Sales metrics in CRM: the three numbers worth checking daily

When a sales manager first opens a CRM dashboard, the eyes go wide. Funnel, sources, average deal size, LTV, ROMI, heatmaps, forecasts. We have seen companies across Central Asia spend a month building those reports, only to make decisions at the morning meeting based on gut feeling. A pretty picture does not answer the question "what should I do today?"

After years of implementations in different cities, we came to a simple rule: if a manager or a sales lead opens the CRM in the morning and cannot find three numbers within a minute, the system is running at half power. Here are those three.

Time to first response

Not the monthly average for the whole team. Not the median across all channels. It is per lead, and especially for those who wrote outside working hours or on weekends.

A customer in Tashkent messages on Instagram Direct at 10:40 pm. The manager replies at 9:15 am. Formally, everything is fine: the workday started, the reply went out. In reality, during those ten hours the person already messaged three other sellers and probably bought from whoever answered first. Time to first response is not about speed for its own sake. It is about whether your reply lands while interest is still hot.

Look at the breakdown by channel, not just the overall number. WhatsApp and Telegram usually get faster replies than Instagram, because the manager sees a notification on their phone and answers from a personal account. Direct gets checked once an hour. If all channels are connected in one CRM window — Instagram, WhatsApp, Telegram, Facebook Messenger — that gap stops being invisible. The manager sees every conversation in one place, and first responses land in minutes, not hours.

Voice messages are a separate story. A customer records two minutes of audio, the manager cannot listen in a noisy office, so it gets postponed. An hour later the moment is gone. If the CRM transcribes voice messages into text, the first response gets faster without any extra discipline. This is not about control. It is about not losing people for no reason.

Stage conversion, not overall funnel conversion

Overall lead-to-deal conversion is a nice number for an owner's report, but useless for daily work. It does not tell you where things break. Deals die at a specific transition, not "in the funnel."

Look at every step separately. Lead → qualified. Qualified → proposal sent. Proposal → invoice. Invoice → payment. Each transition has its own norm, and it depends heavily on the industry. In retail through Instagram, a 15–20% conversion to payment can be normal. In B2B with a long cycle, 3–5% is fine. Comparing yourself to some abstract benchmark from the internet is pointless. Compare a stage to itself a month ago.

Where does it usually break? At "proposal sent → customer replied." The manager sends the file, sets a task to call in two days, and moves on. The customer opens the PDF, gets distracted, forgets. Two days later the call is out of place. The stage sits there, the deal ages, conversion drops.

A funnel with suspiciously smooth edges is almost always a sign that stages are moved by hand before the morning meeting, not in the moment of a real conversation with a customer.

If deals sit at one stage for weeks, that is not a customer problem. It is a problem of the manager not knowing what to do next. Sometimes it is enough to rename the stage: not "proposal sent" but "customer confirmed they reviewed it." A different name changes behaviour.

Repeat purchases

The most undervalued metric in small business across Central Asia. Everyone counts new customers because they are visible. Old ones quietly leave, and nobody notices.

Look not at the overall repeat rate, but at the time between the first and second purchase. If your average customer returns in 45 days and a specific person has not bought in 60, that is a signal, not statistics. You can act on it: message them, ask, offer something. Without a CRM this is physically impossible to track. A manager's memory holds 20–30 active contacts, then the fog rolls in.

The second number next to it is the share of customers who bought a second time within three months of the first purchase. It shows how well the product and service retain people. If it drops, no amount of new traffic spend will save you. Acquiring a new customer always costs more than bringing back an old one. Everyone knows this, yet reports only show the first part.

This is where a unified chat history helps. The manager sees that the customer bought six months ago, what was discussed, how it ended. Without it, a repeat contact starts from zero: "hello, did you order something from us?" — and the customer feels like a row in a database.

Dashboards you can skip

Do not get us wrong: reports matter. But some blocks in small business only create an illusion of control.

• Revenue forecast based on stage probabilities. Until managers move deals based on real conversations rather than a script, the forecast is guesswork. It becomes useful when stages reflect reality instead of the desire to close the month nicely.
• Average deal size for the whole team. One large deal skews the picture, and the sales lead starts hunting for a problem that does not exist. Look at the median or break it down by manager.
• Lead sources in percentages without cost. "40% of leads from Instagram" — so what? Without acquisition cost and conversion to payment, it is decoration, not a metric.
• Manager ranking by number of calls. A call is not a call. Better to look at each manager's conversion to the next stage — it shows more honestly who is working and who is faking activity.
• Hourly activity heatmaps if you have fewer than ten managers. On small numbers they lie: one deal at 11 pm paints an "evening peak" that does not exist.

At shamCRM we deliberately do not build dashboards for the sake of dashboards. More than 120 companies and around 2,000 users is enough to see: people actually look at three or four numbers and a task list. Everything else gets opened once a month, before a report.

What this looks like in practice

A sales lead's morning: time to first response for yesterday's leads broken down by channel, a list of deals stuck at a stage longer than the norm, and customers whose repeat purchase window has arrived. Three screens, five minutes. Then conversations with managers about specific names, not "the team got worse overall."

If you want to test this on your real deals rather than demo data, start with the 14-day trial. Channels connect without a developer: Instagram, WhatsApp, Telegram, Facebook Messenger. An AI agent replies in chats, transcribes voice messages, describes photos, and hands off to a manager so the first response does not wait for someone to free up. Payme, Click, Alif payments, a storefront, and a Telegram mini app live inside the CRM. Integrations are included in the plan.

And do not start with dashboards. Start with one question: how much time on average passes between a customer's message and the manager's first reply? If you cannot answer within a minute, your CRM is still working as a notebook, not as a sales tool.