Avoiding Expectation Traps

Have you ever reached the end of a sprint, a project, or a review cycle and been amazed to find out that stakeholders are disappointed with the outcome even when your team believes it has delivered on all its commitments? Anyone who has worked in software development for any length of time is probably intimately familiar with this experience. In these instances, the differences in expectations between individuals or organizations create an expectation gap. Many traps (e.g. expensive rework, lost trust, missed deadlines) lie in these expectations gaps and closing them is one of the keys to operational excellence.

Expectation gaps exist everywhere in real life, and once you start looking for them you will be shocked how often they occur. Sometimes expectation gaps result in a positive experience, like when your favorite team exceeds expectations, but usually they result in friction between the parties involved and disappointment. Expectation gaps are caused by many factors, but in the business world the main causes tend to be unspoken assumptions, misunderstood discussions, information siloing, and misleading information.

The first three problems (unspoken assumptions, misunderstood discussions, information siloing) are relatively easy to fix. They all stem from a lack of a single source of truth (SST) [1]. When all sides of a partnership have a single source of truth, they can align to it. Just like any practice however, the more energy put into maintaining a single source of truth, the better the results are and the smaller the potential for expectation gaps. For example, a well-documented API prevents different teams from making assumptions about how the API works or building / using it incorrectly. A detailed set of requirements will eliminate surprises about how a solution works when it is finally delivered. A single source of truth is also the best way to eliminate information siloing, allowing information to flow both horizontally and vertically in an organization.

Misleading information has many causes. Sometimes misleading information is caused by the many types of bias individuals and organizations bring to interpreting data. An awareness of common biases can help reduce the impacts. Diligence and organization wide cultural practices are important in reducing the effects of bias.

Frequently, misleading information is caused by loose language. For example, in many organizations the meaning of ‘done’ can be ambiguous. This ambiguity can create a expectation gap between the speaker and the listener. If a developer says a feature is done, a manager might think they are ready to move on to the next feature. The developer may have meant, however, they have code ready for testing. Having and using consistent terminology within an organization or with partners is critical to avoiding these types of expectation gaps. 

Other times misleading information is caused by team members not updating a ticketing system. This goes back to making sure teams keep your SST up to date. “Oh I forgot” is acceptable the first time, but management must instill the discipline to make updating the SST routine. In the end, this is a self-reinforcing practice as team members understand they will have less asynchronous interrupts throughout the day when they keep the SST up to date.

Often, misleading information is caused by poor communication through a communication chain in which information isn’t routinely interrogated and integrated. These communication chains can be intraorganizational or interorganizational. In either case, each layer (especially management layers [2]) should ensure information integrity by interrogating the data, integrating it with other data received, and refining the data [3] before passing it along (or up) the chain.

Occasionally information is embellished as it moves through a communication chain. In these cases, identification of the root and coaching is required. The value of data integrity within an organization is paramount. And if fact, the higher in the organizational structure an individual is, the more important it is that they be held to the highest standards of information integrity for themselves and their organization. The cost of senior leadership making decisions off bad data is too high a price to pay.

Finally, make sure to close expectation gaps with the right parties. Closing expectation gaps at the functional level is a prerequisite for success but is not sufficient. Expectation gaps must be closed with all Deciders[4], especially the Decider who signs the checks (internally or externally). An executive is not going to be bound by a lower level employee signing-off on a change or solution. And, since they sign the checks, they are going to usually get what they want.  

Tips and Techniques

• Ensure that activities are aligned around a single source of truth that is easy to read and understand. This could be a ticketing system for development activities, a requirements document, an API doc, or any written form of communication that is accessible by all team members.

• Change requests (CRs) are your friend. CRs document who requested a change and the scope of the change. It gives a delivery organization a chance to evaluate the change, assign a cost to it, and assess schedule impacts. Don’t allow any changes to requirements, an API, or other SST without an official change request process.

  • For a customer, this could mean approving additional cost and time or "horse trading” features in scope for a given delivery.

• Ensure all the Deciders for a project sign-off on a change request.  

• Creating a culture of data integrity is important.

  • Coach team members to keep SST up to date.

  • Coach team members to use well defined and consistent terminology.

  • Coach team members to interrogate, integrate, and refine data.

  • Coach team members to eliminate embellishment.

  • Eliminate pass through communication chains.

[ 1 ] A single source of truth is a centralized, authoritative data repository that everyone within an organization accesses to ensure they are working with the same, consistent, and accurate information. - Google Search AI

[2] Pass through chains create negative value. They cost money and add no value. Eliminate them with prejudice.

[3] For engineering geeks like me, a good analogy is a Kalman filter or maximum likelihood filter. By integrating several independent sources of noisy data, a user can generate a remarkably accurate prediction of an actual value.

[4] "Decider" is simply shorthand for a person with whom making a decision ultimately lies. For example, an executive with budget would be the decider for a purchasing decision.