Conversion Sciences Prompts for Chatting with Analytics

**PROMPT: Business Health Audit — Traffic, Conversions & Trends**

Act as a senior CRO analyst and business intelligence advisor. Your job is to deliver a complete, plain-language health report on this business using available analytics data. Do **not** change any configuration — analysis only. Present findings layer by layer so a non-technical business owner can understand what's working, what isn't, and where to focus next.

**Step 0 — Establish business context:** Infer the business model (ecommerce, lead-gen, SaaS, content, marketplace, etc.), the primary conversion action (purchase, lead form, trial, demo request, etc.), and the likely customer journey. State your assumptions explicitly before proceeding. Identify the key conversion event(s) that represent revenue or pipeline. Flag any tracking anomalies (duplicate events, missing conversions, bot inflation) that would distort the numbers — correct for them before reporting.

**Step 1 — Twelve-month trend (big picture):** Pull monthly sessions, users, and conversions for the last 12 months. Identify:
- Overall trajectory: is the business growing, declining, or flat?
- Any months with unusual spikes or drops — and what likely caused them (seasonality, campaign, tracking issue?)
- Year-over-year change if data allows
- Summarize in one sentence what the 12-month trend says about the business

**Step 2 — 90-day snapshot (current state):** Pull sessions, users, engagement rate, and conversions for the last 90 days. Compare to the prior 90-day period. Report:
- Overall volume and direction (growing or declining vs. prior period?)
- Conversion rate trend: improving, declining, or flat?
- Any channel, device, or geo shifts between periods worth flagging

**Step 3 — Traffic channel audit:** Break down sessions and conversion rate by channel (organic search, paid search, direct, email, social, referral, etc.) for the last 90 days. For each channel report:
- Sessions and % of total traffic
- Conversion rate (not just volume)
- Engagement rate
- Verdict: is this channel **working** (high CVR, growing), **coasting** (volume but low CVR), **underperforming** (low both), or **at risk** (declining)?
- Highlight the single best and single worst performing channel by CVR, with a plain-language explanation of what that means for the business

**Step 4 — Landing page audit:** Pull the top 20 landing pages by sessions for the last 90 days. For each, report sessions, engagement rate, and conversion rate. Then:
- Identify the top 5 pages by conversion rate (with meaningful volume — flag pages with fewer than 50 sessions)
- Identify the bottom 5 high-traffic pages with low conversion (the biggest missed opportunities)
- Flag any landing pages with anomalous engagement rates (near 0% or near 100%) as potential tracking or bot issues
- Summarize: which landing page is the single biggest opportunity for improvement?

**Step 5 — Products, offers & conversion leaders:** Identify which products, service pages, offer pages, pricing tiers, or CTAs are generating the most conversions. If ecommerce, pull by product or category. If lead-gen or SaaS, pull by the pages or flows closest to the conversion event. Report:
- Top 5 converting offers/products/pages by volume
- Top 5 by conversion rate (with volume guardrail)
- Any notable gap between high-traffic offers and high-converting offers (i.e., are you sending traffic to the wrong things?)

**Step 6 — Audience insights:** Break down the sellable traffic (real potential buyers — exclude obvious bots and data-center traffic) by:
- Device: desktop vs. mobile vs. tablet — CVR and engagement for each
- Geography: top 5 countries/regions by sessions and by CVR
- New vs. returning users: which converts better and by how much?
- Flag any meaningful segment gap (e.g., mobile converts at half the desktop rate) as an insight

**Step 7 — Business owner summary:** Write a plain-language executive summary (no jargon, no metric names, no GA terminology) that answers these questions as if talking directly to the business owner:
- Is the business growing or shrinking online right now?
- Which marketing channels are actually bringing in customers (not just clicks)?
- What pages or offers are doing the heavy lifting?
- Where is the biggest leak — where are you losing potential customers?
- What is the single highest-leverage thing to fix or double down on right now?

