Identifying Repeated Work Across Engagements
A pattern becomes a product feature through four stages: recurring engagements, a selected pattern, a generalized version designed with the product team, and a build into the core product
Two Engineers With the Same Year
Imagine two engineers who both spend a year doing excellent client work. Each one ships solutions, keeps customers happy, and learns the domain well. Yet over time their careers can drift apart. The difference is not how hard they work on individual jobs. It is whether their work also changes the product that every customer uses.
This article follows one idea that makes that difference concrete: taking a lesson from the field and carrying it back into the shared software. The source frames this as paving a highway. A single customer's custom path through a problem is like a dirt track. It gets someone there once. When many customers keep needing the same path, it becomes worth building a road that anyone can drive on without special effort.
From Repeated Work to Built-In Feature
A pattern becomes a product feature through four stages. You begin with the engagements you have already worked on. Among them, you look for a recurring pattern, meaning a piece of work that keeps reappearing. You then turn that specific pattern into a generalized version that is not tied to any single customer. Finally, that generalized version is built into the core product, so future customers receive it without needing a custom effort each time.
Read the diagram as a one-way path. Many individual engagements feed the search for a pattern, but only one pattern is selected for generalization. The generalized version then becomes part of the product itself. That is the point where the work stops belonging to one customer and starts belonging to everyone.
Counting Distinct Customers
The selection rule in the source is concrete: pick the pattern from your engagements that has come up at the most customers. Counting is the whole method. The rule does not ask which pattern is the most technically impressive, or which one you enjoyed building most. It asks which one recurs most often across the people you served.
The word that matters is distinct. Count customers, not occurrences. A pattern that showed up five times for one customer counts once for that customer.
Choosing a Candidate From Six Engagements
An engineer has completed six engagements for four customers. Each engagement involved one or more of three recurring patterns. Which pattern should be chosen for generalization under the source's rule?
- 1List the patterns and the customers where each appears
CSV column mapping appears for customers A, B, C and D. Custom report export appears for customers A, C and D. Invoice date parsing appears for customers B and D.
- 2Count distinct customers, not engagements
CSV column mapping shows up in several engagements, but it is counted once per customer, giving 4. Custom report export gives 3. Invoice date parsing gives 2.
- 3Apply the selection rule
The rule picks the pattern with the most distinct customers. CSV column mapping has the highest count at 4, so it is the candidate for generalization.
CSV column mapping is selected, because it has come up at the most distinct customers. The rule does not depend on which pattern was most interesting to build.
Senior Track and Staff Track Contributions
The assignment's most important idea is that the contribution to the code separates two career paths. Both paths involve doing customer work, and the source describes both as shipping engagements. The difference lies in what happens after the engagement ends. On the senior track, the engineer keeps shipping engagements. On the staff track, the engineer also takes patterns from those engagements and pulls them into the product.
Read the two columns side by side. Both tracks share the engagement work in the left-hand position. Only the staff track has the extra step of moving a pattern into the product. This is why the assignment is not a demand to become a different kind of engineer. It is a demand to make one additional contribution, the one that carries your knowledge into the shared code.
| Aspect | Senior track | Staff track |
|---|---|---|
| Engagement work | Keeps shipping engagements | Keeps shipping engagements |
| Pattern from engagements | Not pulled into the product by this role | Taken and generalized for the product |
| Contribution to the code | Customer solutions | Customer solutions plus shared product code |
The only added contribution on the staff track is the movement of a pattern into the shared code.
Mistakes When Choosing a Pattern
Common mistake Choosing the pattern you enjoyed building most
Why it failsThe source's selection rule does not ask which pattern is most impressive or most enjoyable. It asks which one has come up at the most customers.
Do this insteadCount the distinct customers for each recurring pattern and select the one with the highest count.
Common mistake Counting engagements instead of customers
What it looks likeA pattern that appears in five engagements for a single customer is counted as five engagements.
Why it failsThe source's rule says to count distinct customers, so repeats for one customer do not add to the count.
Do this insteadCount each customer once per pattern, then compare the totals.
Common mistake Treating the generalized version as a custom fix for one customer
Predicting the Selected Pattern
What do you think happens?
Four candidate patterns each appear in different numbers of customer engagements. Before reading on, which one should be generalized?
Reveal answer
The pattern that has come up at the most distinct customers
The source's selection rule is to pick the pattern that has come up at the most customers, which means counting distinct customers. Technical complexity, personal enjoyment, and raw engagement totals are not the rule.
Think of three pieces of work you have done or seen for different customers. Which one has recurred most often across distinct customers? Write down the customer count for each, then explain why that pattern would be the candidate under the source's rule.
Hints
- List each recurring piece of work on its own line.
- For each line, count only the customers, not the number of times you did it.
- The candidate is the line with the highest customer count.
Key Takeaways
- A pattern becomes a product feature through four stages: recurring engagements, a selected pattern, a generalized version designed with the product team, and a build into the core product.
- The selection rule is to pick the pattern that has come up at the most customers, which means counting distinct customers rather than occurrences or engagements.
- Senior-track and staff-track engineers both ship engagements, but only the staff track also pulls patterns from those engagements into the product.
- The source reports faster promotion for the staff-track group in the labs and better vertical AI startups, but gives no figures, so that comparison is the only claim it supports.