CRM metrics reference

Last updated: see the newest file in docs/crm-metrics/log/ Maintenance contract: any change to lib/crm-metric-catalog.ts, lib/sources/amocrm.ts, lib/sources/bitrix.ts, or lib/sources/hubspot.ts that adds, removes, or changes the meaning of a metric must also update this file in the same commit. The publish target for this doc is the Reportagram help site, so customers will read it directly.

Every metric below answers a question a sales manager, media buyer, or CEO asks. The doc groups metrics by what question they answer, not by alphabetical name. For each metric: precise definition, how it's computed, where it's available, when it's useful, and what to watch out for.


Concepts you need first

Lead vs Deal

  • amoCRM unified leads and deals in 2020 — every record is the same entity, just at different stages. "Lead created" and "Deal created" are the same event.
  • Bitrix keeps Leads and Deals as separate tables. A Lead can be converted into a Deal, which then has its own pipeline.
  • HubSpot has Contacts (people) and Deals (sales opportunities) as separate objects. A contact can exist without a deal.

This is why some metrics are present on one CRM but not another — the underlying data model differs.

Window Every metric is computed over the report window (e.g., yesterday, last week, last month). Snapshots like "Active deals" reflect the state at the END of the window.

Qualifying call (a conversation) A call of 45 seconds or more that the CRM doesn't say failed. A phone call < 45 seconds is voicemail, hang-up, wrong number, or no-answer (the sales industry's "Bridge Group rule"). The seconds can include ringing, so when the CRM says a call didn't connect, it doesn't count however long it ran.

CRMWhat it tells us
amoCRMThe call's status, when the telephony sends one: "conversation took place" and 45 seconds or more counts; "couldn't reach" never does. Without a status, 45 seconds
Bitrix24Only whether a call was missed. Otherwise 45 seconds, so a long unanswered ring can count. The real result needs a telephony permission we don't ask for
HubSpotHas a "Connected" outcome; the by-lead lines aren't on HubSpot yet

We use this for any metric that asks "did we actually talk to this lead?"

Stage mapping amoCRM pipelines are 100% portal-custom — every customer names their stages differently. For metrics that depend on funnel position (Contacted, Quality), each amoCRM user maps their stages to canonical funnel positions once per portal in the wizard. Bitrix uses its native STATUS_SEMANTIC_ID (universal), so no mapping is needed there. HubSpot uses its standard lifecyclestage values.


Volume — how many leads / deals?

Chat inquiries (chatInquiries, amoCRM)

Definition: chats and comments waiting in Unsorted, created in the period. amoCRM's own mark (category = chats on /leads/unsorted), so it includes Instagram and Facebook comments. They aren't leads: they're left out of leads created, lead cost, conversion, reached / not reached and top sources. Becomes a lead: when a person moves it out of Unsorted, whether or not it has a phone. A chat with a phone that's still in Unsorted is still a chat. Chat messages on a lead aren't used for this: the only test is whether the Unsorted row is still there. Past days can change: a moved chat counts as a lead on the day it was created, so a re-run of a past day can show more leads and fewer inquiries than the report sent then. Sent reports aren't changed. Leads moved between pipelines can also change a past day (#638). Not a metric to pick: it appears on amoCRM reports with any of those lines when it's above 0. For a report's first 7 days with chats, a note says lead cost changed because of this, not because of the ads. Reports created afterwards don't get the note. Still includes chats: time to first response, the stage report and sales by ad. Bitrix24: not split. This is the exception to using one rule on every CRM: it has no signal for a chat (HAS_IMOL on 0 of 2,506 leads, and source names are each portal's own), and its direct leads behave like other leads. Don't use HAS_PHONE as a stand-in: phones can sit on the contact. HubSpot: not split, no Unsorted.

Leads created (leadsCreated)

Definition: Count of new leads/records created in the window. Computed from:

  • amoCRM: /api/v4/leads?filter[created_at], plus the same dates asked for the pipeline's Unsorted stage ("Неразобранное") by id: amoCRM leaves leads still waiting in Unsorted out of the plain list
  • Bitrix: crm.lead.list?filter[>=DATE_CREATE]
  • HubSpot: contactsCreated plays the equivalent role: a lead from an ad lands in HubSpot as a contact (see below).

Available on: amoCRM, Bitrix. Default: ✓ Yes. Best for: Top-line volume trend. Track daily/weekly to see if your lead-gen channels are working. Watch out for: This counts ALL new leads regardless of quality. Pair with notQualifiedLeads to see how many were unreachable. On amoCRM, leads nobody has accepted from Unsorted yet are counted too, on the day they arrived.

Contacts created (contactsCreated, HubSpot)

Definition: New contacts created in the window. Computed from: HubSpot /crm/v3/objects/contacts/search filtered by createdate in window (a count; no contact details are read). Available on: HubSpot only. Default: ✓ Yes. Best for: The same job as leadsCreated: a lead from a Meta form or your website becomes a HubSpot contact, so this is the number to compare with Meta's leads. It's also what conversion rate and cost per lead divide by. Watch out for: It counts every new contact, including ones added by hand or by import. If HubSpot can't be read, the row shows "—", never 0.

Deals created (dealsCreated, HubSpot)

Definition: New deals created in the window. Computed from: HubSpot /crm/v3/objects/deals filtered by createdate in window. Available on: HubSpot only. Default: ✓ Yes. Best for: Deal flow. In most HubSpot setups a deal opens around SQL, so this is later in the funnel than contactsCreated; it is no longer the "leads" number (changed 2026-10-28).

Leads/Deals won (dealsWon, displayed as "Leads won" on amoCRM)

Definition: Records that closed as won during the window. Computed from:

  • amoCRM: /api/v4/leads?filter[closed_at] with status_id=142 (the system-wide "won" status).
  • Bitrix: deals that moved into a won stage during the window (Bitrix24's stage history, which has the exact time), each once. Not the deal's close date: that field has no time, so a report in a timezone other than the portal's used to count two days. A deal counts by its last closing move in the window: won and then reopened or lost before the window ends, it isn't won for that window; won before the window, it isn't counted again. So a week can show fewer than its days added up, or than Bitrix24's own count by close date. Answers about any dates count the same way as the report. If you picked your own sale column when connecting (for example a "Paid" lead status on a Leads-only portal, or a deal stage), it's the leads or deals that first reached that column or a later one in the window, each once. Failure columns never count. The revenue is their amounts.
  • HubSpot: deals where hs_is_closed_won=true and close date in window.

Available on: All three. Default: ✓ Yes. Best for: The headline "are we winning?" number. Pair with revenueWon for the dollar value. Label note: displayed as "Leads won" on amoCRM (unified Lead+Deal entity) and "Deals won" on Bitrix/HubSpot.

