RUSHA AI for Internal Audits & Proactive Monitoring

This guide covers using RUSHA as your background account manager: two automated scans that run without you asking anything, so problems surface before a client notices them.

When to Use This

• You manage more accounts than you can manually review every day.
• You want to catch a spend spike, a lead drop, or a broken campaign before the client emails you about it.
• You want a defensible, data-checked diagnosis ready the moment a client asks "what happened?"

The Two Background Scans

1. Anomaly Detection

Runs automatically and compares each client's last 7 days against their own historical baseline (4–8 weeks back) across spend, conversions, revenue, contacts, deals, and email metrics. When a metric moves more than 2 standard deviations from that baseline — in either direction — RUSHA writes a diagnosis and files it as a draft.

Typical catches: a campaign that suddenly stopped spending, a lead-form conversion rate that collapsed, an email list that started bouncing harder than usual.

2. Correlation Detection

Runs separately and looks for strong statistical relationships (Pearson correlation, |r| > 0.7) between metrics across all of a client's connected platforms, over an 8-week window — e.g. "email click rate rises in lockstep with open rate," or "Meta spend increases track with organic search dips." Each correlation is cached for 24 hours per client so you're not shown the same relationship on repeat.

How to Review What the Scans Found

1. Go to Dashboard → RUSHA AI → Insights Feed.
2. Click a notification to expand it. You'll see two tabs, exactly like a chat answer:
   • Internal View — the blunt, exact diagnosis, written for you.
   • Client View — the same facts, reframed diplomatically, in case you decide to publish it.
3. Decide: Approve or Dismiss.

This step matters — a draft is never shown to a client on its own. You are always the human in the loop.

Using This as a Daily/Weekly Habit

• Morning check: open the Insights Feed across your book of clients before your first call. Anything sitting there is something RUSHA already found and verified — you're triaging, not hunting.
• Dismiss aggressively, approve selectively. Dismissing costs nothing — the scan itself is always free, whether it finds something or not. You only spend an insight credit when you decide a finding is worth surfacing to the client (see below).
• Use Internal View to prep, Client View to publish. The internal diagnosis is the fastest way to know exactly what happened before a call; the client version is what you'd actually want them to read.
• An unreviewed draft never expires. It sits at pending_review until you act on it — nothing is auto-approved or silently lost.

What This Costs

• The background scans themselves — running, finding nothing, or producing a draft you later dismiss — are always free.
• Approving a draft into something a client can see costs 1 insight credit from your monthly allowance. See Pricing, Quotas, & Billing for how the quota works.

Related Guides

• RUSHA AI for Client Calls & QBRs — using RUSHA live once you already know something's off.
• RUSHA AI for Client Reports & Proof of Value — what happens to an approved insight once it's published.
• Mastering the RUSHA AI Insights Engine — the technical deep dive on how the dual-voice system and guardrails work.