Configure, Price, Quote by the Numbers
Chapter 8 of 8 · 10 min read
Measure what the system changed
Repeat the quoting baseline after the configurator, compare like with like across seasons, separate the system's effect from the market's, and calculate an honest payback.
A configurator project ends the way it began, with measurement. Chapter 1 measured quoting volume, effort, turnaround, accuracy, win rate and discounting before anything changed. Six months after the configurator and the discount policy went live, the bench fabricator measured the same things again. The comparison answers the questions the owner cares about: did quoting get faster, did the firm win more, did margins hold, and was it worth the money? It also shows where the system still falls short.
This chapter makes that comparison. It repeats each measure, compares the same months across years to separate the system’s effect from the season’s, calculates the payback with and without the uncertain sales gain, and looks at what customers and staff say. The failure case is a celebration of a higher win rate that turned out to be mostly a busy season.
Measure the same things the same way
The value of a baseline depends on measuring again in exactly the same way. The firm used the same definitions as in Chapter 1: a quote counts from the date the enquiry arrived to the date the quote reached the customer; a quote is won when an order is received citing it; a concession is any reduction from the configured price. It used the same eight-week windows, and the same assistant ran the comparison.
| Baseline, Chapter 1 | After six months | |
|---|---|---|
| Quotes a month | about 120 | about 130 |
| Estimator time per quote | about 45 minutes | about 10 minutes |
| Quotes sent the same day | about a quarter | about 78% |
| Win rate | about 40% | about 44% |
| Won quotes underpriced | about 6% | under 1% |
| Price given in concessions, a year | about A$29,030 | about A$9,953 |
| Unbuildable configurations quoted | three in eight weeks | none |
Every measure moved in the intended direction. Quotes rose slightly, partly from self-service quoting. Estimator time per quote fell from about 45 minutes to about 10, mostly checking special requests and reviewing larger orders. Most quotes went out the same day.
Compare like with like
The win rate rose from about 40% to about 44%. Before crediting the configurator, the firm compared the same months across years, because its business is seasonal: food factories commission new lines at particular times of year.
The same season a year earlier, before the configurator, had a win rate of about 43%. Against that, the improvement is about one point, not four. Most of the apparent gain came from comparing a quiet season with a busy one. One point is still worth having, and it may grow as customers learn that the firm quotes the same day, but it is far smaller than the first figure suggested.
Count the benefits honestly
The firm separated benefits it could count with confidence from those it could only estimate.
Counted with confidence, per year, excluding GST:
- estimator time: from about 1,080 hours, about A$118,800 at A$110 an hour, to about 260 hours, about A$28,600, saving about A$90,200. Some of those hours were redeployed to special requests and customer visits rather than removed from the payroll, so the cash saving was smaller, but the capacity was real;
- concessions: about A$19,077 less price given away, from Chapter 6;
- pricing errors: about A$5,751 less underpricing, with errors falling from about 6% of won quotes to under 1%.
Estimated, with less confidence: at about 1,560 quotes a year, a one-point rise in the win rate is about 15.6 more wins, about A$37,440 of sales at A$2,400 each, worth about a third of that in gross profit. The firm recorded it but did not rely on it.
Calculate the payback
This is an illustration. The configurator cost about A$62,000 to build and implement, excluding GST, including the price model, the connection to the business system and the generated drawings, and about A$8,000 a year to maintain.
- Build and implementation
- A$62,000
- Upkeep
- A$8,000 a year
- Counted benefits
- 90,200 + 19,077 + 5,751 ≈ A$115,028 a year
- net annual benefit ≈ 115,028 − 8,000 ≈ A$107,028
- payback ≈ 62,000 ÷ 107,028 × 12 ≈ 7.0 months
On counted benefits alone, the configurator paid back in about 7 months. The estimator-time figure is the largest and the softest, because the firm did not reduce its staff. If only half of it is counted as real, the counted benefit falls to about A$69,928 a year, and the payback stretches to about 12 months. Either way, it paid back within the first year.
Ask the people who use it
Numbers do not show everything. The firm asked its estimators, sales representatives, workshop and five regular customers what had changed for them.
The estimators said they spent their time on the interesting work, special designs and large projects, rather than retyping quotes. The sales representatives said they could answer price and lead-time questions on the phone, and that the margin display had changed how they thought about discounts. The workshop said job cards and drawings now agreed with each other. Customers said quotes arrived faster and were easier to read, and two mentioned the sketch.
The firm also heard criticism. Sales representatives found the approval step slow when the sales manager was travelling, and asked for a deputy approver. The workshop wanted the inspection sheet on a tablet rather than paper. Both went onto the improvement list.
A failure case: the win rate that was the season
This is an illustration. Three months after go-live, the owner presented the configurator’s results at a staff meeting: the win rate had risen from 40% to about 45%, and the owner credited the configurator with about A$170,000 a year of extra sales. The sales team were congratulated, and a bonus was discussed.
The accountant, preparing the quarterly figures, compared the period with the same quarter of the previous year and found a win rate of about 44%, before the configurator existed. The busy season had arrived. The firm’s real improvement was about one point, as the comparison above shows.
The detection came from someone comparing like with like. Nothing was lost except some enthusiasm. But the bonus discussion was paused, and the firm set a rule for every later measurement: compare with the same season in the previous year as well as with the baseline, and state both. The configurator was still worth having on the counted benefits alone. The lesson was about evidence, not about the system.
