2015 TOCICO Conference - Final Thoughts

So, I've already written several posts about the learning that went on in Cape Town this past September (see here, here and here) and I guess it's time to move on. Still I want to leave you with a real taste for more and hopefully next time you'll come join us (see the TOCICO calendar for upcoming events near you). This review did not include my sessions and I will share some of that material separately in a future time. It's on my list.

For this final post I chose to review  2 sessions that I found had a high impact on me:

I have already mentioned Rami Goldratt's keynote on forecasting here. He talks about the fact forecasting capabilities in retail are not improving. In fact, they are deteriorating. It is becoming harder and harder to generate predictions that will create good enough forecasts so as to lead to the right decisions. No one is expecting a forecast system to be able to predict accurately what will shop X sell next Tuesday. But the fact of the matter is these forecasts fail at much simpler questions such as how much of product Y will be sold by the entire network over the entire season. This is a simpler question as it is an aggregation, allowing for mistakes to average out.

Rami presented 3 major forces that are eroding the ability for forecast:
1 - Customers are increasingly requiring products to create the exact fit for their needs and tastes. As a result manufacturers and retailers increase their variety. Think of toothpaste, just as an example. It used to be that you were either a Crest user or a Colgate (or another brand). Now each brand has an extensive range of options - different tastes, different usages and problem resolution and many of the possible combinations of these. As the variety grows the volume per SKU shrinks and with it the ability to predict. At the same time the customers learn to expect even more accurate fit and it is clear there is a feedback loop making the situation grow more and more extreme.

Gateway Supermarket Sardines' Section (from Wikipedia)

2 - Customer tolerance time is getting shorter, meaning retailers lose more sales when the exact inventory is not on hand. For example, in a market research Rami cited, people stated that they will pay a higher premium for immediacy that for technological improvement. As a result there is pressure to hold more inventory at the point of sales.
3 - The product life cycle is getting shorter. Customers are forever expecting new products and so there is a on going pressure on NPIs. New products do not have a history we can use for predictions and they have a major impact on the sales of current products (that have a high level of inventory in the shops from item#2) thus practically eliminating the ability to forecast.
On top of all this (as if that was not enough) more and more industries suffer from a product life cycle that is shorter than the supply time.
With all these phenomenas increasing over time, relying on forecasts will lead companies to fail. What we need is a system to minimize the time needed for reading and understanding the signals from the shop floor and to react to them fast. We have the needed IT, we are just using it the wrong way.
Just so you know, Rami promised a video covering all this in more detail - Rami, we are waiting.

The second talk was by Dr. Alan Barnard concentrating on flow. Re-reading Dr. Goldratt's "Standing on the Shoulders of Giants", Alan pointed out that the 5 Focusing Steps are actually an application and should only be used when the boundary conditions dim them relevant. Instead he suggest starting with the concept and principals of flow:

1 - The prime objective of operations is to focus on flow, specifically on reducing flow time.
2 - This objective is to be translated into a practical mechanism that guides operations when not to produce.
3 - Local efficiencies must be abolished.
4 - A focusing process to balance flow must be in place.

The application of these principals will be:
1- Separate work and time buffer.
2 - Aggregate time buffers.
3 - Remove unnecessary buffers
4 - Simplify flow with a single time buffer.
Looking over these lists it is clear that the 5 Focusing Steps, which are used for constraint or bottleneck management, are only relevant when those are the primary causes of blocks to the flow.

This is just the tip of the iceberg, of course. I highly recommend getting access to the conference videos for these sessions, if you weren't there.

If you enjoyed this and want to learn more - share it and subscribe to this blog, great things are coming.

2015 TOCICO International Conference - Cape Town

I've just returned from Cape Town a week ago and I think it's a great time to share some of my conference highlights with those who could not make it this year.

Some Conference Participants (can you spot me?)

First off - join me in sending love and support to Marcia Hutchinson, TOCICO's General Manager, who broke her arm on the first night and needed an operation. Marcia's proffesional management was totally aparent in the fact the conference went on without any interference, leaving her free to take care of herself. GET WELL SOON, MARCIA!

The first day was a full day of keynote speakers interlacing academics and practitioners  and introducing the conference's main thread of the Anti-Fragile vision. Some success stories brought to light the amazing effects TOC can have on a business and how important it is to ensure the process is maintained over time and management switch-overs.

