CRM deals, stage history and users; ERP orders and invoices
Sales pipelines only, win by semantics, real owner and bridge to the invoice
Funnel, conversion, invoiced value and response time by salesperson and store
Daily store card, dashboard and AI agent
What it solves
- Each salesperson's conversion counts as a win everything that was sold, not only the deals that went through the satisfaction survey.
- After-sales, delivery, exchanges and collections leave the sales funnel and stop distorting the win rate.
- Each salesperson's value is what became an invoice in the ERP, linked to the CRM deal.
Business questions it answers
Ask your agent these questions once the case is live.
What you need
- CRM connected to Nekt with deals, stage history, stages and users
- ERP connected with orders and invoices, and a field that links the order to the deal
- The list of sales pipelines, win stages and system accounts, agreed with management
What you get
- A table per deal with pipeline, stage, category, owner, store and dates
- Indicators by salesperson and month: opportunities, wins, conversion, invoiced value and ticket
- Daily store card with yesterday's sales and deals in negotiation by salesperson
- An AI agent that answers about the team with the right win rule
Where to activate
Every morning, on WhatsApp or in the CRM chat: yesterday's sales and quotes by store and deals in negotiation by salesperson, with the time of the CRM snapshot.
Funnel by stage, conversion, invoiced value and response time by salesperson, SDR and store, with loss reasons.
Answers about the team with the win rule, the right pipelines and revenue coming from the ERP.
Mapping pipelines and stages is quick. The bridge between the CRM owner and the ERP invoice depends on a field linking the two, which sometimes has to be created. The detailed view has the business rules, the pitfalls and the acceptance test. Or send everything straight to your agent with the Nekt MCP.
The retail CRM mixes sales, after-sales, collections and partner relationship pipelines in the same place. The final stage of one pipeline may be the satisfaction survey, and of another, the closure of the SDR's service. The deal value is typed by the salesperson, and the ERP, which has the sale, sometimes returns a generic salesperson through the API. Whoever counts the last stage as a win, or adds all pipelines together, delivers rates of 20% or 97% for the same team.
How it is usually done, and the rule
Each row is a mistake that already happened in practice.
Classify every stage of every pipeline in a reference table: in progress, won, lost or postponed. In pipelines where the deal goes through won to be invoiced, won invoiced and then a final satisfaction survey stage, all three count as a commercial win. The CRM semantics help (in Bitrix24, STAGE_SEMANTIC_ID S for won, F for lost, P for in progress), but check the name: a final stage is not always a sale.
Filter by pipeline (in Bitrix24, CATEGORY_ID). After-sales and service pipelines (exchange, delivery, complaint, warranty), scheduling, collections and partner programs are not sales and stay out of conversion, even if they have success stages. A deal created directly in an operational pipeline is not a lead.
Before modeling, measure fill rates: in many operations the service lives in deals, and the leads entity has abandoned SDR and status fields, filled close to zero. The custom SDR field may also be empty; in that case the owner is ASSIGNED_BY_ID. Write the choice in the context document.
When the lead enters an SDR pipeline and, once qualified, is transferred to a sales pipeline, it is the same deal. SDR origin = the deal has any SDR pipeline stage in its history. Handoff = it is in or went through the sales pipeline coming from the SDR. The SDR's final stage (lead closure) is not a sale: the sale is measured in the sales pipeline. Count each deal once in the total.
The salesperson is the deal owner in the CRM, resolved through the users table, without grouping accounts into names. Administrator, marketing, performance, integration and design accounts stay out of the ranking through a list of ids maintained by management; their deals appear in the total as System. If the CRM does not export names, keep an id-to-name mapping.
The CRM is the source of opportunities, stages, origin and activity. The ERP is the source of realized sales. Link the won deal to the order or invoice through an explicit field (deal id saved on the order, quote or cart number), using the composite store and document key. The deal value in the CRM stays as a reference; invoiced value is the sum of the linked sales invoices. A win without an invoice after the agreed delivery period becomes a pending item.
Keep a table of CRM id, ERP salesperson, store, start and end. Some ERPs return, through the API, a generic salesperson on every order: then the CRM owner is the only source of who sold. Without a registry, the month's assignment can be inferred as the store where the salesperson invoiced the most that month. A month with no sales in a store does not mean vacation or dismissal.
First response time = from entry into the pipeline (first record in the stage history) to the first move out of the entry stages, counting only the declared business hours (for example, 8 am to 6 pm, Monday to Friday, America/Sao_Paulo). Convert times once: Bitrix24 stage history comes in UTC and open channel sessions come in UTC+3. Report the median, because the average rises with leads that arrive at night.
