Revenue Over Vibes: The New Standard for E-commerce Analytics

Revenue attribution is the process of assigning credit for a sale to the marketing touchpoints, tools, or interactions that influenced it. In e-commerce, understanding attribution tells you which channels, apps, and experiences are actually driving revenue—versus which ones just look busy on a dashboard. Without proper attribution, you can end up paying for tools that don't move the needle or cutting channels that are quietly driving most of your conversions. For AI tools specifically, attribution is the only honest way to evaluate ROI.

The Last Interaction (or Last Touch) model assigns 100% of the credit for a conversion to the final touchpoint before purchase. It's most appropriate for tools that operate at the bottom of the funnel—i.e., tools that engage customers when they are already considering a purchase and help them make a final decision. An AI agent that answers a sizing question or applies a discount code right before checkout is a bottom-of-funnel interaction, making Last Interaction attribution the most relevant model for measuring its impact. For top-of-funnel awareness channels like social ads, multi-touch attribution models are more appropriate.

Vanity metrics are numbers that look impressive but don't connect to business outcomes. In AI tool reporting, common vanity metrics include: total messages sent, widget open rate, conversation count, and "engagement rate" (undefined). The litmus test is simple: if you cannot draw a direct line from the metric to dollars in your bank account, it is a vanity metric. ShopGuide's philosophy is to report on metrics that answer the question, "Did this make me more money, and by how much?"

AOV Delta is the difference in average order value between customers who interact with the AI agent and those who don't, over the same time period. To calculate it: (1) Segment your orders into "agent-assisted" and "non-assisted" cohorts over a 30-day window. (2) Calculate the mean order value for each cohort. (3) The difference is your AOV Delta. This metric matters more than raw revenue attributed because it directly measures the agent's ability to uplift purchase behavior—upselling, cross-selling, and helping customers find higher-value products. A consistent positive AOV Delta proves the tool is actively improving the quality of your sales, not just counting the ones that would have happened anyway.

Yes—this is the feedback loop that separates good AI deployments from great ones. The "Discovery Heatmap" in ShopGuide analytics reveals exactly which questions lead to conversions and which lead to drop-offs. If a cluster of customers is asking about a specific product feature and then leaving, that signals either a product gap, a pricing issue, or a knowledge gap in the agent. By addressing these analytically identified friction points—whether through product changes, price adjustments, or agent corrections—you systematically improve your conversion funnel over time.

ShopGuide uses a configurable attribution window (default: 24 hours after the last agent interaction) to account for customers who needed time to consider. If a customer had a substantive interaction with the agent—asked about sizing, received a recommendation, engaged with a discount offer—and then purchased within that window, the sale is attributed as agent-assisted. You can adjust this window in the dashboard to match your store's typical consideration period. For high-consideration categories (furniture, electronics, custom orders), a 72-hour or 7-day window is often more appropriate.

A well-optimized AI agent on an active Shopify store should attribute 15–35% of total revenue as agent-assisted, depending on the store's traffic mix and product category. If the rate is below 10%, the agent is likely not positioned prominently enough or is not being used for high-intent traffic. If the rate is above 40%, it may indicate over-counting (e.g., attribution window is too long) or the agent is genuinely central to the purchase flow, which is the goal. Track this number monthly and compare it to period-over-period revenue growth to see if improving agent performance correlates with store growth.

Use three numbers: (1) Revenue Attributed—total revenue from agent-assisted orders in the period; (2) AOV Delta—the percentage difference in order value between assisted and non-assisted customers; (3) Payback Period—the number of days it took for attributed revenue to exceed the cost of the tool. These three figures tell a complete story: the tool is driving real sales, it's improving the quality of those sales, and it pays for itself quickly. Vanity metrics like "chats handled" are not relevant to this conversation.