Keynote Speaker

I especially liked Rami Goldratt's session about the forces eroding our ability to generate forcasts, especially in a retail environment. He makes a valid point - yes, forecasts are inherntly wrong but they should be getting closed and closer to being right. After all, they are based on past experiance, of which we have more and more data points with each day passing, on computing abilities, which grow bigger and faster with each day passing and on algorithms, that should improve over time and experiance. Look at the weather forcast as an example. Over the years it has gotten better, to the point that we trust the 10-day forcast as a good enough directive for packing before a trip. Sure, we do not expect it to be right on the money, just in the vicinity. Shouldn't the forcasts for sales numbers and for purchasing be doing the same? Well, they don't and Rami gave a very clear explenation why. He also promissed to put together a YouTube vid covering his explenation, so keep a look out.

With all the interesting take aways I've had, it seems a series of posts is in order, so be on the look out for my future posts on the subject. (Here is the first one)

Here is a sneak peak:

My presentation about buffering life

So, have a look at the conference schedule, and let me know what is of interest to you.

Shop floor insights - freinds and foes

TOC for retail is based on the statistical characteristics of forecasting. Specifically the fact that the more general the forecast - the better fit you will get between results and reality. The other side of this coin is, of course, that the more specific the forecast - the worse fit you'll get. Therefore TOC calls for holding inventory centralized and moving it closer to the end costumer as late as possible.

Pushing the merchendise into the final storage rooms earlier creates local shortages and surpluses. These increase the workload for the sales staff, of course. Since they can never know what is available and what is not, so they are always checking. Another very significant impact this has is the increase of the customer's percieved risk. Since you can't know ahead of time, even when you know for sure the SKU is part of the store's stock, if it will be available in that store at that moment and if they will have the size and color wanted.

As I described in earlier posts, the chain I worked for pushed the inventory forward as soon as possible. I am sure this seems logical to them, after all - if the item is not in the store it can't be sold, so that seems like the best place to store you inventory, no?
The stores also had access, through the central database, to the tracked inventory of all other stores in the chain.  This data is used to reduce lost sales by cross shipping from oher branches. There were always calls from one branch to the next requsting the relevant SKU (though we were always describing, not using the codes) and issueing a cross shipping. I don't think there was a morning we did not have packages of cross shipments coming in AND going out.
A couple of things I noticed. First off - we never checked the warehouse inventory. This makes total sense in retrospect, since the warehouse was automatically replacing any shortages we had if there was stock. So, if the warehouse had it, we knew we'd get it. Even surpluses were rarely sent of to the warehouse. Everything was worked out between the stores. Second - cooperation between shops was choppy at best. Finding the garment you need to get the sale was always a good thing, but sending off a garment wasn't such a hit. We were directed to limit the outgoing shipments to those garments we had enough stock off and the ones that were not selling well. When a branch "overdid it" the manager would stop answering their calls and direct us to do the same. Then, if one of us inadvertedly did answer, the manager of the other store would give that poor bastard such a talk to.....
From my point of view, as a simple sales clerck, this was a no win situation. If I help out the other store - my manager gets mad at me, if I don't help them I have the phone going on all the time and at the end they manage to get a hold of us and they are mad at me.
Taking the owners point of view this whole situation is not in their best interest, either. Don't you think?

Shop floor insights - daily replenishment

When I first encountered the TOC logistic and supply chain management solution I got the impression that the daily replenishment of stock with what was sold, is the driving force of this solution. The solution is based on the idea that the sales outlet should hold minimal stock on hand - just enough to allow you to sell to any patron who may wish to buy the item + enough to support the display.

For the shop I worked in (some background and a disclaimer - I was a sales clerk, this was not a TOC project) the second part was almost negligible, we sold all garments displayed in the shop which were not on the manikins, of which we had only 3. The first part, however, is quite tricky as we have no idea how many patrons may wish to buy a certain item on a certain day. This is why the forecasts aren't good, remember? we had items of which we sold 1 unit, or maybe even nothing at all for weeks and then one day we'd sell 3 or 4. The TOC replenishment solution answers that with a rule of the thumb. You start high, with enough stock to equal 14-20 days of average sales. Then you adjust as you go by tracking end-of-day inventory using 3 zones. Red zone is the bottom third of your target. If for a few days you continually end up having less than one third of your original target then this is a fast runner and the target has to go up. Green zone is the top third, pointing out that this is a slow runner, the target is too high and has to go down. Yellow zone is in between and where we'd like to be.

