run a synthetic close
Synthetic dataChoose how to run it and press the button. The human review never disappears.
This demo needs JavaScript. The case below reads the same without it.
Problem
Each close meant refreshing sources, rebuilding comparisons and KPIs, and drawing the same set of charts again. Repetitive work with many places to make a mistake.
Decisions
Keep it where the team already works
BusinessThe close runs from a spreadsheet the team knows, in clear steps, so nobody depends on one person to operate it.
TechnicalGoogle Sheets with Apps Script. Data comes from BigQuery and Supermetrics, with a few auxiliary sources.
Rules for insights, a person for judgement
BusinessThe flow prepares the insights, and a person decides what is worth saying to the client.
TechnicalInsights are prepared with deterministic rules. Human review is a fixed step, not an option.
How I checked it
Used in real closes and compared with how long the manual process took.
What I learned
Automating most of a process is more useful than chasing all of it. The part left to a person is where the judgement is.
Role
Co-development, with functional and technical responsibility shared with a colleague, plus use and validation in real closes.
Status
In real professional use and under continuous improvement.
C3
Path-to-Conversion Analyzer
A tool that turns awkward path exports into an analysis you can read in minutes.
A browser tool to analyse path-to-conversion exports from Campaign Manager 360. It detects the format, normalises it and shows the paths that matter.
- Upload
- Detect and validate
- Normalise
- Filter and read
- Date
- August 2026
- Scope
- In the cases tested.
- Source
- Professional estimate
Analyse synthetic conversion paths
Synthetic dataPick an export format, normalise it and filter the paths.
This demo needs JavaScript. The case below reads the same without it.
Problem
The exports arrive in formats that are hard to work with, and analysing them by hand took several hours each time.
Decisions
Accept the export as it comes
BusinessThe analyst uploads the file as exported. No manual reshaping before starting.
TechnicalCSV and XLSX upload with SheetJS, detection and validation of the structure, and normalisation of pivoted and unpivoted layouts into one table.
Answer the usual questions first
BusinessFilters, KPIs, most frequent paths and path lengths are on screen at once, and the result can be exported as a presentation.
TechnicalChart.js for the charts and PptxGenJS for the export. Everything runs in the browser.
How I checked it
Results were compared with analyses done by hand on the same exports. In the cases tested, the work went from several hours to about ten minutes.
What I learned
Most of the time was going into reshaping data, not into thinking about it. Fixing the format problem first made the analysis itself short.
Role
Functional design, adaptation to real use cases and formats, validation and evolution. It grew out of earlier exploratory work with a Gemini Gem, and the implementation was assisted by Claude.
Status
A working prototype with its first real applications.