Why Does It Take So Long to Produce a Quotation?

Why does a quotation that takes less than two hours of work sometimes take four days to reach a customer?

A manufacturer receives an enquiry for a component it has made before. The estimator retrieves the previous costing. Purchasing needs to confirm a material price. Engineering must check a revised drawing. Production planning is asked whether the requested delivery date is achievable. Four days later, sales is still chasing an answer.

Nobody has necessarily been slow. Each person may have completed their part competently. Yet the customer has waited four days for something that required relatively little active work.

I would be cautious about treating this as an estimating problem. Before buying costing software, recruiting another estimator or asking AI to read the drawings, I would want to know where those four days went.

Manufacturers understand the difference between the time a component spends being processed and the time it spends travelling through a factory. The same distinction matters in the information factory: the information, coordination and decisions that enable physical production. I explored that wider argument in The Information Factory: Why Manufacturing Improvement Shouldn’t Stop at Production. Here I want to examine one particularly visible result: how long a customer waits for a quotation.

A quotation is a promise, not just a price

The temptation is to see a quotation as a calculation. Work out material, labour and overheads, apply a margin, and send the figure to the customer. But a quotation can also commit the business to a specification, quality requirement, delivery date and commercial risk.

Imagine a repeat part with a revised drawing and a tighter delivery requirement. Engineering might need to establish whether the revision changes the manufacturing method. Purchasing might need a new price or confirmation of availability. Planning might need to consider existing commitments. A manager might need to authorise an exception to normal terms.

Those checks are not necessarily bureaucracy. An apparently minor drawing change could affect inspection or scrap risk. A delivery date agreed without reliable capacity information could create problems after the order is won. Good judgement protects the customer and the manufacturer.

The objective, therefore, is not to eliminate every check. It is to ensure that the right information reaches the right person in time for a sound decision, without unnecessary searching, waiting or repeated work.

A busy estimating department is not necessarily an effective quotation process. We need to examine the whole system, not simply the people who calculate prices.

Two hours of work can occupy four days

Consider an illustrative quotation with the following active effort:

These are hypothetical figures, not measurements from a manufacturer. Even allowing for other small tasks, they show how a quotation can involve less than two hours of work and several days of elapsed time. Much of the difference may be found in the intervals: waiting for a response, clarifying a specification, locating a record, finding someone authorised to decide, or returning to work that was thought to be complete.

On a factory floor, a pallet waiting outside a machine is visible. A quotation waiting in an inbox is less obvious. It may be described as “with engineering” even though nobody has agreed what engineering must decide, or “awaiting purchasing” when the supplier has never received a complete request.

A useful investigation distinguishes active work, necessary waiting, potentially avoidable waiting and rework. Not every interval is waste. Waiting for a customer to clarify a safety-critical requirement, for example, may be appropriate. But we should know that it is happening and who owns the next action. We should also avoid treating all waiting as the same: time waiting for a customer is not necessarily something the manufacturer can eliminate, while time spent awaiting an internal handover may be more directly controllable.

What Little’s Law tells us — and what it does not

One of the most useful ideas from operations management is Little’s Law. For a system in long-run equilibrium, with consistent boundaries and measurements:

Average unfinished work = average completion rate × average lead time.

Rearranging it gives:

Average lead time = average unfinished work ÷ average completion rate.

Suppose a manufacturer completes ten quotations per working day and has, on average, forty enquiries in progress between receipt and issue. On those assumptions, average quotation lead time is four working days: 40 ÷ 10 = 4.

If the business achieves an average of two working days while still completing ten quotations per day, the corresponding average unfinished workload would be twenty enquiries. This is a relationship between averages, not a recommendation to delete twenty enquiries, refuse customers or impose an arbitrary work-in-progress limit. Such actions could simply hide demand or shift waiting elsewhere.

The practical question is what would allow quotations to progress with less accumulation while maintaining appropriate checks and a reliable completion rate. The arithmetic cannot establish whether a two-day average is feasible: the business would first need to understand the causes of delay and the demand placed on its people.

Little’s Law does not identify the cause of a delay. Nor does it guarantee that changing unfinished work will change throughput or lead time in the way we hope. The system boundary must be clear, the units consistent and the observation period sufficiently representative. Enquiries awaiting customer clarification cannot be counted one way before an intervention and another way afterwards without distorting the result.

Its value is that it encourages managers to look at the whole flow of enquiries, not just the apparent productivity of one department.

Why everyone can be busy while customers wait

A second lesson comes from queueing theory. When arrivals and processing times vary, a resource operating close to its available capacity can experience sharply increasing waits.

Ten straightforward repeat enquiries arriving evenly throughout a day do not place the same demands on an estimator as ten enquiries arriving together, several of which require engineering judgement. Even when the average workload appears manageable, variation can create queues. If every available hour is already committed, there is little room to absorb that variation.

This is not an argument for arbitrary spare capacity or for leaving people idle without purpose. It means that measuring performance solely by utilisation can mislead. Cross-training, reducing unnecessary variation, protecting time for complex reviews or adjusting capacity may be more useful than pressing everyone to remain fully occupied. The department may look efficient while the customer experiences an increasingly slow and unpredictable response.