Well, the store had daily replenishment. Israel is a tiny place, everything is within a few hours drive, there is no sense in replenishing less frequently. The store was also supposed to be replenished according to the last day's sales, with four o'clock as the cutoff time. There was even a general target level of 2-3 units per SKU (that would be model - color - size) at each shop.

Didn't work. As I reported earlier we had surpluses and we had shortages.

The most annoying sensation for me was, after finding myself telling 3-4 different customers that we have run out of the blue dress in most sizes, but was have the black one in all of them and have the customer explain the blue is what she wants and thank you very much, good bye. After all that, the next day we get 3 more of the black dress in a variety of sizes. So the shortage stays and the surplus grows. Now I'm not trashing on the warehouse, they do the best they can. They don't have what we need so, trying to be helpful, they send "the next best thing", something similar, something close enough. But it isn't. 

Sometimes it just became macabre. The day after the new manager finished cleaning up the back room and sending some of the surpluses to another branch (took her 3 days), we got at least one of those surplus models again in all colors and all sizes. We were really low on it.....

My point of view now is that daily replenishment is a great way for minimizing inventory, but you can implement TOC replenishment without it, it is not the most important thing. As long as your central stock is out of whack with the market, nothing's going to help in the long run, this should be the most important factor of your supply chain analytics and your supply chain risk analysis. This is felt much more profoundly in a small country like Israel, where there is simply not enough clout for the "big numbers" rule (that's the statistics rule that shows that if you have a big enough sample everything ends up looking like a normal distribution) to take effect.

Insights from the shop floor

I spent the month of June working as a sales clerk for a local fashion retailer. Not a very big one, but still respectable enough. During that time no one in the chain was interested in what I know about TOC and the chain was not doing any kind of TOC project. It was an entry level job and I was judged only by my ability to ring up the register, and using that parameter I was no star. I'm OK with that, not everyone has to be great in everything and a smart employer will get the best from each employee, even if it means shifting the person around a bit.
Anyway, I found myself working in one of the bigger stores in the chain, one considered a "flagship" store. Yet we sailed through some very rough waters during that short month, with me joining the ranks after most of the sales force quit and the store manager, who turned out to be the reason for all the turbulence, was removed from her position during my second week. Working the floor was a great opportunity to validate and better understand TOC's supply chain management application, presented in the supply chain management book "Isn't It Obvious".

Before I dive into the TOC point of view, some info about the shop is appropriate. The shop is located in a closed mall, this was one of the first closed malls in the Tel Aviv area, built some 20 years ago. The mall has been updated with a movies mega-plex a few years ago, but I think this was not enough to shake off its outdated image.
As for the shop itself - while this shop is considered large within the chain, it is actually quite small, perhaps even very small if you compare it to shops in the US. The shop has a tiny back room to store any inventory that is not needed for display on the floor. The shop's floor is small in comparison to the variety on hand and not all sizes are put on display, some are only available if you asked a clerk to get them from the back room.
The first things to come up were the fact sales were going hard, even though the shop was either in promotion or sale mode the entire month. The second was that we were constantly telling shoppers that we're sorry but the wished for item, in the wished for color and size, has sold out. We would always offer to check and if it is available at another store - the customer can pay for it and we'll call her after it arrives at the store. Some of them did, which only goes to show the competition isn't any better.
I had befriended the one sales clerk who had managed to survive through the turmoil created by that shop manager, so I checked with him if the theory works. A theory is tested by its ability to predict. I used the TOC theory to "predict" what had happened in the store at the season's start, weeks before I joined. As I "predicted", sales were easy and merchandise was flying of the shelves at full price.
This should come as no surprise. At season's start stocks are full, all the high runners are available at all sizes, so sales are easy - with customers easily finding garments they like at the right size they often convince themselves into buying. The sales force is free to help the customer find more items and "deepen" the sale.
By the time I arrived at the store, though, the negative effects of buying according to forecast, predominant in global logistics and supply chain management, have reared their ugly head. The cream of the crop, the best models, have been totally sold out. The fast runners we did have were running short on their best sizes and we had way to much of some dead weights  So even with lowered prices sales were hard and the sales force concentrated on convincing customers they should indeed buy. Sales were "shallow"  with most customers buying only one item.

If you want more details about the damages of working to forecast and the TOC approach to replenishment, here are a couple of links to the "Big Brand" case study:
First is Dr. Goldratt's report on the process (which can also be found in "The Choice", the best book on supply chain management) can be obtained here (just fill in the details and you'll get the PDF in your mail)
Second is the IDEA report on the project, this one includes graphs showing the progress made.

To be continued ....