How we measure our metrics
heysteff is an AI platform for customer support and sales across WhatsApp, Instagram, Messenger, Gmail and Shopify; Steff is the AI agent that runs it. This page summarizes what the three metrics we publish mean — 4.2x, 87% and under 9 seconds — and how they're calculated. The canonical source, with the full definition of each one, is /metodologia: if anything here and there doesn't match, the methodology wins.
4.2x: cart recovery
It's the average multiplier by which heysteff workspaces that enable AI-powered abandoned cart recovery on WhatsApp recover more sales, compared to doing no active follow-up at all. It's an internal average across active workspaces with that feature enabled, not a universal projection — each store's result depends on its catalog, its traffic and how well the recovery conversation is set up.
87%: resolutions without a human
It's the average percentage of conversations Steff closes without a human agent having to step in, measured across active workspaces. It's different from the "70-90% of repetitive inquiries" range we also publish: that range describes the kind of inquiry the AI can take on; the 87% is the real average of resolution without escalation, aggregated across different businesses and inquiry types.
Under 9 seconds: first response
It's the average time it takes Steff to send the first reply from the moment a customer's first message arrives, on any of the supported channels, 24 hours a day. It's measured from the inbound message, not from a later assignment or review.
◆ canonical source
The exact definition, measurement period and criteria for each metric live at /metodologia — this page is just a summary to orient you quickly.
Internal averages, not guarantees
All three figures are internal averages calculated across active heysteff workspaces, not a guaranteed result for any new business. Your result depends on your catalog, your volume, your knowledge base and how much you configure the platform; it can land above or below these averages. We publish them because they're real and measurable, not because they're an assured floor — and that's exactly why we document how they're calculated instead of leaving them as a loose claim.
Why we publish the methodology instead of just the number
Any loose figure — "4.2x", "87%", "under 9 seconds" — without explaining how it's calculated is impossible to verify and easy to inflate. We prefer to expose the criteria: which workspaces it's measured on, what period is considered and what exactly counts as "resolved" or as "first response". If you're comparing heysteff against another vendor, ask for the same transparency before putting the numbers side by side — two figures with different definitions aren't comparable even if they look alike.
Related
◆ next step
Get a demo and measure these numbers in your own operation, not ours.