Incoming opportunities and conversion are counted by the deal's creation date (DATE_CREATE in America/Sao_Paulo), which matches the CRM export. The entry date in the history differs for some deals and creates daily differences. Recent cohorts are still open: show the current month's conversion as partial.
In WhatsApp integrations with the CRM, the customer's incoming message becomes an activity, but the reply the team sends through the chat does not. Automatic tasks also become activities. Activity counts measure incoming demand; the effort signal is first response time and stage progress.
How each stage enters the salesperson funnel
Reference table by pipeline and stage, maintained with sales management.
Definitions
Pitfalls that already happened
The win rate used only the final satisfaction survey stage, which was rarely filled in. Adding won to be invoiced and won invoiced, real conversion for the period was close to 52%.
Classify all sales stages as won in the reference table.
About 97% of SDR pipeline deals ended in the stage marked as won, which was the closure of the service, and the SDR showed up as the company's best salesperson.
Measure the SDR by response, qualification and handoff; wins only in the sales pipeline.
Administrator, marketing and integration accounts owned thousands of imported deals and showed up ahead of the salespeople.
List of system ids out of the ranking and in the total as System.
An old rule grouped several CRM accounts into a few SDR names. The distribution was distorted and part of the deals ended up with no owner.
Use the real owner by id; grouping only as an extra view, if management asks.
The ERP returned through the API the same salesperson on every order of a chain, and the revenue ranking put all sales under one name.
Take the salesperson from the deal owner in the CRM and link the order to the deal.
Almost all activities were incoming, because chat replies did not generate records, and the dashboard concluded that the team did no outbound contact.
Use first response time and stage progress as the effort signal.
Acceptance test
The case is done when every item passes.
- For a closed period, opportunities created by day, owner and stage match the CRM export one by one.
- No deal from after-sales, collections or partner pipelines appears in the sales funnel, and no system account appears in the ranking.
- Commercial wins = sum of stages classified as won, and the percentage of wins with an invoice is reported next to it.
- For a store and a month, the sum of the salespeople's invoiced value plus the Not linked to CRM row equals the store's net sales in the ERP.
- The first response time of a deal that came in on Friday at 5:50 pm and was moved on Monday at 8:10 am is 20 minutes.
Agent package
Paste into Claude, ChatGPT or any client with the Nekt MCP connected.
Salesperson funnel in the CRM (retail) - Sales pipelines (CATEGORY_ID): <list>. After-sales, delivery, exchange, collections, scheduling and partners stay out of conversion. - Commercial win = stages classified as won in the reference table (won to be invoiced, won invoiced and finished after survey). Never only the last stage. - Lead closure in the SDR pipeline is not a sale. The SDR is measured by response, qualification and handoff. - SDR origin = any SDR pipeline stage in the history. Handoff rate = handed off / SDR origin. - Salesperson = deal owner (ASSIGNED_BY_ID). System accounts <ids> stay out of the ranking. - Opportunities and conversion by creation date (DATE_CREATE, America/Sao_Paulo). The current month's cohort is partial. - Invoiced value comes from the ERP: sales invoices linked to the won deal, composite key store + document. The CRM deal value is only a reference. - First response time: median in business hours <hours>, from the first record in the history to the first exit from the entry stages. - Activity count measures incoming demand, not team effort. - The salesperson x store link has validity dates; never apply the current store to the past. Client exceptions: - <client>: win stages of pipeline <id> = <to be invoiced stage>, <invoiced stage>, <finished stage>. - <client>: ERP returns a generic salesperson; the salesperson comes only from the CRM. - <client>: SDR pipeline <id> hands off to pipelines <ids>.
Build the "Salesperson funnel in the CRM" case in my Nekt workspace following the rules at https://academy.nekt.com/en/industrias/varejo/funil-vendedor-crm.md. 1) Find the source tables and confirm the columns the rules use. 2) Create a transformation at the stated grain applying every rule. 3) Run the acceptance test and show me the result of each item. 4) Create the case's context document in the workspace, leaving my clients' exceptions open for me to fill in.
Variations
At car dealerships, the funnel lives in the management system: proposals, approved, approved by financing, sold, canceled and returned by salesperson and month, with average discount per proposal.
When the person who refers (painter, architect, installer) is not the buyer, the sale is attributed to the buyer and to the commissioned professional in separate columns, matching the CRM record with the ERP.
One row per conversation session with routed channel, operator, time to acceptance and rating, for service SLA and productivity.