Leads/Deals lost (dealsLost, displayed as "Leads lost" on amoCRM)

Definition: Records that closed as lost during the window. Computed from:

  • amoCRM: status_id=143 with closed_at in window.
  • Bitrix: deals that moved into a lost stage during the window (stage history, each once), your own lost columns included.
  • HubSpot: hs_is_closed=true AND hs_is_closed_won=false.

Available on: All three. Default: ✓ Yes. Best for: The flip side of won. Pair with the loss-quality split below (notQualifiedLeads + lostQualifiedLeads) to understand WHY you lost them.

Active deals (activeDeals)

Definition: Snapshot count of open (not won, not lost) deals at the END of the window. NOT period-flow. Computed from:

  • amoCRM: all leads with status not in {142, 143} at fetch time.
  • Bitrix: all deals with STAGE_SEMANTIC_ID="P" at fetch time.

Available on: amoCRM (default), Bitrix (optional), HubSpot (default). Best for: Pipeline health at-a-glance. "Our pipeline has 165 open deals right now." Watch out for: This is heavy to compute on large portals — it fetches the entire open pipeline. If your portal has 10k+ open deals, expect slower reports.

Leads in progress (leadsInProgress, amoCRM only)

Definition: amoCRM's own name for open leads: every lead in the report's pipeline that is neither won nor lost at the end of the period — the same number as activeDeals (shown once if both are picked). Available on: amoCRM only. Default: ✗ Optional. Best for: Distinguishing "still being worked" from raw open count. Display-only — no cross-source canonical aggregation.

Deals created/updated/moved in window (dealsCreatedInWindow, dealsUpdatedInWindow, dealsMovedInWindow)

Definition: Window-bounded deal activity counts.

  • Created: deals where DATE_CREATE in window.
  • Updated: deals where DATE_MODIFY in window.
  • Moved: deals where MOVED_TIME (stage change time) in window.

Available on: Bitrix (default for created, optional for others). amoCRM has these as optional too but dealsCreatedInWindow is identical to leadsCreated for amoCRM (unified entity) — usually prefer the latter. Best for: Daily reports — "what changed today?" — when an active- deals snapshot is too noisy on portals with thousands of stale open deals.


Money — what was the revenue?

Revenue (won) (revenueWon)

Definition: Sum of price (amoCRM) or OPPORTUNITY (Bitrix) or amount (HubSpot) across all won deals closed in window. Available on: All three. Default: ✓ Yes. Best for: The headline dollar number. "We closed $48k this week."

Revenue (lost) (revenueLost)

Definition: Sum of values across all LOST deals closed in window. "The deals we missed out on." Available on: All three. Default: ✗ Optional (Bitrix/HubSpot have it default). Best for: Identifying high-value opportunities that slipped away. A high revenueLost with low notQualifiedLeads suggests sales execution problems on real opportunities, not lead-gen issues.

Active deals value (activeDealsValue)

Definition: Sum of price/OPPORTUNITY across all OPEN deals at end of window. Available on: amoCRM (default), Bitrix (optional). Best for: Pipeline value at a snapshot. Often used by CEOs as the "what's in the funnel?" number. Watch out for: Stale deals from 6 months ago pad this number. Pair with dealsCreatedInWindow for a freshness check.

Pipeline value (pipelineValue)

Definition: Alias of activeDealsValue on amoCRM; HubSpot's canonical pipeline value (sum of open deal amounts). Available on: All three.

Average deal size (avgDealSize)

Definition: revenueWon / dealsWon for the window. Available on: All three. Default: ✗ Optional. Best for: Tracking deal-size trends. Rising avg deal size = you moved upmarket. Falling = you discounted or lost big deals.


Telephony — raw call counts

No call records: if amoCRM or Bitrix24 holds no call at all in the 30 days before the report (no telephony connected to it), every call-based line (calls, leads contacted, not qualified / lost qualified) shows —, with one line saying so. Without call data they'd read "0 contacted" and "every lost lead not qualified", which isn't true; we just can't see the calls.

A failed read: when a read from the CRM still fails after retries (it refuses, or times out), the lines it feeds show —, on every CRM, and nothing is stored for them. The report still goes with the numbers we could read. A 0 always means a real zero. On HubSpot, calls show — when the connection has no permission to read call logs.

Incoming calls (incomingCalls)

Definition: Total count of incoming call EVENTS in the window (client → us). Note: counts CALLS, not distinct leads. A single lead calling 5 times = 5 incoming calls. Computed from:

  • amoCRM: /api/v4/events?filter[type]=incoming_call.
  • Bitrix: crm.activity.list?TYPE_ID=2&DIRECTION=1.

Available on: amoCRM, Bitrix. Default: ✓ Yes. Requires: Portal has telephony integration (Zadarma, OnlinePBX, Sipuni, etc.). Portals without telephony get 0. Best for: Inbound call volume — "how much demand are my ads / SEO / referrals generating?" Watch out for: Includes voicemails, hangups, wrong numbers. For "meaningful contact" use leadsContactedInbound instead.

Outgoing calls (outgoingCalls)

Definition: Total count of outgoing call EVENTS in the window (us → client). Counts CALLS, not distinct leads. Available on: amoCRM, Bitrix. Default: ✓ Yes. Best for: Outbound effort — "did my SDRs make their dials?" "Did the team work 200 calls or 30 today?" Watch out for: Same as incomingCalls — a one-minute conversation counts as 1 just like a 5-second hangup. Pair with leadsContactedOutbound for quality signal.


New leads with a phone conversation — by lead

New leads with a phone conversation (leadsReached)

Definition: Of the leads created in the period, how many had at least one call of 45 seconds or more, in either direction, within the period. Each lead counts once, however many times it was called. Available on: amoCRM, Bitrix24. Best for: The number owners mean by "how many of today's leads did we talk to?". It can never exceed new leads.

New leads without a phone conversation (leadsNotReached)

Definition: The rest of the leads created in the period: new leads minus reached. Includes leads closed in the period with no such call. Available on: amoCRM, Bitrix24.

A phone conversation means a qualifying call in the period (see "Qualifying call": 45 seconds or more, and not one the CRM says failed). "Without" means no such call: the team may still have reached the lead by chat or Telegram, which this line can't see.

How these differ from "Leads contacted" below: those count every lead a qualifying call touched in the period, so a call to last month's lead counts, and they split by direction. These count only the period's new leads, so the two numbers aren't the same.

When is it counted? At the end of the report's period, not when the report is sent. A lead that came in at 23:50 and was called at 09:00 the next morning stays "without a phone conversation" in that day's report. So a re-run of the same day reads the same, unless leads or calls are later deleted in the CRM.

