Noise Makers (Part 2)

 

In some engagements, project teams will encounter groups that do not want the project to be successful for one reason or another. If they believe strongly that the project should fail, they are likely to either withhold information or deliberately inject noise into the project (Intentional Noise Mitigation).

In the event that the project must continue regardless of opinions, project teams must map out stakeholder opinions and expectations and tasks need to be planned and executed accordingly to mitigate at-risk stakeholders.

ADVERTISEMENT

Trending Articles

Why Project Managers Need to Push Back by Andy Jordan June 3, 2026 How much do you challenge the directive? If project managers are always going to go along with what they are asked or told to do, then there really isn’t a lot of point in them being there.

Beyond 'Done': Reclaiming Purpose in Agile Delivery by Lissette Pimentel May 29, 2026 “Done” is one of the most powerful words in agile. It brings closure. It marks progress. It gives us the satisfaction of checking a box and moving on. But what happens when “done” becomes hollow, and the value is nowhere in sight?

33 Years of Volunteer Impact by Deborah O'Bray May 28, 2026 Thinking about volunteering? A longtime practitioner reflects on her 33-year journey that began after attending her first conference in 1992. Through her many roles, she gained invaluable skills, friendships, and opportunities, and encourages others to make a meaningful impact.

Options to consider when dealing with difficult groups:

• If you need information, look for people who recently departed that team.
• Bear in mind that not everyone on the team may share the same opinion.
• A trusted third-party may succeed where the project team cannot.
• Groups that have regular interaction with the problem area may have significant insight into the issues in question.
• If the problem is in one area and processes are standardized, then look to other areas and/or shifts for more information.

Unintentional Noise Mitigation

There are times that you will need to generate accurate information based on the content you can pull in and others where you just need to bear in mind that noise can and does exist. The main theme is to recognize that the noise exists and create a system that both recognizes that fact and compensates accordingly. The following are potential solutions for mitigating noise depending on the situation at hand.

Multi-Source Confirmation. Use more than one source for your content, attempting to select sources that offset each other. In other words, select content sources such that the overall portfolio reduces the level of noise. If one person has a belief that affects his perception, then solicit input from another source that lacks that believe and see if the second source confirms the report from the first. You may even include inputs from additional parties depending on cost, timing, sensitivity, etc.

Pattern Analysis and Algorithmic Prediction. Look for patterns in the content. If you receive content that traditionally goes "1, 2, 3, 4, 5…" and all of a sudden, you get "22, 0, 11, 33…" then something must be wrong. Also, if you take the first sample population and then compare it to "1, 2, 3, 7, 5, …" then you may be able to infer that the "7" is in error because the pattern is typically X+1, not X+4.

Statistical Normalization. By using statistics and content that can be quantified, a person can attempt to drop statistical outliers and determine sets with a certain degree of accuracy. For example, "1, 22, 23, 22, 23, 24, 23, 22, 23, 2999." Here, we can easily look and question the validity of the "1" and the "2999" because they lie so far away from the sample mean. However, we must also consider the sample size relative to the population and question if the sample reflects the real world. It may be that there are tens of thousands of values per second and that the 2999 score is the real mean and all of the other data points were brief anomalies caused by a sensor or software design error. Depending on the situation, statistical analysis can aid in the filtering of content.

Constraint Recognition. Identify what forms of noise can enter into the content, document the potential issues and bring them to the consumer's attention. If you are dealing with systems you can create and/or modify, then consider building mechanisms into the system that recognize the constraints automatically. To explain using our earlier sensor resolution, construct software that does not display data beyond the sensor's effective range and/or resolution, plus flag when the range is exceeded. One pitfall to avoid--do not simply drop the seemingly invalid scores.

Omitting scores can be dangerous if in fact there is another factor at play. For example, if you have a sensor that reads up to 1,000 pounds and you periodically get scores that exceed that range, then what should you do? If you drop the data, then you may not recognize that a new production process has gone into affect that is causing larger loads to go through the plant and, in fact, both the sensor and software need to be revised to reflect the environmental change.

Summary

We like to think that we are always dealing with data and truths. The fact is that we are always confronted by content that contains noise and our precious concept of truth is relative. Project teams, even stakeholders, need to be aware that noise exists and come up with strategies to manage it and thus reduce project risk.

How do you strive to reduce the noise both in communications coming to you and in your dealings with other people? If you have thoughts to share, then please contact the author at [email protected].

George Spafford is a project management consultant and instructor living in saint Joseph, Michigan. In total, George has more than 10 years of experience in information technology and project management. His areas of personal interest include project management, software engineering, organizational learning and maximizing the value added by information technology to an organization. He can be reached at [email protected].

Related Content