See What Each Client Costs: 16,900+ Leads Traced for Aura Agency

Aura Agency ran paid advertising on Meta and took in leads through Instagram DMs, Respond.io chats and organic marketing. Each source kept its own records, so the team couldn't see which channel brought in the clients who signed.

An attribution pipeline that pulled every lead into one place:

Matching: the pipeline matched each lead across the 4 sources, removed duplicates and enriched the record.

Storage: 5 linked Notion databases, kept current by daily syncs. New Respond.io messages arrived as they happened, and older conversations came in too.

Questions in plain English: a Claude query layer, so nobody on the team had to write SQL.

Reports: weekly snapshots on a schedule, plus a Telegram alert naming the workflow and the step whenever one failed.

n8n, Notion, Respond.io, Meta Ads, Python, Claude and Claude Code.

The team traced 16,900+ leads to their source and saw, for the first time, which channels produced signed revenue, with the cost per client and the payback for each channel.

I designed the architecture, mapped the data across all 4 sources, built the automations and the attribution logic, added the Claude query layer and documented how to maintain it.

Attribution systems like this one are a Custom Production AI System. Start with the Free AI Systems Fit Call and bring the question you can't answer about your ad spend.