Calls that count: the same as below: saved on the lead, its contact or its company (and on Bitrix24 its deal), and a missed call never counts. The incoming call that created a lead counts even though it's logged a few seconds before the lead.

If we can't see the calls: both lines show —, never 0: when the CRM holds no call records, when it refuses the read, or (amoCRM) when there are too many call notes to read all of them. The reason is logged.

Phone-contacted leads — meaningful conversations

These three metrics count distinct leads with at least one call of ≥45 seconds in the window. Different from incomingCalls / outgoingCalls which count CALLS.

Bitrix24: a call's length is its end time minus its start time, and a missed call never counts. Calls saved on a deal count for the lead the deal came from, even if that lead was converted weeks earlier: the call still reached that person. Calls saved on a contact count for that contact's leads that were open at the time of the call.

amoCRM: many telephony widgets save the call on the contact (or the company), not the lead. Calls saved on a lead count for that lead; calls saved on a contact or company count for its leads that were open at the time of the call. Only leads in the report's pipeline count.

If we can't read the calls (amoCRM or Bitrix24 refuses, or on amoCRM there are too many call notes to read all of them), these lines show —, never 0.

HubSpot note: these distinct-lead variants are not available on HubSpot yet. HubSpot's call data needs an extra batch-association fetch to know which call belongs to which contact, which we haven't implemented yet. HubSpot users see raw incomingCalls / outgoingCalls counts instead.

Leads contacted (inbound) (leadsContactedInbound)

Definition: Distinct leads who called us with a conversation ≥45 seconds. Available on: amoCRM, Bitrix. Default: ✓ Yes. Best for: Media buyer's "real high-intent demand" signal. Leads who called you back are 5-10x more likely to close than ones who just submitted a form.

Leads contacted (outbound) (leadsContactedOutbound)

Definition: Distinct leads we called with a conversation ≥45 seconds. Available on: amoCRM, Bitrix. Default: ✓ Yes. Best for: Sales manager's / SDR's team-effort signal. "How many leads did we actually reach out to and connect with?"

Leads contacted (any phone) (leadsContactedAnyPhone)

Definition: Distinct leads with ≥1 qualifying call in EITHER direction during the window. This is the UNION — a lead with both inbound and outbound qualifying calls counts as 1. Available on: amoCRM, Bitrix. Default: ✓ Yes. Best for: CEO/headline number. "We had meaningful phone contact with X leads this week." Can't be derived from inbound + outbound because of overlap.

Industry note: the 45-second threshold is the "Bridge Group rule" — sub-45-second calls are statistically voicemail, hangups, or no-answers, not conversations.


Funnel — Contacted & Quality (stage-based)

These count leads that crossed into specific funnel stages during the window. Use these when your team tracks contact/qualification via deliberate stage moves (not just phone calls).

Contacted leads (contactedLeads on amoCRM, leadsContacted on Bitrix)

Definition: Distinct leads that moved INTO a "Contacted" stage during the window. Computed from:

  • amoCRM: /api/v4/events?filter[type]=lead_status_changed where the destination status is mapped to contacted in the connection's stage mapping. Requires the user to map their portal's stages in wizard Step 3.
  • Bitrix: leads with STATUS_SEMANTIC_ID="P" (in-process). Bitrix has this built-in — no mapping needed.

Available on: amoCRM, Bitrix. Default: ✓ Yes. Best for: Teams that track contact via stage moves rather than phones (e.g., chat-based, email-based, or multi-channel outreach). Watch out for: amoCRM users must configure stage mapping in Step 3 the first time. The metric returns 0 until mapped — the wizard shows a "configure your Contacted stage" hint when unmapped.

Quality leads (qualityLeads on amoCRM, leadsQualified on Bitrix)

Definition: Distinct leads that reached a "Quality" / "Qualified" stage, or went past it, during the window. Qualified = matches your team's ICP / BANT criteria. Computed from:

  • amoCRM: the stage you mark as qualified or any later stage in its pipeline, won included, lost never. A lead counts once, in the period it first crosses that line (moved from an earlier stage or another pipeline); moving between later stages doesn't count again.
  • Bitrix: the same rule as amoCRM, on the rule you picked when you connected Bitrix24. If your portal uses the Leads kanban, a lead counts when it reaches the lead status you picked or any later one (converted to a deal is later). If it uses Deals, a deal counts when it reaches the stage you picked or any later stage in its pipeline, won included. "Later" follows the kanban's order in Bitrix24. Failure stages never count: lost, junk, or your own (like "didn't come"). Either way it counts once, in the period it first crosses that line. Until a rule is picked, it's the leads created in the window that sit in STATUS_ID="PROCESSED".

Available on: amoCRM, Bitrix. Default: ✓ Yes. Best for: Marketing-to-sales handoff health. High leadsCreated with low qualityLeads = lead-gen attracting unqualified traffic.


Loss quality split — WHY did we lose?

When a deal is lost, was it because we couldn't reach the lead, or because we engaged and got rejected? Different problems need different fixes. Needs call records in the CRM (see "No call records" under Telephony): without them both lines show —.

Invariant: notQualifiedLeads + lostQualifiedLeads = dealsLost in the same window.

HubSpot note: not available on HubSpot yet — same reason as phone-contacted leads above (needs call-to-contact batch associations). On HubSpot, look at dealsLost as the headline plus raw call counts to infer the same signal manually.

Not qualified leads (notQualifiedLeads)

Definition: Distinct leads that closed as LOST AND had no qualifying call (≥45s) in the window — i.e., we tried but never connected. Wrong phone number, no answer, bad data. On Bitrix24 any qualifying call before the close counts, not only calls on the report's day. That's a call on the deal, on the lead it came from, or on its contact while the deal was open. Available on: amoCRM, Bitrix. Default: ✓ Yes. Best for: Lead-gen quality signal. High notQualifiedLeads relative to leadsCreated = your lead source is feeding you bad contacts. Ad targeting too broad? Form not collecting valid phones? Action: tighten lead-gen, not sales training.

Lost qualified leads (lostQualifiedLeads)

Definition: Distinct leads that closed as LOST AND had ≥1 qualifying call in the window — i.e., we engaged but couldn't close. Available on: amoCRM, Bitrix. Default: ✓ Yes. Best for: Sales-execution signal. High lostQualifiedLeads = the team is having conversations but losing them. Pitch problem, competitor problem, product problem, pricing problem. Action: sales training, pricing review, or competitive positioning — not lead-gen.

Concrete example

This week: 20 deals lost. Split = 12 not qualified, 8 lost qualified.

