/ artificial-intelligence

The Need for Modular Data

If we consider Human intelligence, in regard to progressing sales of tires through the multi-distribution locations, we see a bias toward absolute sales (in terms of numbers of tires). The existing evidence supports the position that humans looks at getting the sale at a fixed price with no consideration of using price manipulation to moderate the demand.

Examples of this are:

Example 1

A single specific tire is available, in a product with a substantial demand. The likely buyer will be for a user who needs precisely that tire and the tire is most likely needed to make up a set (why else buy one tire?).

If the tire sold for a moderately higher price, it is certain that the sale would go ahead and the Buyer (Dealer) would be satisfied and not concerned that the price was moderately more than they traditionally pay.

If we believe that, managing the Dealer Relationship is important for the long term good.

Example 2

Products with high demand that do not show at the Dealers Default Warehouse but do show at a nearby warehouse. The tire sale will be priced at the regular price for that Dealer. Regardless of the cost price, of the tire at the nearby warehouse.

To make this clearer, if the part had a cost at the default location that the dealer buys from of $1 and a cost price at the nearby warehouse of $1.5 there is no adjustment to the Dealers buying price (the price they see is still based on the cost of the tire being $1 not $1.5).

The reason, I mention this is to demonstrate that the process we are looking to implement will be more productive than typical human consideration in this matter.

One of the most significant developments that must be built to utilize the AI and allow the ability to use price moderation is a more sophisticated data source.

The existing Tireweb displays a sell price based on the Dealers buying price from the Dealers default warehouse and shows a second column for inventory available from alternative warehouses. This process ignores the variant in buying prices at both locations.

The complexity of the Cost Price Issue can be seen in this example.

Example 3

A Dealer is looking to buy 20 tires, there are 5 tires as the default warehouse that the warehouse bought for $50 there are 20 tires at a nearby warehouse that this warehouse bought for $65.00. The Dealer sees their price for buying the tires at $60 and orders 20 tires The total cost of the tires to the distributor is

Amount Price Total
5 $50 $250
15 $65 $975
- Final Price: $1225

The income is however only $1200 (20 X $60)

This is a serious issue and not an anomaly. It is often that a warehouse has a product that they bought at a cheaper price in bulk and took several months to sell, while the second warehouse may have recently bought a smaller lot and after a price increase from the supplier, making the smaller lot significantly more expensive.

There are several scenarios that we can state with consequences ranging from potentially lost sales where the Default Warehouse has a highly priced product and the supplementary warehouse has substantial stock but at lower prices (The Dealer sees an excessive price compared to normal market expectations), through to lower profit or even profit loss.

The example is based on the nearby warehouse being commonly owned and having an agreed cooperation. However, there are other situations where this warehouse is owned independently (what we call foreign Warehouse), or the warehouse belongs to the Manufacturer. Let's remember the name "Foreign Warehouse" in this discussion.

A warehouse that is commonly owned gets no reward from cooperation with associated warehouses, in terms of the profit and sales metric that it is most likely measured by. Because cooperation is two way the reward is a product of access to the other warehouse and the penalty is the fact that its own warehouse inventory is being made available externally (sales made through the other warehouse grant no benefit).

Reward

When we consider a Foreign Warehouse, the reward is built into the sale. For these warehouses they are fulfilling their purpose and making an external sale with the subsequent profit margin.

A stock transfer from a commonly owned warehouse depletes that warehouse inventory without fulfilling the purpose of making an external sale.

Transfers between commonly owned warehouse are best negotiated. The supplying location may lose an opportunity, and perhaps that opportunity is more valuable to the common owner than the gain from the sale itself.

My own personal experience is worth considering here as a Warehouse Manager. A bulk of OTR tires was promised to a local client that had an excellent payment record and had placed forward orders for the product that would clear the inventory within 6 weeks. An account through another warehouse requested the stock that would be sold at a lower price, with the added burden of a high shipping cost and the account buying the tires had a poor payment record. The common owner of both warehouses decided the transaction, after considering these multiple factors. As we can see there are many influences that need to be negotiated.

Game Theory

A brilliant idea by Miles Flavel, Connections Center Lead, was to introduce the idea of Game Theory into the way Warehouses reveal their inventory.

