Measuring Productivity

B	C	D	E	D	F	G	H	I	J	K
Release	Applica- tion	Sub- Applica- tion	Deliverable	Incomplete	Document title	Total topics or pages	Wild guess?	Complexity of doc (low=1, medium=2, high=3)	Percent new/ changed content	Writer
STL 8.7	STL	–	PDF	50	STL_rpt guide 87.pdf	212	N	2	25%	Alice
STL 8.7	STL	–	PDF	50	STL_user guide _87.pdf	725	N	1	25%	Alice
STL 8.7	STL	–	WebHelp	50	BLYhelp .htm	540	Y	2	25%	Alice
STL 8.7	STL	–	PDF	50	STL BlY upgrade tool.pdf	8	N	1	50%	Alice

Figure 1. Key data points in the tracking spreadsheet

Relative Productivity Can Be Measured

Using the methods below, we can reasonably assess productivity in three areas:

• Current writer workload relative to the team
• Past performance of a writer
• Future team capacity

In this article, we discuss evaluating writers’ current workloads relative to the team. However, you can adjust the spreadsheet formulas to measure past individual performance or future team capacity.

To measure productivity, we:

• Gather data
• Calculate work units
• Normalize the data
• Account for special projects
• Normalize the data again
• Account for job grade

The basic formula we use is this:

(# topics or pages) × (complexity of deliverable) × (% of change) + (% time spent on special projects) × (job grade)

Let’s break it down.

Gathering Data

The team maintains a tracking spreadsheet for several reasons. The main reason is to track the progress of deliverables against established milestones. Among the inputs to the tracking spreadsheet are the following data points, entered by the writers and verified by the manager, if necessary:

• Number of topics (for a help project) or pages (for a Word document or PDF). Early in the document lifecycle, this number is an estimate.
• Complexity of the deliverable. Our team assigns a numeric value from 1 to 3, although you might develop more nuanced values. For example, we might assign release notes a value of 1 and a technical reference manual a value of 3.
• Percentage of new or substantially revised content. For example, we assign a value of 100% to a document that must be written from scratch; we might estimate a value of 10% for minimal updates to an existing document.
• Special projects. The writers record the percentage of time they spend on special projects. Special projects are an optional measure. Most of our writers volunteer for special projects in addition to their assigned deliverables (for example, updating standards and style guides).

Figure 1 shows an example of the spreadsheet in which the writers enter the data for their projects.

The final data point is equally important, but not entered by the writers in the spreadsheet:

• Job grade (for example, entry level, mid-level, senior). Expectations are (and should be) different for each of these levels.

Calculating Work Units

For each deliverable, we multiply these inputs (see Table 1):

(# of topics or pages) × (complexity of deliverable) × (% of new or changed content)

Writer	Number of topics/pages	Document complexity	% change	Work units
Alice	128	x 3	x .10	= 38.4
Don	62	x 1	x .30	= 18.6
Kylie	200	x 2	x 1.00	= 400
Kruz	21	x 2	x .40	= 16.8
Misty	220	x 2	x .60	= 264
Lola	79	x 1	x .20	= 15.8

Table 1. Calculating work units for each deliverable

We call the resulting number a work unit. We total each writer’s work units so that all individual deliverables are included. Now each writer has a number that reflects his or her total workload from all deliverables (see Table 2).

Writer	Total work units
Alice	3181
Don	1477
Kylie	1476
Kruz	1395
Misty	2231
Lola	2520

 

Table 2. Sum of work units for each writer

Normalizing the Data

The next step is to calibrate the team’s average productivity in terms of total work units. You can derive this number using several methods, such as adding the team’s total work units and dividing by the number of writers; however, we prefer a more subjective approach that takes into consideration the productivity level we want our writers to achieve as a team. For example, if we have 12 writers, we identify three or four writers who consistently meet the average level of productivity we expect from the team, and then average their work units.

Yes, this is subjective, but in this way we can adjust for current working conditions, such as an atypical sprint to meet a tight deadline or a lull in company activity.

Determining the Productivity Factor

We take the total work units for those three or four writers and determine what number we need to divide by to make their numbers close to 100; in other words, the expected productivity is 100%. If Writer X has a total workload number of 1,400, dividing by 14 gets us to 100 (1400 / 14 = 100). Thus 14 becomes the productivity factor by which we divide all writers’ total work units.

Applying the Productivity Factor

The next step is to divide each writer’s total work units by the productivity factor you have established:

(writer’s total work units) / (productivity factor)

The resulting number is each writer’s current initial workload (see Table 3). A competent, mid-level writer’s workload number should be around 100%. If it is not, you should reassess your calibration numbers.

