imple chain is basically the sequence of steps without forks and loops. That could be a simplified procedure or a recipe: do this, then this, then this. Each step of the process usually awaits certain inputs/ingredients/prerequisites and may require certain tools to perform the operation.
Examples: cooking a berry pie, going through departure procedures at the airport, and renting a car.
The advantage of this architecture is its simplicity, but the disadvantage is that every step is required, and if one of them is missing or becomes a bottleneck, the whole process will be at a standstill.
Research landscape of the Chain
Steps of the process
The steps of the simple chain are usually easy to identify.
The steps of the chain should have the same scale. If you feel like the steps are too small and detailed for the current chain, it may be reasonable to group several steps together as one macro-step. In this case, a step of the chain can contain a “sub-process” inside.
Example: when you rent a car, the macro-steps could be “choose a car”, “fill out the forms”, “make payment”, and “pick the car from the parking lot”. But the “fill out the forms” step may have a sub-chain inside: “fill out the personal information form”, “fill out the insurance form”, and “fill out the payment method form”, which in turn can have even more detailed sub-chain of “specify the credit card number”, “specify the credit card expiration date”, and so on.
Ingredients/prerequisites are the most non-trivial part. The steps are usually clear and observable, but each step may require very specific inputs. Some of these inputs are the outputs of the previous step of the chain but some additional resources or actions may be required. That can cause problems. For successful process execution, these ingredients must satisfy the criteria that you need to explore. How diverse is the range of acceptable inputs? How often does an inappropriate ingredient come into the process? How does the process suffer from an inappropriate ingredient? What types of ingredients are ideal for the process? Why? Can the process be changed to become more tolerant and consume a wider range of ingredients?
Example: When you cook a blueberry pie and follow the recipe, will it be acceptable to use cherries if you run out of blueberries? Will the end result be delicious enough? Is the process of renting a car with the steering wheel on the right side exactly the same as renting a car with the steering wheel on the left side? How about an electric car vs a hybrid one? Or, a sports car vs a truck?
Some steps of the chain may require external tools or resources to be utilized. Some tools may be expensive to own or maintain, and if these tools are not kept in good working order, they may produce defects and waste the ingredients. All these expenses must be taken into account when evaluating the cost of the chain.
Example: When you cook a blueberry pie, you need an oven.
The chain typically has a weak link or a bottleneck that causes problems. Thus, the goal is usually to identify several weak links and vulnerable steps of the process and understand if they may be replaced with more robust alternatives. Sadly, once you fix the weakest link, the system usually breaks on the second-weakest link. Your efforts to create a state-of-the-art node of the chain therefore do not improve the whole chain to that level, and may not pay back what you put in. You don’t need to replace a bad step with a perfect one. You need to replace it with a good enough solution that is just better than the other remaining weak links.
Possible flaws that make a week link:
- Low throughput
- Expensive processing or ingredients
- Low tolerance to the ingredient variation
- High failure/defect/waste rate
- Poor availability or condition of tools required to execute the step
Improvements may include re-engineering the step or a subsequence of steps so that they have better throughput and are more robust and tolerant to the ingredients variability.
Fidelity and Quality
The High-Fidelity solution cures most of the listed illnesses. The Low-Fidelity solution aims at resolving just one of those at a certain level. The Medium-Fidelity solution aims at solving several problems with the highest impact.
The High-Quality solution operates 24/7 and is able to execute the process smoothly and consistently. The Low-Quality solution requires constant attention and tinkering in case something goes wrong. The Medium-Qualitysolution looks nice but sometimes breaks and may require some attention, though not frequently.
Leveling guide: Simple chain
If you are designing and building a process from scratch:
Create Placeholders of all steps of the process. The chain has zero throughput if at least one of the steps is missing. That means you need all of them before you start investing in one specific.
Upgrade all steps of the process to the Quick & Dirty level (Low Fidelity, Medium Quality). Low Fidelity means that the steps may have low tolerance and work with a narrow range of ingredients. Medium Quality means that if the ingredients satisfy the criteria, this step will be completed not at the best possible speed and efficiency but still successfully.
If you are improving an existing process, skip phases 1 and 2 and go to the next phase:
Target discovery: identify several bottlenecks (weak links of the chain) — the steps with low throughput, delays, high inaccuracy or defect rates, expensive operations or ingredients. These are your targets.
Explore and create several experimental Quick & Dirty (Medium Fidelity-Low Quality) alternatives to the existing target steps. These alternatives must be less expensive or more robust than existing ones.
Once you proved one of these prototypes has potential, upgrade it to a Good Enough (Medium Fidelity-Medium/High Quality) level, pick another alternative if this effort did not work out.
Don’t overinvest and don’t try to create a State of the Art (High Fidelity-High Quality) solution for the weakest nodes of the chain. There are other bottlenecks that will limit the throughput of the entire process. So start experimenting with the State of the Art solutions only when all main nodes are already at the Good Enough level.