**Deliverable format:** business assumptions → 12-month trend → 90-day snapshot vs. prior period → channel audit table → landing page opportunity table → product/offer conversion leaders → audience breakdown → plain-language executive summary → top 3 prioritized action items (ICE scored). Cite specific numbers, page paths, and event names throughout. Flag anywhere sample size is too small to draw conclusions.

Sellable Traffic Prompt

Use this prompt to eliminate traffic that isn’t your prospects and customers:

Bots
Blog readers
SaaS logins
Internal
Job seekers

Paste this prompt into your AI chat connected to Google Analytics.

Click to View and Copy

**PROMPT: Identify "Sellable Traffic" — Buyer-Intent Segment Isolation**

Act as a senior CRO/analytics auditor. Your job is to define and quantify **"sellable traffic"** for this website: the subset of total traffic that plausibly represents real potential buyers. This is the denominator that actually matters for conversion rate, media efficiency, and CRO prioritization. Do **not** change any configuration — analysis only. Build the segment transparently, layer by layer, so I can see what was removed and why.

**Step 0 — Establish business context:** Infer the business model (ecommerce, lead-gen, SaaS, content/ad-supported, marketplace), the buying action that defines a "buyer" (e.g., `purchase`, `generate_lead`, trial start, demo request), and the realistic buyer journey. State assumptions explicitly — the exclusions depend on this. Confirm which conversion event(s) count as the money action.

**Step 1 — Measure the baseline:** Pull total sessions and users for the last 90 days (and note the trend). This is the gross number everything gets subtracted from.

**Step 2 — Identify and quantify each non-buyer segment.** For every category below, quantify sessions, show the filter/definition used, and report its engagement rate + conversion rate so we can confirm it genuinely doesn't buy before excluding it. Never exclude silently.

1. **Bots / invalid traffic** — Direct-channel spikes far above the daily average, data-center cities (Ashburn, Council Bluffs, Boardman, Beaverton, Moses Lake, Des Moines, etc.), suspicious geos with near-0% engagement, 0-second sessions, impossible session counts. Flag day-level anomalies too.
2. **Internal / employee traffic** — known office geographies, internal IP patterns if identifiable, staging/dev hostnames, unusually deep repeat visitors, QA/test paths.
3. **Existing users / logins (SaaS or account-based)** — sessions landing on or hitting `/login`, `/account`, `/dashboard`, `/app`, password reset, or authenticated areas; returning logged-in users who are customers, not prospects.
4. **Low-intent organic content/blog traffic** — landing pages matching `/blog/`, long-tail informational slugs, glossary/resource pages with high bounce and near-0% conversion. Quantify what share of blog sessions ever reach a buyer action before excluding.
5. **Support / careers / investor / PR traffic** — `/support`, `/help`, `/careers`, `/jobs`, `/press`, `/legal`, `/privacy`, unsubscribe pages — visitors with no purchase intent.
6. **Existing-customer service traffic** — order tracking, returns, warranty, "my orders," login-gated reorder flows (for ecommerce).
7. **Any other non-buyer segments you recommend** — e.g., referral spam, self-referrals, affiliate/coupon-only traffic, accidental/misrouted campaign traffic, geographies the business can't sell to (out of shipping/service area). Propose these explicitly and justify each.

**Guardrail:** Before removing any segment, verify it actually converts at a negligible rate. If a segment converts materially (e.g., blog traffic that assists conversions, or logged-in users who upgrade), do **not** exclude it — flag it as "keep, but monitor" and explain. The goal is to isolate non-buyers, not to inflate the conversion rate by cherry-picking.

**Step 3 — Build the waterfall.** Present a subtraction table:

Account for overlap (a session can be both bot and blog) so you don't double-subtract — apply exclusions in a defined order and note overlaps.

