What to do first? Simple prioritization of tasks using rice
Implementation of a project or creation of a product is related to the performance of tasks, testing of ideas and hypotheses. Most often, a huge number of them accumulates, and the original question arises (no, not who is to blame and what to do): what to do first? If no explicit rules are established in the organization, then the “rule of the jungle” will work: the one who shouts the loudest is right. Accordingly, the tasks of the customer who has an advantage in voice and “corporate weight” will be given the highest priority.
In this article, I will tell you how we tried to make the prioritization rules clear and without a knife fight, as well as a simple method RICEwhich is based on the point system.
How we invented the bicycle, once again…
As fate would have it, I landed a position as a program manager for projects that covered changes to a large chunk of the company’s critical operational activities (bloody enterprise). With what we entered the program of projects, where a huge part was occupied by the “Automation” project:
development tasks were set by several customers at the “general minus one” level and their “officers”;
there were no prioritization rules for the customer, the priority was set by the development team after a quick assessment and approved or not approved at special meetings;
the degree of description of the business requirements of the accumulated development tasks was radically different: from simple names to detailed requirements, which often did not take into account system limitations or dependencies on related processes and systems;
the specialized division of business analysis did not participate in the process of setting tasks (don’t ask why – this was the case historically at that time).
the system was legacy and removed from support by the manufacturer;
the incoming number of tasks was more than 2000 and was constantly replenished.
As you probably guessed, the meetings on prioritization and at least some arrangement of the list of tasks were often held in high tones and did not lead to any result.
Something had to be done with the mountain of tasks, and we came up with nothing better than to invent (where are my tautology pills) our own system of assigning priorities: “High”, “Medium” and “Low” – the rules for assigning these priorities and the corresponding sequence of execution . The priority was assigned based on the level of the task (strategic, regulatory, tactical and other), the required implementation period, the team’s assessment in working hours, the status (at which stage of the technical process the task is) and several other criteria. “High priority”, of course, was assigned to regulatory tasks without discussion, but then the game of semitones began: strategic tasks could fall into the second place, and tactical tasks – into the first or third, and with epics that were not decomposed, it is not at all clear what had to do, so they were evaluated on a par with tasks. By the end of one hour of such priority meetings, the effectiveness of the review naturally dropped to almost zero values. And the cost of such fees, as you could guess, was cosmic. There were a couple of such approaches, and then we decided to plan monthly iterations around the short-term goals of business customers, and thereby quite organically rolled out a series of tasks in about 7 months. But it was also possible otherwise.
What is rice for?
The point system helps in prioritization. The assessment of each task ultimately allows for the formation of a consistent plan of action. I propose to consider how to conduct an assessment using RICE on specific examples.
Tasks are evaluated according to the methodology based on four factors:
The first letters form the abbreviation of the name of the technique – RICE.
We define reach (Reach)
What is “coverage”? In simple words, how many users will be affected by the software change. The factor is measured by the number of people/events per period. Moreover, very exemplary grades taken from the ceiling will not do. Take only real data tied to specific product metrics or an already known number of users working with a specific functionality or software module.
example. In our specific case, where the software is already running, and the number of users is unchanged, for a unit of time, we will take a quarter, the number of users is 680, and some number of events:
1. 680 users will pay 3 times a month. So Coverage = 680 * 3 * 3 = 6120.
2. 680 users perform an action in the software once a month. So Reach = 680 * 3 = 2040.
3. 680 users will be affected by the change in software 1 time, and there will be no further result. So Coverage = 680 * 1 = 680.
We enter all the results into a summary table.
We define impact (Impact)
A difficult to measure factor in the methodology: because it is an objective assessment. But such an assessment is still better than a biased attitude towards tasks or many hours of breaking copies. Consider the impact of each specific customer or user or group of users. To quantify the impact, it is suggested to use a multiple choice scale:
An adequate assessment of the impact is extremely important, because the established value multiplies the final result.
example. In our specific case:
1. Each user will have an average impact. So, the score is 1.
2. In the second option, every user will be massively affected by the influence. So, the score is 3.
3. In the third option, there will be minimal impact on each user. So, the score is 0.25.
We also enter all the results in the summary table in the “Impact” column.
We determine the level of confidence in estimates (Confidence)
Tasks and, especially, ideas are interesting and promising, some at a superficial examination seem to be created under hallucinogenic mushrooms, but there is no data to confirm. In order not to get involved in an adventure, determine the level of confidence in the estimates. Confidence is indicated as a percentage. 100% is high confidence, 80% is medium and 50% is low. Everything below 50% is an illusion, which is better to abandon immediately and not waste time or money on them.
example. In our example:
1. In the first case, we know from historical data that our estimate can get 100% for Confidence.
2. In the second case, the situation is similar to the first case: Confidence – 100%.
3. But with the third case, not everything is so concrete. The Confidence parameter here will get 80%.
We translate the percentage expression into natural numbers, since it is easier to calculate the final score according to the formula. Therefore, the consolidated table will look as follows.
Define effort (Effort)
We will have to calculate how much time it will take to implement. In the RICE method, this indicator is measured in months. If the implementation of the task will take 1 month, the table, accordingly, indicates 1. Of course, it is difficult to determine the time costs with high accuracy to the day, but it will be possible to calculate the approximate values.
example. In our specific case:
1. In the first version, the team estimated the effort for 3 months.
2. In the second option, the estimate was 1 month.
3. And the third option also received an estimate of 1 month.
Enter the values in the summary table.
We consider the final score
The final value is calculated according to the formula: (Reach x Impact x Confidence)/Effort. What do dry unbiased numbers tell us:
1. In the first option, the estimate will be = 6120 * 1 * 1 / 3 = 2040.
2. The estimate of the second task will be=2040*3*1/1=6120.
3. According to the formula, the third option will receive a score = 60 * 0.25 * 0.8 / 1 = 136.
Let’s look at the overall picture
The final indicator shows the importance of tasks. The highest indicator – 6120 – indicates that the third task will bring more benefit during the work time, so we implement it first. After evaluating all tasks, you can sort the considered tasks in descending order and once again review the received evaluations. Do they seem overpriced or underpriced? If there are doubts about any tasks, it is necessary to review the assessment of each parameter and add new introductory ones to the matrix.
As a conclusion
Obviously, the RICE assessment is not a panacea. Often, in real production practice, it is necessary to first implement tasks with a low priority due to various circumstances. And, of course, for various reasons, some tasks will lose their beauty contest time and time again. However, RICE is one tool that allows you to perform an assessment fairly quickly, and time is money, as you know. And it is better to take an existing proven rule than to spend precious time inventing your own method.