Author’s note: The following is inspired by an entry in my journal reflecting on recent events at the office. After three and a half years of professional experience, I finally feel entitled to some opinions on how things ought to be done. Previously, holding such opinions would have been premature, even arrogant, and certainly unhelpful. I am fortunate to work in an environment that provides valuable lessons for my future not-as-of-yet-existing business, both in what to emulate and what to avoid.

Introduction: Swordfish

Suppose we’re the operators of a SaaS (Software as a Service) business, “Swordfish”, deliberating on whether we ought to introduce a new feature called “Bebop” to our flagship product “Cowboy”. As operators, we look in the mirror every morning and recite our job description (thrice, lest we forget), which is: to effectively allocate Swordfish’s resources in pursuit of sustainable profit through which we create durable yet divisible long-term value, i.e. make it redeemable to the stakeholders over the established durations. Having donned our value-maximization hats, we put on whatever else we’d like and head for the office.

When considering Bebop, we know in our bones that we must distinguish between “superficial innovation” and “productive innovation”; the choice of modifier depending on the balance of an innovation’s costs against its gains. To classify Bebop, we can enumerate the costs and finally consider under what conditions its gains trump said costs.

Evaluating Costs

The introduction of any feature is riddled with cost: There’s the obvious upfront cost in building it, e.g. wages, capital, and land expenses. Similarly obvious are the regular costs of its maintenance and support, included here too is the depreciation of extant capital. A feature needs usage and so Bebop’s successful adoption may require we assume further costs in marketing and user education. We must also consider the opportunity cost in the alternative projects forgone (the counterfactual nature of this cost means the exact accounting of it is often unclear but nonetheless nonzero). Arguably the least obvious but perhaps most long-term disastrous are those intangible and uncertain costs: brand dilution, a divergence or loss in internal focus, unforeseen disruption in Swordfish’s ability to execute elsewhere, as well as the risk that Bebop alienates core engaged users (our profitable cheerleaders and champions). That these are potentially long-term disastrous shouldn’t need further elaboration.

Having read somewhere that incentives are everything, we must also evaluate the incentive structure behind Bebop’s proposal. We’re well aware of how a certain competitor, “Big Number Co.”, has been criticized by those in the know for its “promo-driven development” where personal advancement comes before product coherence, leading to silly things like: discontinuing services just a year or two after their launch, breaking user experiences without adequate alternatives, nonsensical pricing or offer bundling, needless changes to logos and icons, reckless overstaffing and its subsequent layoffs. Besides not wanting to be silly, clever operators like ourselves should ask “what can we do?” A couple of things potentially: (1) Separate compensation from promotions; that is, reward performance with sizable and frequent bonuses and rely on the established mechanism of market stock growth (for public companies) for compensation beyond the base. If you want to be less reliant on the whims of Wall Street, consider internally publishing guidance and evaluating on a quarterly basis where the business compares against transparent and sensibly informed expectations — give out further discretionary rewards accordingly. (2) Inculcate a culture where the user is the focus, the customer the focus, the product the focus, and human-machine excellence is the focus. Make “busy work” something discouraged and taboo, one shouldn’t feel productive but rather be productive and so must all things have an end-goal. (3) To achieve the former, we have to be more forthright and open about growth and planning, that means proactively communicating any ceilings while working diligently to raise these ceilings and align company-wide efforts, and doing so from the get-go, i.e. by finding more opportunities for real growth at the company level. (Emphasis on company level, which is needed because bottom-up product planning inevitably results in those aforementioned silly things, top-down directives that prioritize consistency and business/product narrative are the way to go).

Conditions that Justify the Cost

Ok, suppose we’ve gone through the above exercise for Bebop. If we stop here we’d be demoralized, freeze Cowboy in maintenance mode and do nothing else besides that needed to counteract entropy. That’d be the wrong thing to do. To fulfill our fiduciary duty and achieve real growth through productive investment and innovation, of which Bebop could be an example, we must consider the conditions that might justify it.

