source: .forge/discovery/meeting-notes/2026-09-09-requirements-followup.md · generated — edit the markdown, not this file

Empire Teas CRM — Requirements Follow-up Meeting Notes

Date: 9 September 2026 Meeting: Discussion — requirements follow-up: leads and follow-up, the trade-show capture app, weekly bulk pricing Attendees: Tharinda Rajapaksha, Dinil Gamage, Biseni Perera, Thilina Abeyrathna, Rukshan Salgado (Altrium); Kavindu Anthony, Praveen, Sajeev (Empire Teas) Source: in-meeting notes taken by Biseni, reconstructed and confirmed with Dinil afterwards. Empire shared their weekly price workbook (Weekly Auction Prices 35.xlsx) after the meeting; it was inspected to describe the pricing section accurately. Cleaned and reproduced here as the discovery record.

Why this file exists: this is the second requirements session after kickoff, following the question set prepared for 3 September (2026-09-03-requirements-gathering-questions.md). It gives partial answers to issued questions Q6 (shared inbox), Q7 (deals going quiet) and Q9 (capture-app scope), and to prepared questions Q21 (badge QR) and Q33 (Buying Standard). Folded into the PRD as Rev 23 (2026-09-11), with OQ-52/OQ-55 updated and OQ-57–OQ-60 raised. Three decisions taken after the meeting are recorded there and supersede the corresponding lines below: one price per standard (the workbook’s cross-tab differences are data drift, not a rule); bulk/group email, newsletters and WhatsApp integration are out of V1 (team decision, Tharinda, 2026-09-11), so the “price circular fits F20” and “bulk email versus F28” readings below are overtaken — the weekly prices go out per account from a send this week’s prices action on the account page; and the role-management refinement (no templates, one role per user, AI not a grantable level) lands as AD #32 in its own PR. The What this changes in the harness list at the end is the record of what was owed at filing time.


Summary

Empire’s core problem is follow-up, not stage movement: leads and deals go quiet and sit idle in a stage with nothing prompting action. Everything else they asked for hangs off that — automated follow-up, tags for grouping, bulk mail-outs (the weekly price circular and newsletters), WhatsApp alongside email, sharing a lead with a colleague, and grouping several companies under one lead when an existing customer brings agencies into a project.

For the capture app they confirmed badge QR scanning, and asked for an existing-client check during the scan, a same-day generic email to stand visitors, and a ranking of the existing backlog of stuck leads.

For weekly bulk pricing they described the mechanics: each customer has their own standards with unique, never-reused numbers (about 35,000 issued to date); the price is set per standard, every week, in an Excel workbook that circulates by email; the system should read that workbook; and the published prices are emailed to each customer’s contacts. The workbook they shared confirms the shape and raises two housekeeping points to settle with the Tea Room (below).

Action items

  • Altrium — walk the shared workbook with the Tea Room and settle the two To confirm points under Weekly bulk pricing before the F26 spec is drafted.
  • Altrium — ask Empire for a photo of a badge from a recent show (what the QR encodes), and confirm whether info@ is a shared mailbox or a distribution list.
  • Altrium management — WhatsApp and a newsletter tool are outside the accepted scope; decide how to answer Empire (change request or a clear no). The weekly price emails are inside scope.
  • Harness — apply the follow-ups listed at the end of this file.

Details

Leads, inbound email and follow-up

Following up is the core problem. Cold leads and idle deals have no follow-up mechanism today. The pain is not moving deals stage to stage; it is items sitting untouched in a stage. Empire wants follow-up automated — a reminder, a drafted email, or a scheduled sequence. (F18 stale alerts, F25 outreach sequences, F3 reminders. A person still approves every send.)