Writer	Sum of work units	Productivity factor	Initial Workload
Alice	3181	/ 14	= 227
Don	1477	/ 14	= 106
Kylie	1476	/ 14	= 105
Kruz	1395	/ 14	= 100
Misty	2231	/ 14	= 159
Lola	2520	/ 14	= 180

Table 3. Each writer’s normalized initial workload

Accounting for Special Projects

Next, we add the special projects percentage to the writer’s initial workload percentage (see Table 4):

(initial workload %) + (special projects %)

Writer	Initial workload	Special project %	Adjusted workload
Alice	227	+ 0	= 227
Don	106	+ 30	= 136
Kylie	105	+ 10	= 115
Kruz	100	+ 40	= 140
Misty	159	+ 0	= 159
Lola	180	+ 0	= 180

Table 4. Workloads adjusted for special projects

Normalizing the Data Again

At this point, we usually normalize the numbers again to bring the average back to near 100. In this case, we multiply all numbers by .8 (see Table 5).

Writer	Adjusted workload	Productivity factor	Total workload
Alice	= 227	x .8	= 182
Don	= 136	x .8	= 108
Kylie	= 115	x .8	= 92
Kruz	= 140	x .8	= 112
Misty	= 159	x .8	= 127
Lola	= 180	x .8	= 144

Table 5. Workloads normalized again

Accounting for Job Grade

Job grade is the final metric we factor in. We assign a multiplier value to each job grade to quantify the assumption that senior writers are expected to be more productive and maintain a heavier workload than junior writers. For a junior writer, we set the multiplier at 1.0; the mid-level writer multiplier is 0.9, and the senior writer multiplier is 0.8. The final calculation is (see Table 6):

(total workload) × (job grade multiplier)

Writer	Total workload	Job grade	Job grade multiplier	Adjusted workload
Alice	182	Mid-level	x .9	= 164
Don	108	Junior	x 1	= 108
Kylie	92	Senior	x .8	= 74
Kruz	112	Junior	x 1	= 112
Misty	127	Senior	x .8	= 102
Lola	144	Mid-level	x .9	= 130

Table 6. Final workload, adjusted for job grade

Graphing the Results

We plot the resulting value for each writer on a bar chart (see Table 7). We find an acceptable range to be within 90–110%.

Balancing Workload

What do you do about writers who are significantly above or below 100%? There are three factors that can be adjusted to change the percentage:

• Shift deliverables from an overloaded writer to an underloaded one.
• Increase or decrease a writer’s participation in special projects.
• Promote an overloaded junior or mid-level writer.

The spreadsheet is very useful as a simulation tool in this situation. See what happens if you move a project to a different writer, how much the number changes if you decrease someone’s participation in a special project, or how much the number changes if a junior writer is promoted.

How Do Writers React?

We have experienced a range of reactions from writers when presented with their metrics. (We only show a writer his or her percentage as compared to 100% and to a team average, not as compared to other writers.) Some writers are doing well and are pleased or impressed to see their suspicions confirmed! (“I knew I was doing the work of one and a half people!”) Writers with low numbers have occasionally expressed appreciation for the tangible nature of the metrics. Writers who disagree with the numbers or are dissatisfied with the process nevertheless find the quantifiable nature of the productivity metrics difficult to argue with. In several cases, consistently poor performance numbers have prompted writers to leave the company on their own, sparing us the time, expense, and legal issues associated with terminating an underperforming employee.

Caveat Emptor

We have used variations of the above metrics and calculations for the past several years to accurately and consistently estimate (a) past performance of a writer relative to his or her peers, (b) current workload of each writer relative to the team, and (c) future team capacity. However, we cannot overemphasize that what we have described is ultimately a subjective process. It must be tailored by each documentation manager to suit the needs and conditions of the specific team. Perhaps you have other factors to consider: Should different work products be evaluated using different criteria? Should writer location be a factor? Are there other metrics to consider, such as compliance requirements?

We have proven that this system works for our team:

• We never miss deadlines.
• Our productivity rate is very high compared to similar documentation teams working on similar products with similar deliverables. This claim is based on the fact that our team has expanded to encompass three existing products and associated writers from within our company. By using this system to periodically evaluate the added writers, we have reasonably objective evidence of each writer’s performance levels and their workloads over time.
• We have used the resulting metrics to successfully identify and cull low performers, set reasonable workload expectations for all writers, and identify and promote top performers.
• We usually have adequate staffing because we can provide reasonably objective metrics to management when requesting staffing adjustments.

By objectively measuring what we can and consistently comparing what can be compared but not easily measured, we have made this system to measure productivity work for us. We hope you can make it work for you, too.

Please feel free to contact us if you would like an example of the measurement spreadsheet that we have calibrated for our team.$$

PAM SWANWICK (Pam.Swanwick@McKesson.com) has worked as a technical writer for more than 20 years, primarily in technology industries. For the past dozen or so years, she has focused on medical software at McKesson. She managed one of McKesson’s product documentation teams for five years.

JULIET WELLS LECKENBY (Juliet.Leckenby@McKesson .com) has worked as a technical writer for almost 20 years, the last five at McKesson. She served as a team lead under Pam and is now the manager of the documentation team.