Webinar Summary
Morten walked through two use cases.
The first: bi-weekly client reporting. Before this setup, the team was pulling screenshots from Data Studio, pasting them into Google Slides, and manually adding commentary — a time-consuming process that couldn't easily adapt in real time. Now, a detailed prompt pulls cross-channel Google Ads and Meta data, generates an interactive dashboard artifact in minutes, and lets Morten adjust any view on the fly during a client meeting.
The second use case was campaign optimization: using Claude projects (loaded with client-specific context, goals, and a custom social media skill) to pull seven-day performance data across TikTok, Meta, and Snapchat and surface a prioritized action list — what to fix now, what to look at soon, what to watch longer-term. What previously took an hour or two of cross-platform digging now takes a few minutes.
A recurring theme throughout the session was trust. Morten spent several weeks stress-testing the data accuracy before using it with clients, starting with just three metrics from Google Ads and scaling up. He's never found an error.
His advice for reducing hallucinations: don't just throw questions at a general AI. Build skills that train it to think like your team, use projects to load client-specific context, and write precise prompts. The architecture you build around the LLM matters as much as the LLM itself.
Layer maintains a strict human-in-the-loop principle: reporting and optimization recommendations can be automated, but no budget changes or campaign edits go through without manual approval. Morten also teased early testing of Supermetrics' new campaign management feature, which would allow making those changes directly from a Claude conversation — without switching platforms.
Key takeaway: The hours saved aren't the most valuable part. It's what those hours get redirected toward. When data retrieval becomes conversational, strategists can spend more time on strategy — and clients notice.
Ready to try Claude for your own reporting and campaign optimization?