Which synced emails become leads. Empire asked how the CRM will tell an inquiry from ordinary correspondence once it syncs everyone’s mailbox. Our answer: synced mail links to known contacts and deals automatically; a rep or the shared-inbox triager promotes an email to a lead with one action; the AI extracts the fields for confirmation. Fully automatic lead creation from inbound mail is a later-phase item in the accepted proposal. (F20, F12, F11)

The same inquiry arrives twice. The first email often lands in both the shared info@ inbox and a rep’s own inbox. If it is addressed to a named person it is that person’s lead; either way it must become one lead, not two. (F11 duplicate detection, F20 sync. Note: “lands in both” is how a distribution list behaves — confirm whether info@ is a mailbox or a list.)

Region-based routing of shared-inbox inquiries. An inquiry to info@ that concerns a region goes to the person in charge of that region. This is a fixed region → responsible rep mapping. (Partial answer to issued Q6.)

Several companies on one lead. An existing customer starts a new project and brings agencies in as stakeholders. Several organisations and people, all on the same email thread, are then connected to one lead, and Empire wants them grouped under it. The parties can be read from the email itself. Empire said this still “needs thinking about”. (Today a deal belongs to one account; this needs a participants concept before the F1/F2 specs.)

Sharing a lead or deal. The owner can share a lead or deal with a colleague, and that colleague then sees everything on it. (A per-record sharing path, in addition to ownership and the reporting line under AD #30.)

Tags. Tags on leads, deals, contacts and organisations, for grouping and filtering. (F1, F2, F32)

Bulk email and newsletters. Send email in bulk from the CRM, newsletter-style; upload a prepared newsletter into the CRM to send it; and send the weekly prices to customers (see pricing below). (The weekly price circular fits F20’s group email and templates. A newsletter tool with consent and unsubscribe handling is F28, deferred from V1.)

WhatsApp alongside email. Empire wants WhatsApp as a channel as well as email. What exactly for — messaging buyers, notifying reps, or logging existing threads — was not said. (Outside the accepted scope: the proposal names WhatsApp only as where information lives today, and the capture app’s notifications are email-only per AD #20.)

The trade-show capture app

Badge QR scanning is needed. Confirmed. What the QR encodes is still unknown; ask for a badge photo. (Q21 answered yes; net-new build in F36.)

Existing-client check during the scan. When a card or badge is scanned, the app must immediately tell the rep whether this company or person is already a customer, and how often Empire deals with them. This happens in the capture flow, not after upload. (Needs a live lookup against accounts from the app; degrade gracefully when the hall has no signal.)

A generic email on the day of the show. After meeting someone at the stand, send them a generic templated email the same day. (New. Outward, buyer-facing mail — decide whether it goes from the rep’s own mailbox or a generic address, and how the rep approves it.)

Rank the backlog of stuck leads. Rank the existing backlog so the team knows which to chase first. (F11’s batch scoring of the Pipedrive backlog.)

Campaign planning was mentioned but is already present in the prototype app, so it is not a new ask.

Weekly bulk pricing, standards and deals

Standards are per customer, and the price is set per standard. Each customer has their own buying standards, each with a unique running number that is never reused. About 35,000 numbers have been issued to date. The price is set for the standard, not for a catalog product. (Confirms OQ-42 and the settled data model.)

The weekly price list is an Excel workbook that arrives by email; the system should read it. Prices change every week. Empire wants the CRM to take the workbook in rather than have prices keyed by hand. (F26 draft → publish, plus the F23 import engine.)

Prices are emailed to customers. After the week’s prices are set, each customer is emailed their prices, using the contacts on that customer’s account record. (F26 + F20 group email — the “price circular”.)

One customer, several products, one standard per product. A customer buys several products; each of those products maps to exactly one standard.

A deal has several samples. (Confirms F6 multi-round sampling.)

A deal or quote must always have an active owner. When someone leaves, their records are reassigned to an active user. (Answers the offboarding default in the 3 September set, Q11.)

