The Fundamental Reason Why You Fail to Deliver on Time (Plus the 3-Step Product Management Guide to Meeting Your Commitments)

Businesses across the globe fail to deliver on time. In the early days of prison warfare, a deadline meant a line that was drawn around a prison to deter prisoners from crossing. If they did, they’d be shot dead on sight! Nowadays, although you won’t get shot if you run over a deadline, it might still have a severe impact on your business. 

In fact, 4 of the top 6 challenges faced by companies, as outlined in the 2020 MHI Annual Industry Report relate to faster delivery and higher customer expectations, including customer demands for lower costs (51%), customer demands on response times (48%), rising customer service expectations (47%), and customer demands for customization (41%).

In times of a saturated market and fierce competition, it’s critical to meet these expectations and deliver on your commitments. So, if you’re still trying to get on top of overdue tasks, it’s worth taking a step back and analyzing the root cause of failing to deliver on time.

The Main Cause of Failing to Deliver on Time

Тhe most common reason behind failing to meet your commitments and deliver on time is setting unrealistic expectations. 

A case study conducted by Calleam Consulting demonstrates how inaccurate estimates led to the significant delay of the completion of the Baggage Handling System of Denver International Airport during the 1990s. The unreliable predictions were off by 16 months, a delay which cost the city of Denver an eye-watering $560 million. Just two decades later, the system has been decommissioned.

This is just one of the many projects that failed due to unrealistic deadlines. 

The Sydney Opera House was supposed to be completed in 1963, for $7 million. A scaled-back version opened ten years later, in 1973, with a final price tag of $102 million.

Managers fall victim to our inbuilt tendency to underestimate how long it takes to complete our work. And how can we possibly predict our exact delivery times with the unpredictability of knowledge work? Are we actually capable of foreseeing the exact time that a feature will take to go through the whole process while handling the rest of our work in progress at the same time? Do we know for sure that there won’t be any additional work coming in between, any dependencies, defects, bottlenecks, or external blockers that might cause a delay?

The problem is rooted in the expectation of answering our customers with a single certain delivery date. Yet, that tends to be the norm in practice. It seems that we’d rather come up with a single certain commitment that ends up being wrong than express any uncertainty.

Pursuing super-ambitious goals may have a far-reaching negative impact on your employees as well. Here are just a few of the ramifications that come with it:

• Burnout & low-quality results. If the expectations placed on your employees are too high, they will be forced to work harder to catch up and their performance will eventually tank. If they’re in the throes of full-fledged burnout, your employees will no longer be able to work efficiently, and their work will be less than stellar. They will rush tasks and cut corners, which will cause mistakes and lead to poor-quality outcomes.
• Missed delivery dates. Impossible deadlines mean it’s unlikely that your team will be able to achieve them, and so they will often miss targets. If they regularly fail to deliver on time, the question on the table will be “Why is your team underachieving?” while they have been forced to push themselves to their limits to meet unrealistic expectations.
• Low morale. Meeting a deadline is a motivation booster. However, if your team constantly misses their deadlines, your entire team can feel like they’re not achieving. Your employees need to know how their work brings value to your company. They need to see how their efforts contribute to business success. If they don’t, this can impact their self-esteem, motivation, and engagement, which would damage your company’s bottom-line.
• Higher staff turnover. Voluntary turnover is a problem that’s costing the US economy $1 Trillion per year, according to Gallup. And this problem is self-inflicted by organizations. If staff feel like they can never and will never be able to meet your expectations, they may realize their only option is to resign. Not only do you lose up to two times their annual salary in turnover costs, but all their expertise and knowledge of your company walk away through the front door too.

Setting realistic expectations and meeting customer’s requirements have to be a priority for every business. Let’s explore the steps you can take to achieving more predictable results without overburdening your workforce. 

The 3-Step Product Management Guide to Meeting Your Commitments

To be able to deliver on your customers’ expectations, you should set and manage realistic goals in the first place. Here is our 3-step product management guide on how to meet your commitments and achieve a more predictable delivery workflow. 

1. Stop Estimating and Start Forecasting

Knowledge work is notorious for its unpredictable nature. Making accurate future predictions allows organizations to define service level agreements with more confidence and deliver value to their customers on time, in a consistent, predictable manner.

One of the most pressing questions every manager faces is “When will this be done?”. And it might be a challenging question if you don’t have the means to give a confident answer.

David Anderson Leader of the Kanban movement "We spend too much energy speculating about the future, rather than studying the recent past and using factual data. We are too ready to believe that our newest work is unique and different when in truth it is much more similar to work we do regularly than we care to admit."

When trying to provide the answer to the “When will this be done” question, there are two approaches you can take – an estimate and a forecast. What’s the difference?

Estimates are predictions based on intuition, guesswork, or judgment. The prediction is communicated as a single value, and it doesn’t involve any probability of its occurrence.

Forecasts, on the other hand, are based on past performance data. The prediction is delivered as a range of values and the probability of those values occurring.

The Cycle Time Scatterplot can be of great help when it comes to making probabilistic forecasts for a work item. This diagram represents your historical performance data, displaying all of your completed tasks as dots scattered on a plot.

The dotted horizontal lines stretching across the graph are called percentile lines. We use percentiles to define the probability of different commitments being met. 

For example, the 85th percentile on our scatterplot points to 10 days. This means that 85% of the tasks so far have been completed in less than 10 days. We can now say that any future task we take on has an 85% chance of being finished in less than 10 days. We can also say with 95% certainty that we can deliver any new work within 13 days. 

