10: Prioritizing

Few companies suffer from a lack of ideas they wish to pursue. More likely, there are an abundance of ideas and a shortage of funding to pursue them all, and the firm is faced with the dilemma of deciding which opportunities to invest in pursuing.

Unfortunately, the most common method of setting prioritization is based on criteria other than profitability: most often, the seniority or political clout of the individual who backs a project determines whether it gets funded, funding is dispensed on a "first come, first served" basis, and budgetary decisions are made once per year only.

While this makes the budgeting process more convenient, it is detrimental to the financial well-being of a firm. If a project that has the potential to generate $2M per month is rejected in favor of one that generates only $1M per month (whose sponsor had greater seniority), the decision costs the firm $12M per year in lost profit. If an idea discovered in March that could generate $1M per month cannot be funded until the following fiscal year, the firm loses $10M for waiting to get started.

The author goes further, describing some of the traits of an organization that has a dysfunctional method of prioritizing work: there is an annual budgeting process when projects are considered for the following year; any new idea must wait until the next year's budgeting season; once approved, a project is not reconsidered even if circumstances change; projects are funded and the work queue is created for a long period of time; projects are worked to completion and seldom cancelled; time/cost estimates become constraints that are very rigid ("success" is conforming to the original plan); after a project, there is no plan in place to monitor and adjust (a new project must be created to correct any problems that are discovered).

This description should be very familiar to most companies - it's how they do business.

Dynamic Prioritization

To address the problem, the author proposes the notion of "dynamic prioritization" - projects are prioritized according to their financial benefit to the firm, and budget appropriations are revisited on a shorter time interval (quarterly or monthly) to ensure money is being wisely invested.

This may be a messy process: a project may be abandoned when it is half-finished because a more profitable opportunity arises that needs the resources - but from a fiscal perspective, it makes good sense: why continue to invest in an effort that generates less benefit, just for the sake of seeing it to completion, when more profitable use can be made of the resources?

(EN: My sense is that the author is leaving the notion of "web analytics" for deeper waters - enterprise planning and resource management - and may well be over his head. Moreover, the egomania of peons merits little attention: if the clerk who orders office supplies ran the firm, he'd do it in a way that made the most logical use of pencils and staplers. Likewise, if a web analyst ran the firm, he'd do it as if the things that are important to a web analyst are the most important things to the enterprise as a whole. Hence, the reason supply clerks and web analysts aren't given much power.)

Dynamic prioritization would make all organizational activates subject to analysis: those that generate the greatest impact in terms of their outcome (whether revenue generation of cost reduction), proportionate to their cast, would be prioritized on the sole basis of their predicted financial return. One need only load them into an excel spreadsheet and sort numerically, with no consideration of any non-financial concern. While this not a novel approach - it's how businesses would operate if accountants ran the ship - it's an approach that is seldom taken.

To undertake this approach, a company would need to change its organizational structure, which locks resources into specific departmental roles, so that the people whose skills were needed could be assigned to the work where their contribution would be of the greatest advantage. Again, this breaks out of the power-structure of the org-chart business where managers "own" their employees and projects must bargain to get the resource they need.

A significant difference of dynamic prioritization is in its shortened planning cycle. As soon as a new project is devised and its outcome predicted, it could be prioritized along with work in progress and implemented immediately, provided no in-flight projects created greater value. (EN: AN interesting notion is that sunk costs would have to be ignored, which would give some advantage to an in-flight project. If the original project proposed $6M in income for a $5M investment, it would have a return of 20% - but if $2M of expenses had already been paid out, the remaining investment would be only $3M for the same $6M benefit, making the return on future expense 200%. A new project would have to be very compelling to take precedence over one that is in flight. So maybe this approach is not as chaotic as it might seem.)

Another benefit of immediate reprioritization is an accelerated release cycle. Where work is planned a year at a time, completion dates are generally arbitrary (end of the month, end of the quarter, etc.) and even if the work is done sooner, it is not released. This is especially true of IT projects, in which rigid scheduling becomes a detriment to productivity and efficiency.

It's also noted that a more analytical approach also emphasizes and plans for follow-on analysis and optimization. The typical approach to business management is "launch and forget" - attention is seldom paid to whether the outcome of a project met its predictions, and there is no resource allocation to after-launch work to improve it. Even if the need to make a change or improvement is recognized, a separate project must be created to implement it. And because it must wait for the next year's budget to be approved, work in production continues to have lackluster results for a long period of time, even if it would take only a simple correction to put it right.

