Building and approving sales quotes at a company sending 5,000 a year: 28% wait for a person

For the executive who decides

• The customer acts
• The machine acts
• A person decides
• Nothing runs

An illustration built from the company's numbers: the times are examples, and the minutes a quote and an approval are estimates.

1The job

Building each sales quote, and sending the ones that break the pricing rules to a person for approval. The company quotes more than 5,000 times a year; in one quarter it sent over 3,800. Pricing lives in spreadsheets, customer details in the CRM, and approvals in a team chat. What they sell, and how big the company is, are not stated.

2The number today

In Q2 we processed 3800 plus quotes and around 28% needed manual approval. That was about 1,064 approvals in one quarter, about 16 a working day. The company also says it sends 5,000+ quotes a year, a slower pace than that quarter; at 5,000, it is 1,400 approvals a year, and the money below uses that lower count. The company says many approvals came from prices edited by hand in the spreadsheets, not from real exceptions.

The staff time: about $126,000 a year to build the quotes and about $24,500 to approve them, at US median pay for a sales rep ($33.65 an hour) and a sales manager ($71.28). About $150,000 a year in all, or about $43 for each quote that needs approval. Then the wait: other teams report 3 to 4 days for an exception, and one deal that sat 3 weeks while the approver was on leave. What the delay costs in lost or late deals is unknown, because the company gives no deal size and does not say what it sells.

Estimate. The $126,000 assumes 45 minutes a quote, the low end of another team's 45 to 60; the $24,500 assumes 10 minutes of the approver and 10 of the rep for each approval. Both use the lower yearly count; at the second quarter's pace they would be about three times higher. The company's own timed week replaces all three.

3The verdict

Hand over the quote and the routing. Keep the exception.

• Hand to the machine: building every quote from a single price list; checking each quote's discount, margin and terms against written limits; sending the quotes that stay inside them; sending the rest to the right approver, with the reasons and the past similar approvals; and keeping a record of every decision.
• Keep with a person: any discount, margin or payment term outside the limits; renewals that carry an old exception; and any change to the price list or the limits themselves.
• Not yet: letting it approve an exception because a similar one was approved before; letting it change a price; and letting quotes for renewals and contract changes go out on their own, until real ones have been tested.

What the people are for now: the rep keeps the customer and the deal. The approver decides which deals are worth a price outside the rules, and says why. The machine does the copying, the checking and the chasing.

4The bill

About $400 to $1,500 a month for 10 people, with the approval rules inside the CRM the company already runs. That is $40 to $150 a seat on the plan that holds approval rules, and up to $3,500 to start, depending on the CRM. A standalone quoting tool is $490 a month plus an approval add-on whose price is not published, and it cannot check margin. Building it is about $25 a month in running costs, plus the build and someone to own the rules. People building the quotes by hand would cost about $3,600 a month for the two it would take, and the approvals would still wait for the manager. Checking: for the first two weeks, the approver reads a list each evening of every quote that went out without a person. The check time is not known until it is counted.

Estimate. 10 seats is assumed, because the company gives no headcount. The build cost assumes about 417 quotes and 117 approvals a month.

5Owner and off switch

The person who signs off on discounts today owns it, because the limits are theirs to set. The off switch: change the approval rule so that every quote needs a person. Quotes then wait for approval as they do today, but nothing goes out unchecked. It takes under a minute. Try it once in week one.

6The brief to hand your builder

The job: build every quote from a single price list, check it against written limits on discount, margin and terms, send the quotes inside the limits, and send the rest to the right approver with the reasons.

Today: 5,000+ quotes a year; 28% needed manual approval in the second quarter, many because prices were edited by hand in spreadsheets. Pricing is in spreadsheets, customer details in the CRM, approvals in a team chat, and people keep asking which version is correct.

Done when: for two weeks, every quote is built from the one price list; no quote goes out outside the limits without a named approver; every approval request reaches its approver within 5 minutes with the reason; no request waits more than 4 hours without going to a backup; and both the share of quotes needing approval and the hours from request to decision are lower than in the week before switch-on.