You can also filter your data by classes of service. Using Classes of service (CoS) is an approach to prioritizing your work effectively. It is highly likely that the 85th percentile for Standard CoS comes with a different cycle time than the 85th percentile for Expedites. That way, you can provide different Service Level Agreements (SLA) for different work items you’re committing to. You can do the same by filtering your data using types of work items like Stories, Tasks, or Bugs. The results will still be valid.

Of course, the percentile you use to define SLAs largely depends on your context. You and your clients may be comfortable settling for an 85% level of certainty, or, you may both need a confidence level that’s higher than that.

The Cycle Time Scatterplot provides a range of commitments, along with a probability of each commitment happening, by using your past performance data. Probabilistic forecasting is amongst the most effective methods for predicting delivery times with maximum accuracy.

2. Start Your Work on Time, Neither Too Early nor Too Late

The best chance for delivering on time comes when you make sure you start on time. When you start working on your assignments too early, you are taking from the time that could be otherwise spent realizing other business opportunities. You’re both wasting capacity on work that is not yet due and running out of capacity for work that is due.

When you start too late, though, you risk delaying your work and breaking your commitments. The longer you wait, the higher the chance of a delay. So when is the best time to start?

First, you need to look at your probability forecast for the specific class of service you’re interested in. Here is the forecast produced by the Cycle Time Scatterplot for tasks of Standard CoS:Now that you know your expected delivery times and the probabilities that come with each of them, the ideal time range to start your work is any time between the 99th percentile and 85th percentile.

99% > start > 85% before the delivery date

Let’s say you need to deliver on 30th May, and your forecast says there is a 99% chance to finish a Standard task within 23 days and an 85% chance to finish it within 11 days. This means that the ideal time to start your work is between 23 and 11 days before the delivery date, or any time in the range of 7th May – 19th May.By initiating your work within the “Normal” start date range, you’ll have at least an 85% chance of delivering it on time. 

Following this approach, you’ll observe over time that those 11 days already point to your 95th percentile, and the cycle time that the 85th percentile represents is going down. By shifting all the percentiles down and keeping those lines close to each other, you reduce the variability in your system and improve your performance.

3. Manage Your Work in Progress

As the start date approaches and you begin working upon your commitments, keep track of the age of your work. Tools such as the Aging Chart offer a detailed overview of where your tasks are in your process and how much time they’ve already spent in progress.

The colored zones draw the timeline of how your tasks have advanced in the past. For example, the green zones show the times that 50% of your previous tickets have spent in each process state. By observing how your current work is moving through the zones, you have a pretty good chance of meeting your commitments. Make sure your work items don’t cross the percentiles you use to define your SLA. The higher the dots, the larger the chance of delay. 

We recommend taking a closer look at the tasks that move to the yellow zone. These tasks have already spent more time in your process than half of the work items completed so far. 

If you’ve committed to the 85th percentile and your work item just moved to the orange zone (crossing the 70th percentile), don’t fall into the trap of cutting from the scope or sacrificing the quality of your work to be able to deliver it on time. Instead, in order to make sure you keep your commitment, expedite your task.

By expediting a task, you’re effectively suspending another work item with lower priority that’s currently in progress. Essentially, you borrow time that would otherwise be reserved for another task in order to fulfill your commitment. Keep in mind that expediting work generates flow dept and make sure that the short-term benefits are worth it in the long run.

Achieving 120% Due Date Performance Improvement in Less Than 2 Months

Your due-date performance is a measure of reliability. Furthermore, it’s a factor determining the quality of your initial forecast. The accuracy of your forecasts strongly depends on the stability of your system. Stable systems produce more accurate forecasts.

You can evaluate your due date performance by comparing the time between the start date and due date of a task with the actual time it took to deliver your work. The Due Date Performance Chart can be of great help here.

For each work item, the Due Date Performance Chart visualizes the predicted time vs the actual cycle time and plots a regression line of your performance. The predicted time is calculated as the difference between your commitment point (the moment when the work entered your workflow) and the due date of the ticket, while your completion time is the difference between the same starting point and the ticket’s actual done date.

The line plotted through the chart is called a regression line. It runs roughly through the middle of all the data points. The slope of the line is the correlation between your estimated time and the actual time you spent finishing your work. It helps you assess the accuracy of your predictions and evaluate how efficiently your system is running.

The most important part of analyzing your due date performance is assessing whether your work has been delivered on time. The On-Time Delivery widget will display the percentage of times that you managed to deliver before your due date approached and how such trends have developed. For example, if you released 10 features today and 4 of them were overdue, your due date performance for that day would be 60%. 

Following the product management guide steps, we managed to increase our due date performance from 33% to 74% in the period from July to August. We improved our on-time delivery by 120% in less than two months. By switching from estimating to probabilistic forecasting, starting our assignments on time, and actively managing our work in progress, we eliminated our primary source of delays – setting and pursuing unrealistic deadlines.

We have been producing more accurate future predictions and keeping our commitments ever since, which has ultimately resulted in higher customer satisfaction.

Avoiding the Black-Swan Risk

Getting rid of uncertainty and having absolute control over everything is just impossible. The real challenge is dealing with the uncertain and unknown in an effective way. That is firmly located in the realm of probability.

If you have the data, use the data. If you do not have the data, then collect the data and use the data. How many work items do you need to first complete before you can get started? The answer is not many, probably around 20. The emphasis here is on quality, not quantity. The more stable your system is, the less data you need to produce accurate results.

Following our 3-step guideline will help you achieve more predictable and consistent results over time. Most importantly, you will meet your customers’ expectations by delivering on time, every time.