The idea is to challenge "Remote Warehouses" (Commonly owned nearby warehouse) , give that warehouse information about the sale being offered to a Dealer and allow that Warehouse to respond with the inventory that it will offer.

If we look at a few examples:

  1. Default Warehouse has 10 of Item A that it bought at $1 and is offering them for sale at $1.2 to the specific Dealer. A nearby warehouse has 100 of the same part and has a cost price of $1.50 and is challenged to show their inventory. In this situation the Nearby Warehouse will respond with no available inventory (It will not benefit from a sale at $1.20 when its own cost is $1.50)

  2. (in reverse) Default Warehouse has 100 of Item A that it bought at $1.50 and is offering them for sale at $1.8 to the specific Dealer. A nearby warehouse has 10 of the same part and has a cost price of $1 and is challenged to show their inventory. In this situation the Nearby Warehouse will respond with 10 (all) available inventory (It will benefit from a sale at $1.80 when its own cost is $1.50)

  3. Default Warehouse has 10 of Item A that it bought at $1 and is offering them for sale at $1.2 to the specific Dealer. A Foreign Warehouse has 100 of the same part and will sell them to the Default Warehouse at $1.50. The data challenge against the Foreign Warehouse does not send the sale offer price (this is privileged information). In this situation the Foreign Warehouse will respond with 100 (all) available inventory.

  4. Default Warehouse has 10 of Item A that it bought at $1 and is offering them for sale at $1.2 to the specific Dealer. A nearby warehouse has 100 of the same part and has a cost price of $1.1 and is challenged to show their inventory. In this situation the Nearby Warehouse will respond with 25 available inventory. This is based on the fact that the Remote Warehouse has an expectation that it can sell 75 parts with its existing price to its clients and in effect 25 of the parts are surplus to immediate need.

  5. Default Warehouse has 1 of Item A that it bought at $1 and is offering them for sale at $1.8 (Note this price is based on the new buy price of $1.50) to the specific Dealer. A Foreign Warehouse has 100 of the same part and will sell them to the Default Warehouse at $1.50. The data challenge against the Foreign Warehouse does not send the sale offer price (this is privileged information). In this situation the Foreign Warehouse will respond with 100 (all) available inventory.

We can see that the data provider is treated differently, they are either Foreign or Remote. Remote data will get data on the sale offer price and will respond with Inventory that it will commit too. Foreign Data will respond with a price and all available inventory.

There is an anomaly here that the remote data will vary it's response if it can influence the sale price:

There is a single tire of Part A at the Default Warehouse with a cost at that Warehouse of $1, with an offer to the dealer at $1.20. The Remote Warehouse has 100 of the same part has a cost at that warehouse of $1.20. The Remote Warehouse will not offer any product with these metrics, however it may volunteer 50 of the product if the sell price was $1.40.

The data response is a separate opportunity. We can see an example where there is one of a tire and the responses. The significance in this is that we send data on the sale opportunity and the process of the data provider decides on how to participate. In the reward discussion the indication is that the data provider is motivated by reward. Foreign Warehouse will always commit to showing their inventory in order to meet their purpose (External Sale). There needs to be a motivation for a "Remote Warehouse".

We see the opportunity to introduce Game Theory. Each warehouse will act independently judging for themselves the data they will reveal.

This discussion is to deal philosophically with an innovative approach. Changing the way, we get data takes some getting used to. In the past data has been static, in this approach it will become Dynamic.

Current Tireweb and Connections Center Data

Tireweb is a data management process. The major component of the system is the retrieval of data from disparate systems and the process of firing orders back to the accounting software.

However, from an administrator's point of view the software is a filtering agent:

  • Create special offers;
  • Grant User Permissions and access;
  • Group Dealers;
  • Apply Make Filters; and
  • Others.

The new approach including AI changes this function dramatically:

  • Users will now be defined by their relationship to AI (influence by make and other aspects);
  • Special Offers will be dynamic;
  • Data will depend on Remote / Foreign affects; and
  • Many others.

Reliable Tire is a good example of the complex nature of this situation where:

  • It may be considered a foreign warehouse by a specific tire wholesaler who purchases products from them; and
  • They would be considered remote to their own warehouses.