It may never, on its own: approve a discount, margin or term outside the limits; change a price on the list; send a quote that was edited after it was approved; or approve a renewal that carries an old exception.

You get from us: the price spreadsheets and which one is right; the discount, margin and term limits for each role, written down; 30 recent quotes and their approvals, messy ones included (a hand-edited price, a renewal, an odd payment term); and a named backup for every approver.

We own: the price list, the limits, the CRM, the accounts and the billing.

We get back: each morning, a list of the quotes that went out without a person, the approvals still waiting and for how long, and every exception approved, with its reason and whether it was a one-off.

Ask whoever sets it up	A good answer	A bad answer
What happens when the approver is away?	Every rule has a backup, and a request waiting more than 4 hours moves to them	"They'll get a notification"
What happens when two rules fire on one quote, or a rule still points at a product we dropped?	One person has the final say on every quote, and the rules are checked against the product list each quarter	"The system handles it"
In six months, how will we know why a discount was allowed?	The approver records a reason, and whether it was a one-off, in the same click, and it can be found by customer	"It's all in the chat thread"

For the operator who builds and runs it Paid layer

7The working set

• Trigger: a rep starts a quote on a deal in the CRM.
• Steps: pull the customer and deal from the CRM; price each line from the one price list; take the rep's quantities and discount; work out the discount, the margin on each line and the total, and the terms; check the rules (discount limit by role, margin floor, non-standard payment terms, a renewal with an old exception, any price typed over the list); inside every rule, approve it and send it or hand it to the rep to send; outside any rule, hold it and post it to the approver for that rule in the chat app, with approve, change and reject buttons, the rule broken, the margin, and the last three approvals for similar quotes; take the decision with a reason code and "one-off or precedent"; send any quote edited after approval back for approval; log every step.
• Data: one record per quote: quote ID, deal, customer, rep, lines, list price, discount %, margin %, terms, rules fired, auto-approved or routed, approver, backup used or not, request time, decision time, decision, reason code, one-off or precedent, version, sent time, and whether the price was typed over the list.
• Review step: for two weeks, the approver reads each evening's list of quotes that went out without a person. After that, the sales lead reads the week's exceptions by reason every Monday, and the price owner checks the rules against the product list every quarter.
• Alerts: a request waiting more than 4 hours (it moves to the backup); a quote going out with a price typed over the list; a rule that names a product no longer sold; a quote passed back and forth between approvers more than twice; and the wrong-but-green count, quotes the machine approved that a person later changed or credited.
• If it breaks: what you see: quotes stuck on "pending", approvers not told, or quotes going out without a check. Check first: the price list is the current one and still connected; the chat connection is authorized; the rule list has an approver and a backup on every rule. Safe fix: set the rule so every quote needs a person; approvals go back to the chat as today.
• Cost per quote: about $0.96 to $3.60 inside the CRM ($400 to $1,500 a month over about 417 quotes); about $0.06 in running costs if built, plus the build.
• Baseline, before switch-on: one week of: quotes built; minutes a quote (time it, do not ask); share needing approval, split by cause (a hand-edited price, a discount, a margin, terms); hours from request to decision; and quotes sent at a price different from the list.

Estimate. The cost per quote assumes 10 seats and 5,000 quotes a year (about 417 a month). The baseline week replaces both.

Tools by name: inside the CRM, Zoho CRM Enterprise ($40 a user, approval process; CPQ from Professional), HubSpot Sales Hub Enterprise ($150 a seat, quote approval workflows, $3,500 onboarding; quote approvals also on Revenue Hub Professional and Enterprise), or Salesforce Revenue Cloud Growth ($150 a user, quoting and configurator); standalone, PandaDoc Business ($49 a seat, approval workflows; conditional approvals a paid add-on; conditions on discount and document value, not margin); people, Assistiq ($1,797 an operator a month); built, n8n Starter (20 euros a month, 2,500 runs), Slack for the approval buttons, Claude Sonnet 5.5 for the approver's summary, and one shared price table.