Sales
Sheet 4/12 · Medium · 3–10 weeksQuotes take two days, follow-ups depend on memory, and the pipeline is a group chat.
- A quote needs the owner, the price list and a spare hour; customers buy elsewhere in the meantime.
- The pipeline is in a WhatsApp group; "did anyone call Mr Moyo back?" is a real question.
- Prices change with the exchange rate and quotes go out at last month's numbers.
- Salespeople spend evenings typing what they said into the CRM — or do not.
- Nobody knows which quotes are open, which were won, and why the rest were lost.
- Produce a quote in the customer's WhatsApp thread from the current price list in a minute, for a person to approve.
- Log calls and messages into the CRM automatically and draft the follow-up.
- Remind, chase and re-quote on a schedule you set — and stop when the customer buys or declines.
- Show the pipeline honestly because the data is captured at source, not retyped.
- Negotiate. Anything with discretion, relationship or risk stays with a person.
- Issue a fiscal tax invoice — that must go through your fiscal device and ZIMRA's FDMS. A quotation is not an invoice.
- Make a price list current if nobody maintains it; the AI quotes what it is given.
- Replace the CRM you do not have; it fills one, it does not become one.
| Effort band | Medium · 3–10 weeks |
|---|---|
| Data you need | A maintained price list; discount rules; a CRM or a contacts sheet; sample quotes |
| Integration points | WhatsApp Business Platform; CRM; price list source; (invoice: fiscal device via S-04) |
| People | Sales lead who owns rules; salespeople who approve drafts |
| Measure | Quote turnaround; follow-ups completed; open quotes visible; win rate |
| Zimbabwe constraint | Dual-currency pricing; quotations vs FDMS fiscal invoices; WhatsApp as the deal channel |
| Code | Shape | Fit | Why |
|---|---|---|---|
| S-04 | Integration | Primary | Price list + CRM + WhatsApp connected is most of the value; the AI is a thin layer on top. |
| Tool | Listed platform with quote/invoice generator and CRM tags | Primary | For a small team, a subscription that generates quotes from a catalogue and keeps contacts tagged covers steps one and two. |
| S-02 | Pilot | Secondary | Run quoting and follow-up for one product line for six weeks; measure quote turnaround and follow-ups completed. |
| S-03 | Implementation | Secondary | Approval rules, discount limits, multi-currency quotes and sequences built around your process. |
| S-06 | Managed support | Later | Price lists and sequences change monthly; someone must own them. |
- Where does the price list live, who updates it, and how does a USD/ZiG rate change flow into quotes?
- Show a quote produced in a WhatsApp thread, approved by a salesperson, and logged in the CRM.
- What stops the system quoting below our floor price or promising stock we do not have?
- How are quotes kept separate from fiscal invoices, and who issues the invoice through the fiscal device?
- Which CRM do you support with an official connection, and what happens to our data if we change CRM?
- How does a salesperson correct a wrong AI draft, and does the correction stick?
The full bank of 60 questions, grouped by risk, is at /vendor-questions/.
Quote, then invoice — two different machines
A sales assistant can draft and send a quotation instantly. It must not send an invoice: since 31 May 2025, ZIMRA requires fiscal tax invoices to carry buyer details and a verifiable QR code transmitted through the Fiscalisation Data Management System, and input-tax claims are only allowed on compliant invoices. The sales flow therefore ends at “accepted quote”, and a fiscal device (or a platform integrated with one) takes over. Ask every vendor to draw that line for you.
Dual-currency is a data problem first
Zimbabwe trades in ZiG and US dollars; the official rate has been broadly stable in the mid-20s to the dollar through 2026, with a parallel premium that moved around 20% in August. Quotes issued in one currency at last month’s rate are how margin disappears. The cheap fix is a single price-list source with a rate field and a date, and a rule that any quote older than n days is re-issued. The AI adds nothing if that source does not exist.
Follow-up sequence, illustrative
| Day | Trigger | Draft for approval |
|---|---|---|
| 0 | Quote sent | Confirmation with validity date |
| 2 | No reply | “Any questions on the quote?” with one clarifying question |
| 5 | No reply | Alternative (smaller quantity or payment plan) |
| 9 | No reply | Final reminder before validity ends |
| Any | Customer replies | Sequence stops; salesperson notified |
Five touches instead of one is the whole effect. It requires no model cleverness, only the CRM connection.