The Retention Equation

Imagine you’re a hungry young company trying to squeeze into a crowded marketplace. How do you unseat the entrenched competition? You’ll probably start by leveraging the skeleton staff of entrepreneurial generalists that you’re paying with stock options and pizza. You’ll ask them to work late hours and take on the sorts of projects which your competitors wouldn’t touch with a ten-foot pole: clients they’ve fired, projects they’ve shelved as too ambitious, campaigns that have bounced around between agencies because no one’s willing to tell the client that the idea’s idiotic.

(I’m looking at you, New York-based wireless startup named after a sitcom catchphrase.)

As your company matures, you’ll start letting go of those bottom-feeder projects, but every once in awhile something will come in that’s just too tempting to pass up. It’s a potential trainwreck, but you want to believe it’s a foot in the door for additional, pricier work in the future. From a superficial cost benefit perspective, it’s a no-brainer: the margins might be anemic, but the risk of a lean month pales next to the opportunity for future growth of the account. So you do it. But what you’ve forgotten to factor into your calculation is depreciated retention.

Bad projects cost you people. The more bad projects you take on, the more time your staff will spend on Monster looking for the next job. Quantifying this cost is difficult: revenue and salary varies widely across industries, and the amount of abuse one might be inclined to accept in a young industry is clearly different than the expectations of those in a more mature sector of the economy. With a bit of generalization, though, we can create a set of formulas that help us quantify this depreciation within a rough order of magnitude, and then run some sample numbers through them to see what we get.

Let’s start by defining a few variables:

L : This is the approximate length of the project, extrapolated, for simplicity’s sake, into years (e.g. a six-month project is 50%)
eS : This is the average annual employee salary of all those tasked with the problematic project
tS : Team size, or number of staff engaged during the life of the project
rS : This is the combined annual salaries of recruitment resources, including departmental management and HR

Now let’s ballpark some constants — any of these percentages can be tweaked depending on the specific industry or individual project.

M : The profit markup on an average employee’s salary, expressed as a percentage : 100%
pD : Productivity drop associated with working on a project everyone hates; we can ballpark this conservatively and assume we’re mitigating this risk by closely managing the team : 25%
uS : Unworkable source materials, or the percentage of specifications or materials that can’t be easily modified in the future because they were created sloppily under duress; this number can vary wildly depending on how tight the timeline is — it’s not unusual for an entire codebase or architecture to get thrown out the door because there wasn’t time to design a proper solution set : 50%
rF : Recruiting fees; in Toronto, generally 15-20% : 15%
hC : Hiring cycle, or the amount of time to recruit and train a new employee, expressed in years : 20%
rU : Recruiting utilization, or the percentage of each day which recruitment staff will dedicate to finding a specific replacement : 25%

Based on these speculative values, we can create some algorithmic relationships between variables and constants. (Note that in the formulas below, parentheses are used both for order of operations and simply to provide clarity through conceptual segmentation.)

Lost productivity = eS x tS x L x pD
In other words, the cost of lost productivity is equal to a percentage of the prorated salary of all those engaged on the project.
Equity loss = eS x tS x L x uS
Lost intellectual equity (e.g. bad code, sloppy architecture, poor planning documents) is another frequent byproduct of a problematic project. We can sneakily define equity loss again as a function of the prorated salaries of the project team, since the same resources would theoretically need to be engaged to properly refactor the project.
Recruiting costs = [(eS x rF) + (rS x rU x hC) + (eS x hC x M)] x 33%
Recruiting a new employee often involves three costs: the cost of external recruitment services (average salary times headhunter fee), the cost of internal recruitment services (total internal recruitment salary multiplied by a reasonable amount of time spent every day for the duration of the hiring cycle), and the cost of lost business as a result of being shorthanded (estimated as the markup on that employee, were s/he available). Finally, we can chop the total down significantly by assuming that it takes three problematic projects before an employee leaves.

If we add up those three costs (and I’m sure there are more — feel free to add your own, or trade with your friends), we wind up with a simple high-level formula that looks like this:

Retention cost = lost productivity + equity loss + recruiting costs

Or, to write it out in full:

Retention cost = {eS x tS x L x pD} + {eS x tS x L x uS} + {[(eS x rF) + (rS x rU x hC) + (eS x hC x M)] x 33%}

Now, if I’d started off there, you’d have thought I was crazy, right?

Let’s run some numbers through the algorithm and see what happens. Assume a bad project comes in the door with a fixed budget of 50k and a lifecycle of three months. It requires a team of five who are each paid an average salary of 50k. Let’s also assume our company involves only two people in the recruiting process: the project manager and the HR manager (each of whom are paid 65k). If we replace the variables and constants with actual data, we get:

{50,000 x 5 x 0.25 x 0.25} + {50,000 x 5 x 0.25 x 0.5} + {[(50,000 x 0.15) + (130,000 x 0.25 x 0.2) + (50,000 x 0.2 x 1)] x 0.33}



In other words, the retention cost of the project actually exceeds the overall project budget. When you also add in the cost of producing the project (which we can conservatively estimate at $25,000, assuming normal margins), we arrive at a total cost of $79,795 to produce a project budgeted at $50,000.

I’ve built a Google Docs spreadsheet to make it easier to experiment with these formulas; simply create your own copy from the template provided, but bear in mind, of course, that this is an illustrative thought experiment only — if you decide to feed this data into your accounting software, you do so at your own risk.

Comments are closed.