  • Insight 1: 60% of losses are unreachable — fix lead source
  • Insight 2: Among reachable leads, conversion is 8/(12+8)=40% loss rate — sales execution OK but pitch could improve

Velocity — how fast?

Time to first response (timeToFirstResponse)

Definition: P50 (median) time between a lead's created_at and the FIRST outbound response from your team in the window. Expressed in human format ("16h 4m"). Computed from:

  • amoCRM: /api/v4/events?filter[entity_id][]=lead1,lead2,... — one batched query for the entire sample; finds the first event of type outgoing_call/outgoing_chat_message/lead_status_changed/ common_note_added/task_added per lead. System-fired events (created_by=0) and deltas < 30 seconds (automation echo) are excluded.
  • Bitrix: crm.activity.list batched via filter[@OWNER_ID][] for the entire sample. Same 30-second floor + same outgoing TYPE_ID set (1=Meeting, 2=Call, 4=Email).
  • HubSpot: not computed yet: the row shows "—". It's optional on HubSpot and out of the HubSpot presets until we read HubSpot's own first-engagement data, checked on a real portal (2026-10-28).

Available on: amoCRM, Bitrix (HubSpot shows "—" for now). Default: ✓ Yes on amoCRM and Bitrix; optional on HubSpot. Best for: Sales responsiveness. Industry research: leads contacted in <5min are 21x more likely to convert vs 30min. Track P50 to know where you stand. Watch out for: P50 (median) is reported, not average. Half your leads got a response faster, half slower. We use median because one slow-responded lead won't blow up the metric.

Time to won (timeToWon)

Definition: Median cycle length from lead created to lead won. Available on: amoCRM, HubSpot. Default: ✗ Optional. Best for: Sales velocity tracking. Rising time-to-won = deals stalling in pipeline.


Intra-CRM rates

These are computed within the CRM data — no ad spend needed.

Win rate (winRate)

Definition: dealsWon / (dealsWon + dealsLost) for the window. Expressed as percent. Available on: All three. Default: ✗ Optional. Best for: Sales execution efficiency at the close stage. Watch out for: Computed only on closed deals in window. Long sales cycles can make a single week's rate volatile — weekly+ windows give cleaner signal than daily.

Conversion rate (conversionRate)

Definition: dealsWon / leadsCreated (intra-CRM math). Percent. Available on: All three. Default: ✗ Optional. Requires: Ad source connected (UX gate; math is intra-CRM). Best for: Top-of-funnel efficiency. "Of every 100 leads, how many close?" Watch out for: Cohort mismatch — leads created this week may not close until next month. Use weekly+ windows; daily conversion rates are meaningless on long sales cycles. Won deals aren't only the window's leads, so the ratio can pass 100%. The line shows — then.


Questions about past dates (amoCRM, rolling out owner by owner)

Questions about a period can be answered from numbers we store per day, with no read from the CRM. Stored: leads, chats, won, revenue and lost, incoming and outgoing calls, leads reached by phone (by direction and in all), new leads reached, and qualified leads, for each pipeline. Numbers and manager or source ids only, never a lead's data.

  • Right after you connect amoCRM the last 30 days start loading, newest first, so last week is ready within minutes and last month soon after (longer on a very busy CRM). Older days, up to 90, load at night.
  • The answer says "as of" the last refresh. A past day can still change in the CRM (a chat moved out of Unsorted, a deal reopened, a backdated lead), so a stored week can differ from what that week's report said. The last 14 finished days are refreshed daily; older days stay as they were read.
  • Won and lost use today's status, by when the deal closed. Split by manager uses today's responsible user.
  • A period with days still loading is never added up. The bot reads the CRM live if it can, and otherwise says how many days are ready.
  • Reached is as of the last refresh: a lead from Oct 3 reached on Oct 9 counts on Oct 3. In a report it counts only if reached by the report's end.
  • Qualified counts a lead on each day it crosses the picked stage, so one that drops back and crosses again on another day counts twice in a stored period; a report counts it once. Changing the picked stage needs no reload.
  • Per salesperson (by the CRM user's id, names read live): leads and reached by the lead's responsible person now, and calls by who logged them, any length, as the Salespeople chart counts them. Lost leads by reason, by the CRM's reason id ("none" when none was given).
  • Calls are the whole portal's, as in reports. A portal with no call records shows "—", not 0.
  • Reports asked for in a chat ("report for last month") use the stored days too, when the period is over and every line the report shows is stored. The report then says "as of" the refresh. Otherwise it reads the CRM live, as before.
  • Cost per lead and cost per qualified in such a report are the period's own ad spend over the stored count, the same rule as a report's period. Cost per qualified follows the stored qualified count, so a lead that crossed the stage twice lowers it slightly. Sales by ad is read live, so those reports stay live.
  • Charts (Salespeople, Lost reasons) for finished days read the stored days too, and Forgotten leads reads last night's snapshot. Names are read live and never stored. A missing day or snapshot means the chart reads the CRM live.
  • Numbers not stored (contacted leads by stage, response time, open deals) aren't shown for that period, and the answer says so.
  • Forgotten leads are a nightly snapshot, in all, per pipeline and per salesperson: open leads from the last 7 days no one has worked on for your stale days (3 working days unless you changed it), as the neglect alert counts them, as of that night. A snapshot is never added up or refreshed: an answer takes the latest night's.
  • Open leads by age are a nightly snapshot too, for each pipeline: open leads by how long ago they came in (0–7, 7–30, 30–90 and 90+ days), how many of each have a task still due (an overdue one doesn't count), and leads still in Unsorted counted apart.
  • Before the last 90 days: the bot says it keeps 90 days.

Cross-source rates — need an ad source

These require ad spend data (Meta, Google) in the same job alongside the CRM source. The driver computes a rolling 7d/30d denominator to avoid cohort-mismatch bias.

Reports for a period you ask for in the chat ("make a report for October 1–20") are different: every rate here (CAC, CRM ROAS, conversion rate, cost per lead, cost per qualified lead) uses that period's own ad spend and that period's own CRM counts, with no rolling window and no "(7d rolling)" in the label, so the report adds up with itself. amoCRM today.

CAC — Customer Acquisition Cost (cac)

Definition: adSpend / dealsWon over the rolling window. Spend and sales cover the same days. In a report's first 30 days the spend is read from Meta; if that read fails, the line uses the report's own period, with no window in its label. Expressed in your ad-account currency. Available on: All three. Default: ✗ Optional. Best for: "How much do we pay to acquire each paying customer?" The headline marketing-efficiency number.