The two ideas are related but distinct. Little’s Law describes the average relationship between unfinished work, throughput and lead time. Queueing theory helps explain why variability and high utilisation can cause waiting. Neither tells us, without investigation, which intervention is appropriate.

There is also important precedent here. Rajan Suri’s Quick Response Manufacturing applies lead-time thinking across the business, including office activities. His wider QRM work explicitly identifies quoting and order processing as areas of application. The extension of flow analysis beyond the shop floor is established work, not a new theory I am claiming to have invented. What interests me is how we can combine it with an explicit examination of information quality, organisational decisions and the commitments a manufacturer makes to its customers.

Five questions that expose different kinds of delay

When examining quotation performance, I find it useful to separate five questions. They are investigation lenses, not a scientifically validated classification or five independent causes. A single delay can involve several at once.

1. Flow: where does the time go?

What happens between receipt of an enquiry and issue of the quotation? How much time is spent doing the work, waiting for it, or repeating it? Where do enquiries accumulate? Does the business know how many are open and how long they have been open?

2. Information: what is missing or unreliable?

Does estimating have the approved drawing revision? Is the material price current? Are customer requirements complete? Can people find the authoritative record without searching several inboxes or spreadsheets? A handover cannot work reliably if the recipient must first reconstruct what has already been agreed.

3. Capacity: where is demand exceeding the ability to respond?

Is there a genuine constraint in engineering review, purchasing or estimating? Are peaks of complex enquiries overwhelming a team that has sufficient capacity on average? Would moving work earlier simply create a longer queue at the next stage?

4. Decisions: who can make the next judgement?

Does somebody know that a decision is needed, understand the options and have authority to decide? A quotation can be technically complete yet remain stationary because nobody is clear who can approve a price exception or delivery commitment.

5. Commitment control: can we deliver what we promise?

Does the proposed price reflect the correct specification and risk? Can the production system meet the quoted delivery date? Have necessary checks been preserved? A fast quotation that leads to an unprofitable or undeliverable order is not an improvement.

These questions matter because different causes call for different responses. Another estimator cannot repair contradictory drawing records. An automated reminder cannot grant commercial authority. A faster costing tool cannot resolve an overloaded engineering review. And removing a technical check may reduce turnaround at the expense of quality.

A worked example: the quotation that kept returning to engineering

Consider a small precision-engineering manufacturer. A repeat customer asks for a price on a familiar component but supplies a new drawing revision and requests an earlier delivery date.

Sales forwards the enquiry to estimating. The estimator retrieves an earlier costing and sends the drawing to engineering. Engineering notices that a tolerance has changed, but the email does not say whether the customer expects a different inspection method. The enquiry returns to sales for clarification. Meanwhile, purchasing requests an updated material price. Once the clarification arrives, engineering reviews the implications and planning checks whether the requested delivery date can be met. Finally, a manager approves the offer.

The estimator’s calculation may be a small fraction of the elapsed time. The quotation has moved repeatedly because the business needed information and decisions that were not available at the outset.

The five lenses reveal different things. Flow shows the returns and waits. Information identifies the unclear drawing requirement and the need for a current price. Capacity asks whether engineering was delayed by competing work or by the clarification itself. Decisions asks who was responsible for obtaining the customer’s answer and authorising the offer. Commitment control explains why engineering and planning checks mattered in the first place.

A plausible improvement might be an initial completeness check that flags changed drawing revisions and missing delivery assumptions before the enquiry reaches estimating. Another possibility is to agree an explicit owner for obtaining customer clarification and to distinguish quotations ready for technical review from those awaiting information. But we should test that hypothesis. If the check adds work to every routine repeat quotation, it might make the overall system slower. The investigation should guide the intervention, not merely justify a new form.

The Five-Quotation Test

A manufacturing director does not need a major software project to begin. I would start with five recently completed quotations and reconstruct what happened to them.

Choose two that were straightforward and completed quickly, two that took unexpectedly long, and one involving substantial technical or commercial uncertainty. These are purposively selected cases: the purpose is to expose mechanisms, not to produce a statistically representative sample. If the business handles different kinds of work, note which type each case represents.

For each quotation:

1. Establish the boundary. Record when the customer enquiry arrived and when the quotation was issued. Apply the same definition to every case.
2. Reconstruct the sequence. Use emails, drawings, costing records and conversations with the people involved. Record activities, handovers, waits and returns for clarification.
3. Distinguish evidence from recollection. Where times are unknown, say so. Do not invent precise processing times from memory.
4. Ask what was missing. Identify information gaps, competing work, unclear decision ownership and necessary controls. Check the current drawing and source of price information.
5. Find a hypothesis worth testing. Look for a possible repeated cause, including cases that contradict your first explanation. Choose one proportionate improvement, name an owner and agree when its effects will be reviewed.

You do not need specialist software to begin. A spreadsheet or a simple table is sufficient. For each quotation, record when the enquiry arrived, when the quotation was issued, where it waited, who was responsible for the next action and what evidence supports your explanation. The following hypothetical example illustrates the approach.