Share the results with the people who made them
The configurator changed the daily work of the estimators, the sales team, the order desk and the workshop, and they had all put up with the disruption of learning it. The owner presented the six-month comparison to the whole firm, with both win-rate comparisons, the counted benefits and the improvement list, and thanked the people whose suggestions had been adopted.
That meeting mattered for two reasons. It showed that the measurements were honest, including the uncomfortable correction to the win rate, which made staff more willing to report problems later. And it connected each person’s work to a result: the estimators saw that their reviewed special requests were the main source of the remaining errors, and the workshop saw that its inspection findings had changed the product model.
The firm also told its regular customers, briefly, that quotes would now normally arrive the same day, that every quote included a sketch and an honest lead time, and how to configure standard benches themselves. Several customers who had drifted to faster competitors returned for their next project. The firm could not measure exactly how many, and did not try to claim a number for it.
Hold the gains
Improvements fade when the attention moves elsewhere. The firm protected its gains in three small ways: the monthly measures page stayed on the agenda of the production meeting; the reference run from Chapter 7 remained a required step for every change; and new staff learned the discount policy and the configurator in their first week, from the people who used them every day.
What the system did not fix
The comparison also showed what had not changed, which is as useful as what had.
Lead times. Quotes now stated lead times honestly, but the workshop was no faster. In the busy season, the quoted lead time reached six weeks, and the lost-quote calls again found customers choosing faster suppliers. The configurator had made the constraint visible. It had not removed it. The owner used the workshop-load figures from Chapter 5 to justify a third fabricator during the busy months, which was a capacity decision, not a software one.
Special requests. About a fifth of enquiries still needed an estimator, because they fell outside the standard range. That share was stable. The special-request log showed a spread of one-off needs rather than a few repeated ones, so extending the model further would cost more than it saved. Special requests remained hand-priced, with the configurator’s cost tables as the estimator’s reference.
Trade customers. Fit-out contractors used the self-service configurator less than expected, preferring to send drawings of whole kitchens and ask for a project price. For them, the estimator’s ability to quote a project of mixed benches quickly mattered more than self-service, and the project view described in Chapter 4 was improved in response.
Decide what comes next
Each improvement idea was written down with an estimate of its value from the measures and its cost, in the same way Chapter 1 had justified the configurator itself:
- a deputy approver for concessions, nearly free, removing a delay the sales team reported weekly;
- inspection on a tablet, a few days’ work, removing the paper sheet and recording measurements directly against the order;
- a connection from the configurator to the firm’s customer records, so follow-up calls on unanswered quotes are scheduled automatically, worth an estimated one or two more wins a month given Chapter 1’s finding that a quarter of quotes went unanswered until chased;
- extending the model to trolleys, the firm’s second product family, which would repeat the whole journey of this book for a smaller range.
The owner approved the first three for the next quarter and asked for a baseline measurement of trolley quoting before deciding on the fourth.
When the numbers disappoint
Not every configurator project shows improvement, and the firm discussed what it would have done if the comparison had been flat. The first step is to check that the system is actually being used as intended: quotes produced outside it, prices overwritten, or rules worked around all show up in the records, as the manual price field did in Chapter 6. The second is to check whether the baseline problem was the one the system addressed. A business whose real problem is capacity will not see its win rate move because quotes go out faster. The third is to ask the people using it what gets in their way. A configurator that is technically correct but awkward to use will be avoided, and the companion book on engineering tools people use covers that in detail.
Keep improving
The measures continue monthly, on a single page the owner reviews with the sales manager and production manager. The page shows quotes, turnaround, win rate against the same month last year, concessions, the special-request log and the realised margin from Chapter 7’s monthly check. The improvement list holds the requests from staff and customers, prioritised by what the measures say matters most.
How this applies to a small Australian business
Small businesses rarely measure the effect of a system they have bought. That makes it hard to know whether to extend it, replace it or stop paying for it. Measure before, measure after with the same definitions, compare the same season and count only the benefits you can defend. Many Australian industries are seasonal, from food processing to construction, so a before-and-after comparison that crosses seasons can mislead in either direction. A configurator that pays back on time saved and margin protected is a good investment even if the sales gain turns out to be small. A sales gain claimed from one quarter’s figures is not evidence.
Exercise: the case for a smaller firm
A smaller firm spends A$28,000 building a simpler configurator with A$4,000 a year of upkeep. It measures that estimator time fell by 350 hours a year at A$95 an hour, concessions fell by A$6,500 a year and pricing errors by A$2,200 a year. Calculate the payback on counted benefits, and again if only half the estimator time is counted.
Worked answer. Estimator time is worth 350 × 95 = A$33,250 a year, so counted benefits are 33,250 + 6,500 + 2,200 = A$41,950, and the net benefit after upkeep is A$37,950. Payback is 28,000 ÷ 37,950 × 12 ≈ 8.9 months. Counting half the estimator time, benefits are 16,625 + 6,500 + 2,200 = A$25,325, net A$21,325, and payback about 28,000 ÷ 21,325 × 12 ≈ 15.8 months. The configurator is worth having in either case, and the firm should be clear in its own records which figure it is relying on.