Brand Networks Cross-platform Advertising Optimization
From budget settings to a traceable decision.
- Role
- Project Management Intern
- Timeline
- Oct–Dec 2024 · 8 weeks
- Shipped
- Web app · 3 core workflows
- Stack
- Next.js · PostgreSQL · Docker
I prioritized three operator jobs for the move from a spreadsheet—set budget rules, follow campaign shifts, and trace earlier actions—then aligned the scope with the client and coordinated an eight-week delivery.
- Follow the budget trailSpreadsheet settings and crowded logs made changes hard to trace.See the problem
- Organize around three jobsConfiguration, execution, and log follow an operator's next question.Explore the views
- Put charts beside recordsOperators can scan shifts, then inspect exact numbers and earlier actions.See what shipped
- Preserve the operator's logicFamiliar parameters remain while the history becomes searchable.Read the trade-off
01 · Context
The spreadsheet held the rules, but buried the trail.
Brand Networks' media team set the parameters for a cross-platform budget optimizer in Google Sheets. That sheet carried the working rules, but it did not make input fields easy to distinguish, and a busy day could add hundreds of entries to the action log. When a budget moved, operators needed to understand both the settings behind it and what had happened across channels.
The team set out to carry that operating model into a web application without losing the detail people relied on. The eight-week scope centered on clear inputs, a readable view of campaign movement, and a way to retrieve a specific past action.
02 · Feature set
Let the operator's next question shape the next view.
Interviews with Brand Networks' media team and technical stakeholders found three points of friction: the sheet did not clearly mark what an operator should enter, large histories made troubleshooting slow, and the working file was too dense to use in client conversations. I used those findings to prioritize a practical sequence with the client: set the optimization, see where budgets moved, then locate an earlier event when a result raised a question.
We gave each job its own view. Configuration exposes campaigns, benchmarks, weights, active channels, and refresh times as explicit controls. Campaign Execution pairs budget-shift charts with platform filters and detailed tables, retaining a familiar way to inspect the numbers. Log keeps past actions in a chronological, searchable record, with persistent navigation linking the three views.
03 · What shipped
Make budget movement visible without losing the record.
The team built an execution view where operators can scan budget shifts on a chart, change the channel or metric, and return to the underlying table for detail. Keeping both forms mattered: the visual makes a movement easier to notice, while the rows preserve the familiar evidence needed to examine it.
We moved the growing action history into PostgreSQL and exposed it through a filtered, creation-time log. An operator can look up earlier activity instead of working backward through hundreds of spreadsheet entries, with the configuration still available when a setting needs to change.
The application connected to advertising platforms including TikTok, Facebook, and YouTube, with OAuth for account access. I coordinated client review and delivery while the team used Docker and Brand Networks' AWS registry for the QA handoff.
Monitor the optimization
Engineering noteMove the action history out of the spreadsheet
A single day could produce hundreds of optimization events, so an operator needed to find one action without scanning the entire history. The team stored events in PostgreSQL and made the history filterable and chronological.
- The log orders events by creation time so an operator can reconstruct when a change happened.
- Filters and adjustable row counts narrow the visible records when investigating an event.
- The execution chart and table retain the budget context while the log provides the audit trail.
04 · Reflection
Keep the operator's logic while changing its form.
The sheet was cumbersome, but it also held budget logic the operators knew. Moving to a web application let the team separate controls from reporting and history while preserving familiar configuration parameters. That kept the migration focused on the decisions inside the spreadsheet rather than on changing its appearance.
Retaining tables beside charts was the useful compromise. A chart makes a shift apparent, but the table keeps the exact numbers an operator needs to question it. The new log extends that principle to earlier changes, preserving detail while making the record easier to navigate.