Dynamic Prioritization in Action

The author provides a few examples of the way in which DP can be put to use to improve business operations in specific circumstances.

One example is lead generation. Generally, a sales department works leads in batches, going through a complete "set" of leads before moving on to the next. Naturally, this means taht they work the good leads first (those with highest closure and income potential) and then work their way through a mass of duds before even considering the next batch, even if the next batch contains more promising leads. Sales would be more productive if it always worked the most promising leads available at any moment, regardless of what order they came into the shop.

Work on an e-commerce site is also organized at random. A major initiative is planned (improve product information by adding demo videos), and if a more profitable opportunity arises in the meantime (fix a problem that's causing people to bail out of the checkout process), it is side boarded until the current work is completed. Dynamic prioritization enables serious issues to be addressed more quickly, preventing the loss of revenue that would occur while a more profitable initiative must wait in line behind a less profitable one.

Call centers can tend to handle their inbound calls in a first-come, first-served order, seeking to minimize the average amount of hold time. But this fails to consider that not all customers are created equal, and it's far more valuable to the firm to be responsive to high-value customers than low-value ones. If call queues were prioritized according to the value of the customer, overall customer service might take a hit, but the service quality to the highest-value customers would improve.

Forecasting Potential Impact

Forecasting the cost savings of an efficiency initiative is generally a straightforward calculation because there is significant historical data upon which calculations of value can be based. But when assessing potential opportunities, there is often no historical basis for deriving value, and it's more difficult to monetize their potential future impact.

Historical data ca be used to derive a fairly reliable estimate of certain elements: you can look to the past to determine the value of a lead, the value of an additional sale, the cost of servicing a call, the value of a visitor to a retail store, etc. (EN: A good idea, but it's noted that this is still historical data. If your proposal is merely to get more of the same, it may be reliable. If your proposal is to improve - to make each visitor to the store more likely to buy more product in future - it's a little less cut-and-dried, though I suppose it's easier to express numerically: e.g., if 10% of customers buy 5% more product as a result of better online information.)

He also suggests considering the change in you are attempting to effect. A change in customer behavior may have an immediate impact (we will prevent 20% of people from dropping out at this point in an ordering process). A change in attitude might require more extrapolation (we will improve customer satisfaction scores by X%, which will result in a Y% increase in sales based on statistical correlation of one to the other).

Where you are doing something that is new to your company, but which has been done elsewhere, competitive data may be of some assistance (according to a third-party source, our competitor boosted their conversion rate X% by doing something similar).

The author suggests that experience will help: as you do more forecasts, you'll get better at it. And variance is to be expected when forecasting -you'll "nail" some forecasts and completely mss others, and "that's OK" (EN: whether it's "OK" depends heavily on corporate culture. Some organizations have little tolerance and a bad forecast could be a career-ending move, so proceed with caution and a lot of disclaimers.)

It's also noted that the potential impact of a change is only one factor to consider. You must also consider the cost of the initiative, the likelihood of success, the potential impact to other channels, the payback period, and the total impact on brand experience. Some of this will be plugged into the calculations of the return (the time value of money is of particular importance - failing to discount future income is a common mistake), others can preempt possible objections (if you don't indicate that you considered cannibalization, a detractor can easily suggest it as a counterpoint, even without evidence).

Pushing Dynamic Prioritization

Getting an organization to adopt this methodology is a difficult prospect: it is a dramatic shift that takes the decision-making power away from senior executives (EN: and places it in the analytics department, naturally), which is a cultural change that those who currently hold power will resist.

A good approach would be to start small: demonstrate the value of analytics to assessing the profit potential of projects. You definitely should not attempt to sell it as a method for prioritizing all work until you have demonstrated its suitability and accuracy. This will get people used to seeing analytics, and those whose projects gain approval based on analytics can be counted on for support.

From there, use analytics as a basis for making new proposals based on findings, demonstrating that analytics can not only assess the ideas others have suggested, but be used as a method for discovering ideas that would otherwise have been overlooked. This will also gather support.

Eventually, analytics will come to the attention of the executives, who can be convinced of its accuracy and validity, such that analytics becomes the standard against which projects will be assessed. Specifically, to gain approval and funding, any new project must be backed by a projection based on analytics.

From there, it is a much smaller step to making analytics the primary criteria for selecting projects, and to negotiate shortening the budgetary cycle.