Two hundred hours to send a price

The customer was willing to pay a premium and could not, because the quote was late

An 8 to 10 week sprint. One senior analyst and two juniors. Twelve months of the client's own quotation data, four process stages, three service packages, eleven customer interviews, and an audit of 302,438 supplier pricing records.

A vessel held in port by a failed piece of cargo handling equipment is losing money by the hour. When something breaks, the ship's purchasing manager sends a request for quotation to two to five suppliers and buys from whoever comes back first with a workable price and lead time. Speed is not a service nicety in that market. It is the sale.

An earlier customer study had already told the client what was wrong, in the customers' own words: they were prepared to pay a premium for a critical spare part and the quote arrived too late for them to place the order. The revenue did not go elsewhere in the sense of being lost to a cheaper bid. It went elsewhere because nobody answered in time.

The average time to quote was six days. The target the leadership team had set itself was 48 hours.

An average of six days describes nothing that anyone can fix

Time to quote is a single headline number covering four sequential stages and three quite different service packages. Averaged, it is useless. A damage repair and a dry dock refit have almost nothing in common except that both end in a price, and the stage that dominates one is not the stage that dominates the other.

Almost all of the elapsed time is waiting, not working. That is what makes it invisible to the people doing the work, each of whom is busy the entire time.

The third is that the biggest cause of delay is not in the process diagram at all. It is in a data table nobody owns, and you only find it by asking what the quotation preparer is actually waiting for.

Two hundred hours to send a price, and half of it is one stage

Split by service package, the six day average dissolves into three quite different problems. Damage repair, the most urgent category, averages 121 hours. Running spares averages 144. Dry docking averages 208. The target for all three is 48.

The dominant delay is a person sitting at a screen unable to finish a quote, because the price they need is not in the system and they have to email a supplier to ask.

Quotation preparation is the largest stage in every package. For dry docking it alone is 103 hours, more than double the entire target. The leadership team's hypothesis had been that the delay came from handovers between departments and from an undifferentiated funnel. Both of those are real and both show up in the data, as the notification queue and the review stage. Neither is the biggest.

A third of the queue is in the wrong order

Requests are worked largely in the order they arrive. To find out what that costs, every quote in a twelve month window was scored retrospectively on a 0 to 100 scale, weighted 70 per cent on customer value and 30 per cent on the opportunity itself, and sorted into four priority bands.

Where the 10 per cent comes from. It is a deliberately conservative assumption, not a measurement. The other 90 per cent of losses are attributed to causes the study did not attempt to separate: delay elsewhere in the process, relationship strength, uncompetitive pricing, lead times and delivery constraints. A different assumption gives a different number, and the calculation is shown so that leadership can substitute its own.

The reordering is severe. Under the current logic 26,773 quotes sit in the top band, at an average value of 9,100 euro and a win rate of 0.51. Under a value-based logic the top band holds 1,035 quotes at an average value of 17,900 euro and a win rate of 0.65. The work that deserves to be expedited is roughly one twenty-fifth of what is currently being expedited, and it is worth twice as much per quote.

Across the three service packages, more than 30 per cent of quotes are mis-prioritised. The cost is measurable: a mis-prioritised quote waits about an extra day in the notification stage before anyone picks it up.

Of the high priority quotes that were lost, the total value was 16.4 million euro. Attributing 10 per cent of that loss to delay, which is an assumption and is labelled as one, gives a potential revenue uplift of 805,000 euro a year from ordering the queue on value rather than on arrival.

Two thirds of the supplier prices had expired

The quotation preparer needs a supplier price and a lead time for every line on the request. Those live in purchasing info records. If the record is current, the quote takes minutes. If it has expired, the preparer emails the supplier and waits.

The records were being refreshed reactively, when a request came in and someone noticed. So the refresh cost was being paid inside the quotation clock, every time, on exactly the parts that get quoted most.

Of 302,438 active records, 204,268 had expired and a further 53,138 carried no validity dates at all. Fewer than one in six was current.

That alone would be a housekeeping problem. What makes it a commercial one is which records had expired. Among the expired records, 51,265 had been quoted three or more times in the year, and 77 per cent of those had been sold at least twice. These are not obscure parts. They are the parts the business sells repeatedly, and their prices are stale.

Six fixes, and only two of them carry most of the money

Six solutions were designed, one for each stage where the analysis found time or value leaking. Each was sized against the client's own data rather than against a benchmark.

Automating quotes for requests that contain only stock parts is the single largest item at a modelled 1.4 million euro. It also produces the clearest operational result: on-time delivery against a 24 hour target rises from 40.5 per cent to a modelled 68.7 per cent for the client's top tier customers, an improvement of 28 percentage points, with the remaining 31 per cent being non-stock requests that the automation does not touch.

The after-hours coverage solution is the one that has to be read carefully. At full effectiveness it models a 334,000 euro uplift against the cost of four additional people. At 25 per cent effectiveness the modelled uplift is roughly 83,000 euro and does not cover the cost. It is a solution that only works if it works well, and the study says so rather than presenting the best case.

What the numbers rest on

Every uplift on this page can be rebuilt from two inputs: the stage timings in the first finding and the hit rate regression described above. Where a figure depends on an assumption, the assumption is stated next to it rather than in a footnote.

On anonymity. The client is not named, nor is its owner, nor any of the eleven customers interviewed. The seven product groups are counted rather than listed, and the software systems involved are described by function rather than by product name, because naming any of them would identify the client to anybody in the sector. Every figure and the full method are given.