The Analytics section of the admin panel shows what clients do in your bot's storefront: how many people opened it, which products they viewed, what they added to the cart, where they got stuck at checkout, what they searched for and how much they ordered. Data is collected automatically from the mini app and from orders, nothing needs to be connected.
Open it: bot menu → Analytics (right after "Reports"). Switch the period with the buttons at the top: today, 7 days, 30 days or a custom date range. Figures are compared with the previous period of the same length, e.g. the last 30 days with the 30 days before them. Amounts are shown in the shop currency.
Analytics is not included in every tariff. If the page shows "Analytics is not included in the current tariff" instead of reports, change the bot tariff on the billing page. Events are collected from the moment analytics is enabled; history is not reconstructed retroactively.
Four tiles at the top of the page, each with the change against the previous period:
| Tile | What it shows |
|---|---|
| Storefront opens | How many times the storefront was opened in the period |
| Orders | How many orders were placed in the storefront |
| Conversion to order | The share of opens that ended with an order. The change is shown in percentage points |
| Abandoned carts | How many carts with items never became an order, and their amount. More abandoned carts is a bad sign, so the up arrow is red here |
The funnel is the client's path from opening the storefront to payment. A bar shows how many reached the step, the hatched part above it shows how many left on the way, and the "N left" badge gives the exact number. The biggest drop is highlighted in red — start there. The dashed line is the previous period level.
| Step | What counts |
|---|---|
| Opens | A cold start of the storefront. Returning from a minimized mini app is not a new open |
| Product views | A product card was opened. Repeated opens of the same card within one visit are not duplicated |
| Added to cart | The client added a product to the cart |
| Checkout | The checkout screen was opened |
| Orders | The order left the cart: placed with payment on delivery or sent to online payment |
| Paid online | The online payment succeeded. Pay-on-delivery orders never pass this step, so it can be lower than "Orders" even when every payment goes through |
Above the funnel there is a Visits / Unique clients switch:
Below the funnel:
Other events, if any (searches, category views, checkout blockers, link clicks), are shown as separate counters.
What to look for. A sharp drop between two adjacent steps is where clients leave:
Next to the funnel there are hints based on the period data: where most people leave the funnel, the main checkout blocker, a query that finds nothing, abandoned carts and orders from shared links. Each hint links to the section with details. While there are few opens, no conclusions about the funnel are drawn: drops on small numbers are random.
Daily dynamics of opens, product views, add-to-cart and orders for the period. Handy for measuring mailings and promotions: a spike of opens on the mailing day and whether it reached orders.
Products ranked by card views for the period: views, unique viewers, add-to-cart, conversion (the share of views after which the product was added to the cart) and orders. Tabs:
The "Hidden" mark means the product is currently not shown in the storefront but was viewed during the period.
Carts with items that were not turned into an order: the client, last activity time, number of items and amount. Tabs Last 24 hours, 1–7 days and All: clients come back best within the first day. The "Message clients" button opens a new mailing with the clients from the list — remind them of the cart or offer a promo code.
The tile at the top counts all carts of the period, while the table shows the latest 100.
Where exactly clients get stuck on the checkout screen. Each row is a readable reason, for example "phone is not filled in", "no payment method selected", "the address is outside the delivery zones", "the amount is below the minimum order".
If one reason repeats across many visits, simplify that field, extend the delivery zone or lower the minimum order.
What clients searched for in the storefront and how often the search found nothing. Zero-result queries are a direct list of what is missing in the catalog or named differently from how clients search.
Statistics for product links created with the "Share" button in the storefront: who shared, clicks, storefront opens, orders and the paid amount. An order is attributed to a link when the client came via the link and placed the order on the same day.
Raw events are written immediately, while reports are built from summaries recalculated once an hour, so figures for the current hour may lag. Summaries are kept indefinitely, detailed events for the current month and the two previous ones. That is why the "Unique clients" mode is available only for periods within these limits.
The same reports are available to an AI agent via MCP (see AI agents): tools get_funnel (the funnel with unique clients, revenue, average check and abandoned carts totals), get_top_products, get_abandoned_carts, get_checkout_blocks, get_search_stats, get_share_link_stats. Ask the assistant "at which step do we lose clients this week" or "what was the average check this month compared with the previous one".
If Metrica or GA counters are connected in the shop settings, the storefront sends the same events to them under standard names (view_item, add_to_cart, begin_checkout, purchase and others). This lets you build reports in familiar tools, but the Analytics section in the admin panel works without them.