What the shared workbook shows (Weekly Auction Prices 35.xlsx — “35” is most likely the auction sale number):

  • Ten tabs, one per market, most named after a destination country (Turkey, Russia, Iran, Syria, Iraq, Hong Kong). One row per standard, columns CLIENT · STD · TYPE · GRADE, then one column per weekly auction date (18 Aug, 25 Aug, 2 Sep, 9 Sep in most tabs). About 1,450 distinct standard numbers in this file, running up to 31,422.
  • TYPE appears to be the blend recipe — component codes with percentages, e.g. 55R : 10%, 57 : 18%, 58 : 10%. GRADE is the tea grade; PEKOE dominates, then FBOP, OPA, FBOPF1, OP1, BOP1, DUST.
  • Not every standard is priced every week: about half the week cells are blank or -.
  • The weekly numbers are the price quoted to the customer for that standard (confirmed by Dinil). They carry no unit or currency label; one tab converts them to USD at 327.5 rupees per dollar, so they appear to be in rupees per kg. Currency and unit to confirm.
  • The same standard number appears in more than one tab in about thirty cases, and this looks like a data-quality issue rather than a pricing rule. In about half of them the two rows carry the same recipe and grade, so one standard was entered twice and its prices have drifted apart by a small amount (10 to 40 in the same week). In the other half the two rows carry different recipes or grades, so one number is standing for two different teas (one such pair differs by 150 to 200). Empire were clear that a standard has one price, and the workbook does not contradict that. The import must detect both patterns instead of silently taking one row.
  • It is a hand-maintained sheet: CLIENT is filled only on the first row of a group and carries working notes (“No Sample”, “CHECK & UPDATE”); one tab has a shifted header; one tab has no week columns; the IRAQ tab still shows March dates; there are a few #REF! errors and some text in price cells. Reading it automatically means either tolerating this shape or agreeing a template with the Tea Room.

To confirm with the Tea Room:

  1. The currency and unit of the weekly numbers (rupees per kg?) and how the USD figure is derived. Empire told us in August that published prices are FOB Colombo in USD.
  2. The duplicated standard numbers described above: where two rows share a number, which row is right, and whether very short numbers such as 3 and 73 are real standards or placeholders. A ten-minute walk of the list with the sheet open.

What this changes in the harness (follow-ups, not yet applied)

  • Resolve OQ-38 as one price per standard — Empire were clear on 9 September, and the workbook’s cross-tab differences are data drift, not a rule (decision: Dinil, 2026-09-11). The August answer that price differs by destination market becomes a fact about which customers sit in which market, not a pricing axis. Consequences to carry through: the price grain in forge-et-be/docs/data-model.md §2.2 drops market (owner: Tharinda, AD #31); F7’s “unpriced market blocks quoting” rule becomes “unpriced standard blocks quoting”; the prototype’s per-market price board (harness PR #63) needs the same change. Record it as a PRD revision via the three-writes procedure in .claude/rules/prd.md.
  • New open questions to log: several companies on one lead (participants); owner-initiated sharing of a lead or deal; region → rep routing for info@; one lead from a message that reaches two mailboxes; reading the weekly workbook into draft prices; the weekly price circular to account contacts; existing-client check and interaction frequency during a scan; the same-day show email; WhatsApp as a channel (change-request candidate); bulk email versus F28.
  • F36 grows: QR scanning, the scan-time client check and the same-day email are more than F36 as written — re-run the R-ARCH-001 band.
  • F25’s Phase-2 placement deserves a look against the stated core problem.
  • The prototype is being reworked on this branch to show each customer’s standards and prices on the account page, with a Tea Room pricing screen; the workbook’s shape (one row per standard per week, market or customer-group tabs) is input to that work.
  • Issued question list (client/open-questions.md): Q6 and Q7 partially answered; Q9 partially answered.

Two fragments in the raw notes were set aside on Dinil’s confirmation: “quotes can be turned into multiple deals” (not meaningful) and “planning the campaign” (already present in the app).

Generated from .forge/discovery/meeting-notes/2026-09-09-requirements-followup.md by .claude/lib/render-docs-html.js. Markdown is canonical.