Which option is right for automating sales quotes?
Start with the option that solves the problem without adding unnecessary steps. If your CRM or ERP already generates quotes, check its features before buying another tool.
| Option | When it makes sense | What to check |
|---|---|---|
| Template and product catalog | Standard products and few exceptions. | Price updates and version control. |
| Existing CRM or ERP | Your system already holds customer, product, and pricing data. | Configuration, permissions, and discount rules. |
| Automation platform | You need to connect tools and route approvals. | Available connections, failures, and ownership. |
| Custom software | Existing tools can't handle your rules or integrations. | Maintenance, data access, and code ownership. |
AI can supplement any of these options if interpreting free-form requests takes time. It doesn't replace the product catalog or business rules.
How does an automated quote work in practice?
This example is illustrative; it doesn't describe a Quarl client.
A sales team receives an email: “I need twelve boxes of medium gloves for delivery in Medellín.” The system identifies the product, size, quantity, and city. If several product codes could match, it asks a person to choose before continuing.
It then checks the authorized product catalog and the customer's price list. It calculates the amount using defined rules, adds delivery terms, and prepares the quote. If a term is missing or a requested discount needs approval, it holds the draft instead of making up an answer.
The result is a draft for twelve boxes of the confirmed product, with the current price and available terms. A person reviews the details and sends the quote. If there isn't enough inventory, the system flags it. It doesn't promise a date without an authorized source.
The rules system, not the AI model, should do the calculation. For example, Business Central calculates a line amount by multiplying the unit price by the quantity. It also updates totals when lines are created or changed.
Availability needs a separate check. Microsoft warns in the Sales Order Agent setup documentation that disabling this check can let the agent quote unavailable quantities.
How can you start sending quotes faster?
Get the process in order before connecting tools. Automation won't fix an outdated price list on its own.
- Define the authorized product catalog. Gather product codes, sales units, and current prices. Assign someone to update the catalog, and make each version's effective date visible.
- Document the business rules. Define which price list applies to each customer, which discounts need approval, and who can approve them. Include taxes and delivery terms where applicable. The Dynamics 365 Sales documentation shows how the catalog, line items, and discounts affect calculations.
- Automate the repeatable steps. Use a template for standard requests. If the data is already in a CRM or ERP, check whether you can generate the draft there. To connect apps or route a request for approval, consider an automation platform.
- Give AI a specific job. AI can extract products and quantities from emails in different formats. It should flag missing data and request a review when a product code is ambiguous. In Sales Order Agent, Business Central calculates default prices and discounts. Requested changes go to a person.
- Decide who reviews and sends. Start with a human review of the complete draft. Keep the request, catalog version, and approval so you can explain where each term came from. If you reduce reviews later, define which cases must still be held.
How do you measure whether automation works?
Plan a before-and-after test using real requests. These metrics support a decision; they aren't guaranteed results.
- Prepare a sample for review. Gather common requests and exceptions, such as ambiguous products or special discounts. For each request, an assigned reviewer records the correct response using the catalog and terms in effect on that date.
- Measure time to send. Record when each request arrived and when the quote was sent. Use records from the current process as your baseline. Afterward, measure comparable cases and separate time spent waiting for approval.
- Count price and quantity errors. Compare each draft with the reference answer before correcting it. Record price errors and quantity errors separately, along with the total number of quotes reviewed.
- Calculate the human correction rate. Divide the number of drafts that needed changes by the total number reviewed. Approval without changes doesn't count as a correction. A misinterpreted product code or incorrect price does.
- Repeat the test using the same criteria. Run the saved requests through the new workflow without sending them to customers again. Use the same reference data and keep the differences for a person to review.
What affects cost and timeline?
The work increases when you have multiple price lists, incomplete data, or rules based on individual negotiations. It also matters whether the catalog is in a shared spreadsheet or needs an ERP integration.
A template may be enough to start, but someone will need to maintain it. An automation platform adds work to configure and operate connections. Custom software requires you to decide who fixes failures, updates rules, and maintains the code.
Before choosing, check whether the tool can access the data it needs and what happens when a connection fails. The system should hold the request and notify the person responsible, not send an incomplete quote.
If the rules are already clear, integrations between systems can keep sales from copying data by hand. If the rules still change depending on who's handling the customer, agree on them before development starts.