Illustrative example only — not actual manufacturing data.

Do not interpret five cases as a reliable estimate of the average lead time for the entire business. To measure performance, also count open enquiries, completions and turnaround over a representative period, separating routine and complex enquiries where appropriate. Be consistent about how cancelled enquiries and customer-held requests are treated. For Little’s Law, the average number open and the completion rate must refer to the same population of enquiries. Where the system is changing rapidly or arrivals exceed completions over the observation period, avoid presenting the resulting ratio as a dependable lead-time estimate.

The next step is a small, reversible trial. If missing information repeatedly causes engineering delays, test a better enquiry handover. If nobody knows who can approve an exception, clarify the decision rule. If current supplier prices are difficult to find, agree one maintained source. Then examine subsequent quotations, taking account of changes in enquiry mix and workload. Agree in advance what improvement would look like and what would count as a failure; a few better examples after the change are not proof of causation.

Crucially, measure more than speed. Did quotation accuracy improve? Was there less rework? Were delivery promises feasible? Did the new check create a delay somewhere else? Faster output is valuable only if the business can stand behind what it offers.

Choosing an improvement before choosing a technology

Once we understand the problem, the intervention may be surprisingly ordinary.

A clearer handover, an agreed owner for each outstanding action, a sensible distinction between routine and technically uncertain enquiries, or a reliable location for current drawings may achieve more than a new system. These are organisational design choices.

Existing software may then help. A shared view of open enquiries can make status and ownership visible. Structured capture can reduce repeated requests for missing information. Automation can notify a responsible person when a clearly defined action is overdue. None of these tools should create a second, conflicting record that staff must maintain by hand.

AI may also have a role. It could assist in extracting requirements from correspondence, highlighting possible differences between document revisions or drafting clarification questions. But these are assistance tasks, not permission to make unreviewed technical or commercial commitments. An AI-generated comparison can be wrong; a missing tolerance or incorrect material assumption can have consequences well beyond the office.

Before using external AI services with customer drawings, prices or correspondence, the business must also establish appropriate confidentiality, contractual and data-handling controls. Technical drawings and commercial prices may be confidential even when they contain no personal data. If the information is not suitable for the chosen service, it should not be uploaded merely because the tool is convenient.

This is why I prefer to use AI on the business before deciding where to put AI in the business: first understand the system, then decide what work genuinely benefits from technological assistance. That investigation need not become a lengthy consultancy exercise: for a well-understood, low-risk task, a small trial may be entirely proportionate.

What should a manufacturing director measure?

I would start with a small set of measures, defined consistently:

• Quotation lead time: receipt to issue, with a clear treatment of customer-held enquiries.
• Open quotations: how many remain unfinished at regular intervals.
• Completion rate: how many are issued per working day or week.
• Turnaround variation: the spread between routine, complex, quick and slow cases, not just one average.
• Quality of commitment: rework, pricing corrections, specification errors and whether quoted delivery dates prove achievable.

If the business can reliably link quotations to orders, it may also examine conversion and subsequent profitability. But a rise in conversion cannot automatically be attributed to faster quotations: price, customer mix, demand and competition matter too.

Measures are useful when they help people ask better questions. They become harmful when teams learn to improve the number by moving enquiries out of the system prematurely or sending quotations before essential checks are complete.

Questions worth asking before changing the process

Would a new estimator solve the problem? Possibly, if estimating capacity is the constraint. But if quotations mainly wait for customer clarification, technical decisions or approval, adding estimating capacity may have little effect.

Should every enquiry follow the same route? Not necessarily. A repeat part with an unchanged approved specification may need different checks from a new engineered component. The business must define the conditions for any simpler route, rather than assuming that a familiar customer makes a changed drawing safe.

Could faster quotations increase sales? They might improve responsiveness, but the commercial result depends on price, customer needs, capacity and the quality of the offer. Conversion and margin should be observed, not promised.

Start with the next five quotations

The next time somebody says quotations take too long, I would resist beginning with the assumption that estimators need to work faster.

I would take five completed enquiries and ask where the time went. Which waits were necessary? Which arose because information could not be found? Which decisions lacked an owner? Where did work accumulate? And which checks protected the business from making a promise it could not keep?

The answer may be a capacity problem. It may be unreliable information, unclear responsibilities or avoidable handovers. It may justify better software, automation or AI. We will not know until we have looked at the work.

A quotation is more than a calculation. It is a promise to a customer. The challenge is to make that promise without unnecessary delay, while preserving the judgement that makes it trustworthy.

Further reading

• Rajan Suri (1998), Quick Response Manufacturing: A Companywide Approach to Reducing Lead Times, Productivity Press — established treatment of company-wide lead-time reduction, including Part III on office operations. Publisher information.
• Rajan Suri, It’s About Time: The Competitive Advantage of Quick Response Manufacturing — describes quoting and order processing as QRM applications.
• MIT OpenCourseWare, Introduction to Manufacturing Systems — lecture materials for further study of manufacturing systems; this is not presented as a direct citation for each formula above.