ROAS (CRM) (roasCrm)

Definition: revenueWon / adSpend over rolling window. Expressed as multiplier (e.g., "3.5x"). Available on: All three. Default: ✗ Optional. Best for: "For every $1 in ads, we earn $X in revenue." Compare across channels to find your highest-ROAS sources. Watch out for: when your CRM's currency differs from your ad account's (e.g. revenue in so'm, ads in dollars), ROAS shows "— (different currencies)". We don't convert currencies yet.

Cost per lead (costPerLead)

Definition: adSpend / leadsCreated over rolling window. Available on: Bitrix24, amoCRM, HubSpot. Default: ✗ Optional. Best for: Top-of-funnel cost efficiency. Cheap channels can still be bad if leads don't convert — pair with cac.

Cost per qualified lead (costPerQualifiedLead)

Definition: adSpend / qualityLeads over a 7-day rolling window. Spend and leads cover the same 7 days. In a report's first week the spend for those days is read from Meta; if that read fails, the line is left out for that report rather than shown on part of the spend. On amoCRM, qualified leads are counted exactly like qualityLeads: leads that first reached the stage you picked as qualified (in the wizard or in the bot's /crm questions) or any later stage. On HubSpot, qualified leads are contacts that entered the stage your /crm answer picked (SQL by default, or MQL). Available on: Bitrix24, amoCRM, HubSpot. Default: ✗ Optional. Best for: "How much do we pay per qualified lead?" Filters out noise from junk leads that inflate CPL but don't convert.


Sales by ad for instant-form leads (amoCRM, in testing)

What it answers: which Meta ad each instant-form lead, qualified lead and sale came from, and what each sale cost per ad. How it works: amoCRM's own Facebook integration names a form lead's deal "Facebook №<lead id>" (or the account keeps the id in a "Meta Lead ID" field). We read only that id, never the deal name's other text, and ask Meta which ad the lead came from (its ad, form and time; never the form answers). Only the exact "Facebook №<digits>" name or a lead-id field counts, and Meta must confirm the lead: a lead Meta can't confirm stays untraced, never guessed. In the report (Meta + amoCRM reports, accounts on the lead-access list): a "📍 Sales by ad" block, up to 5 ads from the report's own ad account(s), each as "leads → qualified → sales · cost per sale". Ads in your other ad accounts aren't listed:

  • leads: instant-form leads created in the period;
  • qualified: of those, the ones now at your qualified stage or later (or won);
  • sales: deals won in the period, whenever the lead came in;
  • cost per sale: the ad's spend in the period ÷ its sales. Leads are matched on every Facebook Page you manage; there's nothing to set up. Coverage: shown with the numbers ("Traced to an ad: 21 of 25 form leads (84%)"), counting only what the block is about: form leads traced to the report's ads, out of form leads not traced to another ad account. Website and call leads have no Meta lead id and are not in it. Watch out for: Meta serves a lead's ad for 90 days after it's created, so older leads can't be traced the first time. Not live yet: it waits for Meta's review of lead access, and runs only for accounts on the lead-access list.

Lead quality and sales by campaign and ad, in answers (amoCRM, in testing)

What it answers: questions under a Meta + amoCRM report like "which campaign brings the best leads?", "which ad sells?", "what does a sale cost?". How it works: the form leads Meta traced to an ad, counted as a cohort: the leads created in the window (the dates asked about, or the last 30 days to yesterday), read live from amoCRM when someone asks (your report's pipeline, Unsorted included), and what became of them so far:

  • qualified: crossed your qualified stage (the report's own rule, read from amoCRM's stage history until today), sit at or past it, or won. A lead that qualified and was lost later still counts;
  • sales: won by now, whenever, including deals won an hour ago. A recent window keeps selling, so its numbers grow;
  • costs: Meta's spend in the window per ad (ads that spent and brought no lead count too), summed per campaign, ÷ leads, qualified leads or sales.

The answer can point out what code found, such as cheap leads that rarely qualify, an ad with fewer leads but at least twice another's sales, the cheapest sale, or a big share of the spend with no sale. Each needs a minimum number of leads or sales. Only the report's own ad account(s). Coverage: campaigns and ads are ranked only when at least 70% of the window's form leads (every lead in it with a Meta lead id) trace to the report's ads. Leads Meta hasn't confirmed yet count against it, so a window we can't trace well gets "not enough traced" rather than costs that look too high. Below that, the answer says how many are traced. If the CRM data couldn't be read in full, the answer says the numbers may be a little low. Rankings by cost per sale need at least 2 sales, and by qualified share at least 10 leads. Watch out for: costs divide the ad's whole spend by its traced leads, and website or call leads aren't traced, so the 70% floor keeps this close. Nothing is stored: it's read when someone asks. Accounts on the lead-access list only.

Where the leads are now, by stage (amoCRM and Bitrix24; weekly and monthly reports)

What it answers: "of the leads we got last week, which stage is each one in today?" Today's rows count what happened in the period ("30 became qualified this week"). This block shows where one batch of leads is now. How it works: the leads created in the report's week or month, in the report's pipeline (Unsorted included). That's the same number as the report's "Leads created" row. They're read live from amoCRM when the report is built. Each lead counts once, at its stage now:

  • the stages use amoCRM's own column names, in pipeline order. Stages with no leads from the batch are left out;
  • ▸ Qualified and later: leads now at the stage picked in /crm or past it, won included. "Later" follows the board's order today, so the line sits right before the picked stage even if stages were moved after setup. A lead that qualified and was later lost counts under lost, not here. Pipelines with no qualified stage picked get no line;
  • Lost: the count, with amoCRM's loss reasons. The top 3 reasons are named, and the rest are added up as "other". A lost lead with no reason picked shows as "no reason given";
  • weekly reports add one line for the week before's leads: their qualified-and-later, won and lost counts now. Those leads are a week older, so they have had more time to move;
  • several pipelines: one block per pipeline that has leads from the batch, up to 2; the rest add up in one "Other pipelines" line. Each block names up to 6 open stages, the fullest ones, in pipeline order; the rest add up as "Other stages". Won and lost always show. Long names are cut at 40 characters, to keep the message within Telegram's limit;
  • by campaign (Meta + amoCRM reports, lead-access accounts only): the form leads Meta traced to an ad in the report's own ad account(s), per campaign. It's shown only when at least 70% of the period's form leads trace, the same gate as above. Below that, only the coverage line. In the rich Telegram report each block is collapsed under its title. Slack and plain Telegram show the same lines as text. If amoCRM's stages can't be read, the block is left out rather than showing stage ids. Daily reports don't get it, because a day's leads are too fresh. In answers: "where are last week's leads now?", "which campaign's leads are stuck?", "why did we lose them?". These use the same counts, for the dates asked about or the report's own period. A daily report uses the last 7 days. "Stuck" ranks campaigns by the share of their leads still open before the qualified stage, from 10 leads up. Watch out for: the counts are a snapshot at send time, so the same week's leads look different a week later. Nothing is stored, and no lead's name or phone is read into the report. HubSpot comes next. On Bitrix24: the same block, on the kanban your qualified rule runs on (set in the chat's /crm):
  • the Leads kanban for "lead reached a status" or "lead converted"; a Deals pipeline for "deal reached a stage". Without a rule it follows the portal's CRM mode: Leads in classic mode, Deals in simple mode (so a quiet week never switches the board);
  • the columns use Bitrix24's own names, in its SORT order. The success column counts as won (a lead's "converted"); failure columns (junk, lost, a portal's own) count as lost;
  • the line sits at your picked stage. With "lead converted", it sits at the success column. On Deals, only the rule's pipeline gets a line;
  • Bitrix24 keeps the reason a lead was lost as the failure column itself, so each failure column is a reason: "Lost: 7 (Junk lead 4, Too expensive 3)";
  • deals: one block per pipeline (your report's pipeline filter applies);
  • no per-campaign split: Bitrix24 leads carry no Meta lead id to trace. Read on the portal's own clock, live, counts only. No titles, names or contacts are read.

