How do you Calculate On Time Delivery?

I just found out that our calculation of on time delivery might be skewed. We run a tally at the end of each month and that is the number I present to Top Management. I had always assumed that it was the number of dollars shipped during the month divided into the number of dollars that was due but not shipped times 100. This would give us our percentage of late materials. I now find out that it is figured by dividing the number of LATE dollars SHIPPED by the total dollars shipped times 100. This means we can have 20 late jobs at the end of this month and still show a 100% on time delivery. Next month would look bad of course.

We go by dollar volume because some jobs are $300 and some are $100,000. Now I am not even sure that this is the correct way to calculate this. Maybe jobs would be better, but a lot of our sales are KANBAN so I’m not sure how we would handle that.

I am thoroughly confused on this and would appreciate some Cove input.

Marc, Please feel free to move this if you think it will do better elsewhere.

Re: How do you figure On Time Delivery?

What you're calculating isn't on-time delivery, at least the in the way that customers look at it. Each order has a promised delivery date, usually with some kind of +/- cushion. On-time delivery is a measure of how well you meet those dates.

What you're measuring is some kind of dollars-lost (or payments delayed) due to late deliveries, which is a different kettle of fish.

We used to do the calculation by both dollars and units, but quit using dollars. Sometimes our $300 order holds up a $1m project, so the fact that it was a small order doesn't necessarily correlate with customer satisfaction.

Now we only use late-units/total-units, but after reading your post, I'm re-thinking that. If we have some May deliveries that ship in June, it should be May's numbers that look bad, not June. I'm not sure if it matters (depends on the root cause) but thanks for the process improvement idea.

dkusleika said:

We used to do the calculation by both dollars and units, but quit using dollars. Sometimes our $300 order holds up a $1m project, so the fact that it was a small order doesn't necessarily correlate with customer satisfaction.

Now we only use late-units/total-units, but after reading your post, I'm re-thinking that. If we have some May deliveries that ship in June, it should be May's numbers that look bad, not June. I'm not sure if it matters (depends on the root cause) but thanks for the process improvement idea.

Click to expand...

It shouldn't matter what month it is if a thing is delivered late. Tracking on-time delivery should be a method of monitoring how well you're keeping customers happy. Although there's probably not much you can do about it, having anyone other than accountants tracking monthly billing will lead to "business decisions" regarding nonconforming or suspect product sooner or later.

dkusleika said:

We used to do the calculation by both dollars and units, but quit using dollars. Sometimes our $300 order holds up a $1m project, so the fact that it was a small order doesn't necessarily correlate with customer satisfaction.

Now we only use late-units/total-units, but after reading your post, I'm re-thinking that. If we have some May deliveries that ship in June, it should be May's numbers that look bad, not June. I'm not sure if it matters (depends on the root cause) but thanks for the process improvement idea.

Click to expand...

There are customers who do not consider early delivery as better, just because you have stocks on hand. An other aspect is the transit time that is taken for the actual delivery at customer specified place, if this is your responsibility. Considering these two, you will have to assess how many of your shipments touch base on the stipulated date (customer warehouse or your warehouse as the case may be) and how many have missed.
Finally the purpose of this is to find out the reason why you have missed so that you can decide how you can correct them going forward.

I think the phrase I overlooked was “Customer Satisfaction”. This is what we are really measuring, not dollars & cents. I now think we should go by the job, not the value. A $300 job that is late may be just as important to customer A as a $300,000 job is to customer B. I will work on this, but would still appreciate any input that anyone has. .

Here's the over-the-top example that the OP got me thinking about. I have three shipments in May and 100 in June. One of the May shipments goes in June and everything else is on time. Now I calculate 100% on time in May and 99% on time in June. The other way it's 67% in May and 100% in June.

In the former, I don't investigate because 99% is above my threshold. In the latter, I do investigate. In reality, if my shipments really fluctuate that much I shouldn't be using a percentage to determine when to investigate, but it's just an extreme example to illustrate a point.

I guess it comes down to finding the right metric to alert me when there's a problem. Probably I shouldn't look at one month in isolation and look at trends. If the on times are trending worse, it's time to look at the details. If they're steady or improving but with an aberration in a particular month, maybe committing resources would be a waste.

When we work to months, we tend to get into 12 blocks and see deliveries more as our monthly targets rather than when customer requires. The better way we follow is to work on production weeks (week number 01 to 52) of the calendar, and plan the complete supply chain and subsequent deliveries over these 52 weeks aligning with customer delivery schedules. The performance and monitoring gets better as well as the controls.

dkusleika said:

Here's the over-the-top example that the OP got me thinking about. I have three shipments in May and 100 in June. One of the May shipments goes in June and everything else is on time. Now I calculate 100% on time in May and 99% on time in June. The other way it's 67% in May and 100% in June.

In the former, I don't investigate because 99% is above my threshold. In the latter, I do investigate. In reality, if my shipments really fluctuate that much I shouldn't be using a percentage to determine when to investigate, but it's just an extreme example to illustrate a point.

I guess it comes down to finding the right metric to alert me when there's a problem. Probably I shouldn't look at one month in isolation and look at trends. If the on times are trending worse, it's time to look at the details. If they're steady or improving but with an aberration in a particular month, maybe committing resources would be a waste.

Click to expand...

If you're concerned about whether or not deliveries are on time, using extrinsic parameters such as month of shipment doesn't help anything. A thing is either on time or it's not.

Jim Wynne said:

If you're concerned about whether or not deliveries are on time, using extrinsic parameters such as month of shipment doesn't help anything. A thing is either on time or it's not.

Click to expand...

I agree that it is quite arbitrary to pick a given period of time, but I have to use some method to catagorize our data to present to Top Management. A month seems a logical choice. As I see it now I would say, "We shipped 100 orders last month - 99 orders were shipped on time and one was shipped after the due date". This would give me a 99% delivery rating. There is no way I could go longer than a month without incuring the wrath of Management. Doing it weekly would become very cumbersome.