1. In Response to User Demands

Direct requests can drive improvements that enhance user satisfaction and eventually improve the bottom line. However, the finesse lies in distinguishing between what users say they want and what they are willing to pay for or use. The noise-to-signal ratio from soliciting users is also apt to be poor: their requests can be too vague, too expensive, or downright irrelevant. As operators, we may hold two conflicting mental models in mind: (1) the user is king, always right, and worth listening to, and (2) the user is an idiot who must be guided to an understanding of their own desires.

2. In Anticipation of Future Needs

As I imagine is often the case, users generally don’t actually know what they want and no amount of market surveys or group user studies will entirely lift that fog. Rather, we must believe that they’ll instinctively know what they want when they see it; and further, that we can meet these latent desires by observing them as they are, not as they claim. If we then have strong reason to both believe that they’ll want Bebop in the near future and that (1) will eventually apply, then Bebop may be worth the effort. The strong reasons here could be trends, regulation or macroeconomic changes, etc. With the (at times seemingly unethical) powers of marketing, we can alternatively substitute “strong reasons” with “a strong confidence in our ability to later convince them they want Bebop”, i.e. shape new behaviors and desires.

3. Ecosystem and Brand Integration

If Cowboy belongs to an ecosystem that includes HammerHead and Jet, then feature additions like Bebop might be more about fortifying or consolidating the broader ecosystem rather than immediate revenue. It’s important to emphasize that in this case we expect an increase in overall market appeal to (indirectly) positively influence profitability down the line, notably in the holistic, final accounting instead of within individual product units. Cowboy improved by Bebop might then be called “loss-leading” for HammerHead and Jet collectively.

As I’ve only subtly alluded to, accounting structure and clarity of communication matters here. Consider a setup where the “loss-leading” Cowboy product is established as its own business unit responsible for its own PNL (Profit and Loss), but still expected to function as a loss-leader in the broader ecosystem (which is to the detriment of its PNL). The misalignment of incentives means the Cowboy unit is pulled in opposite directions as the leadership and employees at the various stages of product operations move in disconcert. As inconsistencies compound, the collapse of product coherence comes suddenly and violently, spoiling user trust in the broader ecosystem with it.

4. As a Strategic Imperative or the Result of Competitor Dynamics

As Swordfish, we appreciate that staying competitive demands strategic agility. And so sometimes, whether to develop Bebop comes down to strategic imperative: we either need to develop a competency in some area since it transfers well to other projects that we’re certain will grow the bottom line, or we must safeguard against obsolescence by matching market standards, i.e. quickly achieve at minimum a parity with competitors.

We have to be careful here however, on the one hand, we don’t want to be reacting to competitors as we’re ideally accountable only to our customers and users. An obsession with competitors can mean dropping ourselves to their level and, God forbid, introducing anti-patterns which we justify by pointing to their being precedence for it. That said, there is such a thing as being left behind, and so you might have to suck it up and build, even if what you’re building purportedly does nothing good for anyone, except perhaps for a couple consumers who temporarily enjoy a larger than normal consumer surplus. (This is how the “stuff AI into everything” of the current moment feels and how Cryptocurrency felt two years ago). Consider two similarly priced products, Cowboy and the competing Syndicate, both of which identically meet user requirements and are equally pleasant to use. Syndicate however has some additional irrelevant feature, “Useless Bebop”, that Cowboy does not. Ceteris paribus, Syndicate will see more usage than Cowboy simply by virtue of the fact that Syndicate seemingly provides more. Unless of course, both products belong to a particular class of products where more is in fact bad, something we conveniently avoid and can ignore given our assumption that the additional feature “Useless Bebop” is irrelevant and useless.

A Final Thought

In the end, if we’re to be responsible operators, the decision to introduce Bebop shouldn’t be taken lightly. It requires a thorough understanding of its potential impact, a careful consideration of costs and benefits, and a deep alignment with long-term strategy and cultural values. Once again, we might feel productive, but let’s ensure we actually be productive by innovating productively.