Where the leads leaked (Meta + amoCRM; weekly summary and answers)

What it answers: "leads but no sales: is it the ads or the sales team?" Each step of the week is compared with your own usual week, the middle of your previous 4–8 weeks. It's never compared with other businesses. The step that cost the most sales is named:

🔎 Where the leads leaked: qualification. 15 of 100 leads reached the qualified stage (15%), against 30% usually. That cost about 3 sales. The other steps were at their usual level. The steps:

  • ads: the cost of a Meta lead (lead campaigns);
  • arrival: Meta's leads that reached amoCRM, meaning leads carrying a Meta lead id, out of Meta's leads (at most 100%: CRM duplicates can push the count past Meta's);
  • contact: leads called in time. Not judged yet, because the CRM's calls aren't read reliably (amoCRM logs many calls on the contact). The summary says so and never says "nobody called";
  • qualification: leads that crossed your qualified stage in the week, out of the week's leads;
  • sale: sales won in the week, out of leads qualified in the week. Each is counted when it happens, as the report counts it, so a recent week isn't judged before its leads could sell. Lost sales: for each step, how many more sales the week would have had if only that step had been at its usual level, with the later steps at theirs. For the ads step, it's the extra spend divided by your usual cost of a sale. A step is a leak only from about 1 sale (or 10% of a usual week's sales); otherwise "no leak this week". When it says less: fewer than 30 leads in the week, or no sales in a usual week, gives "too few to say". With under 4 weeks of history, it only flags Meta leads not reaching the CRM (under 80%). In answers: "where is the problem?", "где теряем лиды?", "muammo qayerda?". These cover the same 7 days ending on the report's day (or on the asked dates), against the 8 weeks before. Watch out for: read live: one Meta call, plus amoCRM's leads and wins over the 9 weeks and its status history week by week. "Or later" follows the board's order today, as the report's qualified row does. If a week's status history hits amoCRM's read cap, qualification and sales aren't judged that time, and the summary says so. These weekly counts aren't the report's own rows: a lead that crosses the qualified stage twice counts once here. Nothing is stored. Bitrix24 comes later.

Data-quality notes in answers

When a question about the CRM is answered from data that isn't complete, the answer ends with one short note, counted by code, never estimated:

  • won deals with an amount (amoCRM): "based on 17 of 20 won deals with an amount (85%)", on revenue and average-deal questions;
  • leads with a source (amoCRM, when the report has Top sources): on questions about sources;
  • form leads traced to an ad (lead quality and sales by campaign and ad);
  • a read that hit a page limit: "the numbers may be a little low".

A note shows only below 90%, and only when the question is about that part. The 70% line for ranking campaigns and ads stays as it is. Counted from the leads the report already read, with no extra requests. Kept with the report from 24 Oct 2026; older reports have no notes.

"Based on" line. Every answer that gives numbers ends with one small line saying what it rests on, for example "Based on: 12–18 Oct · campaign "Umra" · form leads in amoCRM: 64 · traced to an ad: 82%". It shows:

  • the window of the numbers used: the asked dates, the report's own period, or the section's own window (lead quality: the last 30 days; where the leads are now: the cohort's dates);
  • the one campaign, ad or ad set the answer is about. With several, none is named;
  • the CRM counts behind it (form leads, or leads in the cohort), how many are traced to an ad, and which CRM it is ("Meta Ads data" when no CRM number was used). Code builds the line from the same numbers the answer used; the AI never writes it. Answers with no numbers (a greeting, "I can't tell from this report") have no line. Shown in Telegram, rich Telegram and Slack, in the answer's language. Long names are cut at 40 characters.

Aggregates — who & where?

Top managers (topManagers)

Definition: Top 3 responsible managers by won-deal count, with their deal counts and total revenue. Computed from:

  • amoCRM: groups won leads by responsible_user_id, resolves names via /api/v4/users.
  • Bitrix: groups by ASSIGNED_BY_ID, resolves via user.get.

Available on: amoCRM, Bitrix. Default: ✗ Optional. Best for: Recognition + diagnosis. "Anna closed 8 ($24k), Boris closed 5 ($18k), Carla closed 2 ($4k)." Renders as a single line.

Top sources (topSources)

Definition: Top 3 lead-source values by lead count. Computed from:

  • amoCRM: heuristic — scans created leads for the first custom field with code matching /utm_source|source|lead_source/i. If found, groups by that field's values. If not found, returns empty.
  • Bitrix: groups by SOURCE_ID, resolves names via crm.status.list.

Available on: amoCRM, Bitrix. Default: ✗ Optional. Best for: Identifying which channels drive volume. "Meta: 28, Google: 14, Referral: 6." Watch out for (amoCRM): Many amoCRM portals don't track source in a standard custom field. If topSources shows "—" (no data), your portal doesn't have a UTM/source field with a conventional code. Either rename your existing field or skip this metric.


Bitrix-specific lead funnel

Bitrix has a native lead workflow that amoCRM/HubSpot don't share: NEW → IN_PROCESS → PROCESSED → CONVERTED → JUNK. Two extra metrics are exposed for Bitrix because they describe Bitrix-native concepts.

Leads converted (leadsConverted)

Definition: Leads in Bitrix that were CONVERTED into deals during the window. Bitrix-only concept (Lead and Deal are separate tables; "convert" is the moment a lead becomes a deal record). Available on: Bitrix only. Default: ✗ Optional. Best for: Funnel breakdown. "Of 60 leads created, 25 contacted, 12 qualified, 8 converted to deals."