**Step 4 — Report the result:**
- Final **sellable traffic** count and what % of gross traffic it represents.
- **Conversion rate on gross vs. sellable traffic** side by side, so the impact is obvious.
- Channel mix, device, and geo breakdown of the sellable segment only (this is the audience CRO should actually optimize for).
- Recommended **reusable segment definition** (as a set of GA4 filters / HogQL conditions) so this can be applied consistently in future reports.

**Deliverable format:** business assumptions → per-segment evidence table → exclusion waterfall → sellable-traffic summary with gross-vs-sellable CVR → reusable segment definition → caveats and open questions. Cite specific numbers, page paths, and event names as evidence. Where sample sizes are small or a segment's non-buyer status is ambiguous, say so before excluding.

Weekly Email Brief

Use this prompt to get a weekly review of your online business based on data.

Paste this prompt into your AI chat connected to Google Analytics.

Click to View and Copy

## Weekly Analytics Check-Up — Email Brief Prompt

**Role:** Performance-focused digital marketing analyst producing a weekly health check. Compare the **last complete 7 days** to the **prior 90-day daily trend** (use the 90-day daily average as baseline so day-of-week noise doesn't distort the read). Surface what changed, whether it matters, and what to do. Lead with signal.

**Detect the business model first** (ecommerce, lead gen, or SaaS) from the events, key events, and ecommerce metrics present. State which you detected. Run shared checks for all; add the model-specific checks that apply. If ambiguous, report on whatever conversion events exist and say so.

**Always investigate:**
1. **Traffic** — Sessions, users, engagement rate: 7-day vs. baseline. Flag changes beyond ~±15% or clearly outside normal daily variance.
2. **Channel mix** — Sessions and conversions by `sessionPrimaryChannelGroup`. Call out material moves; separate volume (sessions) from efficiency (conv. rate). Note if one channel drives the overall change.
3. **Conversion rate** — Overall and by channel. Distinguish "fewer conversions from less traffic" vs. "lower rate."
4. **Landing pages** — Biggest movers by sessions and conv. rate; flag high-traffic landers that deteriorated.
5. **404 / errors** — Detect spikes in Page Not Found views. Report URLs, volume vs. baseline, and top referrers — usually a broken link, bad campaign URL, or deploy issue.
6. **Device/geo anomalies** — Only when a segment drives the change or looks like bot/spam.

**If ECOMMERCE:** Revenue & AOV vs. baseline — decompose revenue into traffic × conv. rate × AOV. Purchases/transactions. Cart & checkout funnel (`add_to_cart` → `begin_checkout` → `purchase`): step drop-off and increases in cart/checkout abandonment.

**If LEAD GEN:** Lead volume and lead conv. rate. Form funnel (view/start → submit): flag increases in form abandonment and which form/page. Lead-quality proxies if available (directional only).

**If SaaS:** Signups/trial starts and signup conv. rate. Activation funnel (signup → activation → paid): flag drop-off. Trial-to-paid movement if trackable.

**Standards:** Weigh statistical reliability before alarming — call out small samples; prefer "worth watching" over "urgent" when data is thin. Apply known data-quality rules (bot/blog/internal exclusions, attribution artifacts) and state any exclusions explicitly, never silently. Quantify every claim (absolute + %) and attribute causes only where data supports it.

**Output:**
- **TL;DR** — 3–5 bullets: top changes and likely drivers.
- **Metrics table** — last 7 days, 90-day baseline, % change, direction. No emojis in tables.
- **What changed & why** — prioritized narrative, biggest impact first.
- **Watch list / actions** — concrete next steps by urgency (Urgent / Monitor / FYI).
- Concise, skimmable, plain language.

30,000 Tests for 300+ Websites

Still Have Questions?

Let’s talk. Call us at 888-961-6604 or click below to schedule a Free 30 minute strategy session with our Conversion Optimization Experts.

Get My Assessment

Conversion Sciences® and Conversion Scientist® are federally registered trademarks of Conversion Sciences LLC. Any unauthorized use is expressly prohibited.

Privacy