Leads junk (leadsJunk)

Definition: Leads marked as JUNK / spam / rejected. Available on: Bitrix only. Default: ✗ Optional. Best for: Lead-source quality signal (similar to amoCRM's notQualifiedLeads but stage-based, not phone-based).

Lead conversion rate (leadConversionRate)

Definition: leadsConverted / leadsCreated. Bitrix lead-funnel specific. Available on: Bitrix only. Default: ✗ Optional.


HubSpot-specific lifecycle stages

HubSpot has a standard contact lifecycle (subscriber → lead → MQL → SQL → opportunity → customer) that's universal across portals. Three metrics surface contact movement through this funnel — they count contacts that ENTERED each stage during the report window.

These metrics live on Contacts (the lifecycle stage is a contact property), parallel to the Deal flow metrics (which live on Deals). HubSpot users typically read both funnels — top of funnel via these lifecycle stages, bottom via Deals created/won/lost.

MQL contacts (mqlContacts)

Definition: Contacts whose hs_lifecyclestage_marketingqualifiedlead_date falls in the window. The moment a contact reached the Marketing Qualified Lead stage. Computed from: HubSpot /crm/v3/objects/contacts/search filtered by the per-stage date property BETWEEN window bounds. Available on: HubSpot only. Default: ✓ Yes. Best for: Marketing-to-sales handoff health. High contactsCreated with low mqlContacts = your sales team isn't surfacing contacts to marketing-qualified-lead status fast enough (or your lifecycle stage discipline is loose).

SQL contacts (sqlContacts)

Definition: Contacts whose hs_lifecyclestage_salesqualifiedlead_date falls in the window. Contacts the sales team has accepted as qualified for direct outreach. Available on: HubSpot only. Default: ✓ Yes. Best for: Sales pipeline-filling rate. Falling SQL counts with flat MQL counts = SDRs rejecting marketing's leads (could be a lead- quality conversation or a sales-team-saying-no conversation).

Opportunity contacts (opportunityContacts)

Definition: Contacts whose hs_lifecyclestage_opportunity_date falls in the window. The contact has been promoted to "open business opportunity." Available on: HubSpot only. Default: ✗ Optional. Often unused in CIS/SMB HubSpot setups where the Opportunity stage feels redundant with Deal creation. Best for: Enterprise B2B teams who treat Opportunity as a distinct qualification gate before opening a Deal.

If HubSpot can't be read, the MQL, SQL and Opportunity rows show "—", never 0.

Cost per SQL (costPerSql)

Definition: Ad spend in the report's window ÷ contacts that entered SQL in the same window. Available on: HubSpot only, in a report with an ad account. Default: ✗ Optional; in the Media buyer and Executive presets. Best for: What a sales-qualified lead costs you, in HubSpot's own word for "qualified". Watch out for: The SQLs are all of the portal's, not only the ones from your ads. No row when there's no spend or no SQL; "—" when HubSpot's SQL count couldn't be read.

Cost per MQL (costPerMql)

Definition: Ad spend in the report's window ÷ contacts that entered MQL in the same window. Available on: HubSpot only, in a report with an ad account. Default: ✗ Optional. A report shows it in place of cost per SQL when your answer to "what counts as qualified" in /crm is MQL, and goes back if you change the answer. Picked by hand next to cost per SQL, both show. Watch out for: The same as cost per SQL.

Qualified on HubSpot (rolling cost per qualified lead)

HubSpot's "qualified" is a contact entering the stage you picked in /crm: SQL by default, or MQL. The rolling cost per qualified lead uses it, and conversion rate uses contacts created. A window HubSpot couldn't count is left out, never counted as 0.


Quick-start presets

The wizard offers role-based presets at Step 3 — click a card to preselect the metrics that role typically reads, then tweak from there. Manual checkbox edits switch the active card to "Custom."

Each platform has three presets. Sales manager carries the ⭐ Recommended badge — first-time users land on it automatically when the CRM source is added.

amoCRM presets

PresetMetrics in order
⭐ Sales managerleadsCreated, leadsContactedOutbound, qualityLeads, dealsWon (Leads won), revenueWon, incomingCalls, timeToFirstResponse, topManagers
Media buyerleadsCreated, leadsContactedInbound, leadsContactedOutbound, qualityLeads, notQualifiedLeads, dealsWon, revenueWon, topSources
ExecutiveleadsCreated, leadsContactedAnyPhone, dealsWon, revenueWon, notQualifiedLeads, timeToFirstResponse

amoCRM Sales manager intentionally drops the stage-based contactedLeads — it's too dependent on per-portal stage mapping + team discipline. The phone-based leadsContactedOutbound is system-recorded (call duration), reliable on every portal.

Bitrix presets

Mirror amoCRM with Bitrix-native IDs (leadsContacted and leadsQualified instead of contactedLeads/qualityLeads, since Bitrix has these natively via STATUS_SEMANTIC_ID without any per-portal stage mapping).

PresetMetrics in order
⭐ Sales managerleadsCreated, leadsContactedOutbound, leadsQualified, dealsWon, revenueWon, incomingCalls, timeToFirstResponse, topManagers
Media buyerleadsCreated, leadsContactedInbound, leadsContactedOutbound, leadsQualified, notQualifiedLeads, dealsWon, revenueWon, topSources
ExecutiveleadsCreated, leadsContactedAnyPhone, dealsWon, revenueWon, notQualifiedLeads, timeToFirstResponse

HubSpot presets

Use HubSpot-native vocabulary (MQL/SQL) instead of unified Contacted/Quality terms — HubSpot users speak that language. Phone- contact and loss-quality split are deferred for HubSpot, so the presets use raw call counts + lifecycle stages instead.

PresetMetrics in order
⭐ Sales managercontactsCreated, mqlContacts, sqlContacts, dealsCreated, dealsWon, revenueWon, outgoingCalls, incomingCalls
Media buyercontactsCreated, mqlContacts, sqlContacts, costPerSql, dealsWon, revenueWon
ExecutivecontactsCreated, sqlContacts, dealsWon, revenueWon, costPerSql

Cost per SQL shows only in a report with an ad account. Response time returns to the HubSpot presets once it has a HubSpot source.

How metric order in the delivered report works

The order metrics appear in your Telegram/Slack/Email message follows the order they were added in the wizard:

  • Picked a preset? The metric order is exactly the order shown in the table above (each preset is designed as a coherent funnel narrative).
  • Manually checked boxes? Metrics appear in click order — the first one you checked is the first row in the report.
  • Mix of both? The preset's metrics keep their order; any extra metrics you check afterward append to the end.

This matches the Meta Ads wizard's behavior. The delivered report respects your intent, not a hard-coded sequence.


How to combine metrics — playbook by role

Most users won't need this section — the wizard now ships role- based presets (Sales manager / Media buyer / Executive) that implement these guidelines automatically. This section explains the reasoning so you can build your own custom mix if a preset doesn't quite fit.

Media buyer (performance marketer)

Picks: spend, cpm, cpl, leadsCreated, qualityLeads (or mqlContacts on HubSpot), leadsContactedInbound, conversionRate, roasCrm. Hides: outbound metrics, top managers, time-to-first-response.

Head of sales / SDR manager

Picks: leadsCreated, contactedLeads, qualityLeads, dealsWon, dealsLost, timeToFirstResponse, winRate, topManagers, outgoingCalls, leadsContactedOutbound, lostQualifiedLeads, notQualifiedLeads. Hides: ad-spend granular metrics, search impression share.

Sales rep / AE

Picks: leadsContactedOutbound, qualityLeads, dealsWon, revenueWon, timeToFirstResponse. Hides: cross-team aggregates, ad metrics.

Exec / CEO

Picks: spend, leadsCreated, dealsWon, revenueWon, leadsContactedAnyPhone. Total: 5 numbers max. Headline only.


Lead waiting alert (amoCRM, Bitrix24)

Not a report line: one message, at most every 2 hours per chat, when new leads have had no answer for 15 working minutes (5 in the US and western Europe). It goes to the owner only, unless the owner chose Also the group or Off.

The first one asks whether to keep sending. Until the owner says yes, no more go out, and that covers the count of chats and no-phone leads too. "Leads not moving" asks the same on its own. Private alerts have a button to turn them off, and the owner can also write in the private chat to stop or restart them.

  • Answered means the team reached for the lead after it came in:
    • an outgoing call (even unanswered: they tried), or an incoming call that connected, saved on the lead, its contact or its company (including the call that created the lead, logged up to 5 minutes before it);
    • on amoCRM, a person's stage move, note, task, a call they took, or a card field they edited, including what they did as they made the lead themselves (amoCRM's own automation doesn't count; assigning the lead to someone doesn't either; nor does a move into the pipeline's first working stage, which can be sorting rather than contact);
    • on amoCRM, a chat reply on the lead or its contact, a minute or more after the lead came in. amoCRM doesn't say who sent a chat message, so this is a best guess: auto-replies come within seconds, but a bot on a longer delay would count as a reply;
    • on Bitrix24, an outgoing email or message on the lead or deal, or its contact, a minute or more after it came in (robots send SMS in the first minute, under the managers' names), or the card moved past the first stage. A robot's later SMS, or a robot's stage move, would still count.
  • Won or lost leads never wait.
  • Working hours are learned from when the team calls (4 weeks, per weekday; the last week on very busy portals). A quiet weekday gets the team's usual hours; only a day with almost no calls is closed. Until there are enough calls: Monday to Saturday, 09:00–19:00. The clock only runs in working hours.
  • Leads that come in after hours are due an hour after opening.
  • The CRM is read again as the message goes, so a lead handled by then isn't named. Each lead gets its link once per chat. The message shows the oldest 15; the rest are in its count and get their link in the next message.
  • Several reports going to one chat share one message, with each lead once and labelled by report and CRM. Only me is the owner's one private chat. With Also the group, each chat gets its own message on its own 2 hours.
  • The owner's count of inquiries is separate. A grouped message waits 30 minutes after it, so the two don't arrive together.
  • A reply under the message asking about a lead's stage lists the stages of the leads shown, in order.
  • The message carries the time the lead came in, the wait, the manager and a link to the card, never the lead's name or phone.

Conversions API readiness check (amoCRM)

Not a report line: tells an owner whether sending CRM stages back to Meta would help, by Meta's own Conversion Leads rules (about 200 instant-form leads a month, a stage that 1–40% of leads reach within 28 days).

  • Leads, last 30 days: all of them, and the instant-form ones (Meta's lead id in the deal name or a field).
  • Missing lead id: of the deals named for Facebook / Instagram, the share without the id. A floor: a form lead with nothing pointing to Meta can't be seen.
  • Stage rate: leads of the picked stage's pipeline created 29–58 days ago (each had its full 28 days), the share that reached the picked qualified stage or a later one within 28 days of coming in.
  • Verdict, first that applies: no stage picked, mostly not form leads (under 20%), the id missing on 30% or more, stage under 1% or over 40%, fit (200+ a month), almost (100–199), low volume (under 100).
  • Every lead of the last 30 days is counted. When a read still hits amoCRM's limit, the numbers are shown without a verdict.
  • Read-only. The verdict and the counts are logged; never a lead's name.

CRM neglect alert (amoCRM, Bitrix24)

Not a report line: one message in the report's chat when the CRM is going stale.

  • Which leads: created in the last 7 days, still open, in the report's pipeline.
  • Stale: no stage move, and no call, note or reply in the CRM by a person on the lead or its contact, for 3 working days (closed days, learned from the team's calls, don't count). A lead nobody touched counts from when it came in.
  • When it says so: 5 stale leads or more, and at least a quarter of the open ones. Once it has said so, it counts as still stale until under 4 and a fifth, so a CRM sitting at the line doesn't flip each day. Checked once a working day, two hours after opening. It is sent when things cross that line, then again only after a week if they stay stale: never a daily repeat.
  • The message has the counts, the counts per manager and the three longest-waiting leads with a link to the card, never a lead's name.
  • Inquiries (see the lead waiting alert) are left out of this alert: its counts, its list and whether it fires. The owner's morning count covers them for their first 3 days.

Update log

One file per change in docs/crm-metrics/log/, named <YYYY-MM-DD>-<pr-or-slug>.md, so the file names sort by date (newest last in a plain listing). Each file is one short entry.


How to update this doc

When you change lib/crm-metric-catalog.ts, lib/sources/amocrm.ts, lib/sources/bitrix.ts, or lib/sources/hubspot.ts:

  1. Find the section above that describes the affected metric
  2. Update the definition, computation, availability, default state, and use-case prose to match the new code
  3. If adding a metric: pick the right section (Volume / Money / Telephony / etc.) and add it in the same shape as siblings
  4. If removing a metric: delete its block (don't leave dead docs)
  5. Add a log file docs/crm-metrics/log/<YYYY-MM-DD>-<pr-or-slug>.md (one short entry: what changed, with the PR number); never edit a shared list
  6. Commit doc + code together — they're a single unit of change