Develop the right product

Recently I was asked to research early start-up companies developing innovative products. I asked them whether they think their products are getting better and better, and the answers are both positive.

Then I would ask why, they basically answered, adding product features, improving quality, and implementing the product design roadmap. Every month the number of users, revenues and usage are rising every month. Judged in turn, their products are on track.

I further asked: If the product development team leave a month, how will the company grow? Sales team will continue to sign new users, Word of mouth will make the website traffic further increased. If the product development team working overtime and leave the same result, which can make the product better and more efficient?

Many product teams do not know whether they make their products better or worse, so many consumers have liver fibrillation before they need to upgrade or update. However, due to the continuous strengthening of network effects, companies are gaining prominence in other channels or their industries are rapidly growing . Even if the products are getting worse, these companies can still achieve rapid growth .

I have set up a start-up company. Over time, we value the conversion rate from new subscribers to paid subscribers. Although the number of new subscribers has risen day by day, the analysis of discontinuities reveals the problem. We divided the users into different groups and grouped the daily new subscribers into one group. The daily comparisons with the previous day made us depressed and the conversions were basically the same. This situation remained unchanged for several months.

Although we are constantly "improving" products, but there is no change in consumer behavior or showing the problem, we did not make the product better and better, but to do worse and worse. The solution to this problem is to adopt stricter measures than to deceive people with beautiful numbers.

Most people think of A / B testing, this method is directly applied to product development will be more effective. In my new book, I cite a lot of examples of companies that have transitioned to rigorous metrics, such as Grockit, an online education company who, using the A / B tests to test new features, was surprised to find that most of the new features were completely uninfluential to consumers behavior.

Because new features often complicate product use, they have to be able to provide consumers with enough benefits to count on features that do not have "neutrality", "no change" or worse.

When product changes have no real impact, we should be brave enough to recognize that this is a good opportunity for us to know ourselves and our consumers. If we think we are improving our products, and consumers do not care, then we need to do some new experiments to find out where the problem lies.

The product itself is an experiment

Consumers do not understand their needs. There are many psychology studies that show that people can not accurately predict their future behavior, so asking consumers "would you buy if the product had these features?", "How do you feel if we change the product?" Is a waste of time Because they do not know at all.

Experiment does not mean that the product to market, look directly at the market feedback. Because the market is bound to have feedback, even if the market response is bad, but also from a variety of analysis of the curve to find a right-upward slash. As mentioned earlier, these vanity data may have been good, but you have been ruining their products.

Science needs predictions and assumptions about the results, which in turn compare the predictions to the actual results of the experiment, which is why vision is crucial to startups. We need to predict what happens when consumers encounter a product.

We can even use it to make quantitative predictions. How can we determine if the product can spread quickly? Then at the product / market fit stage (specifically, how many products you produce, how many products the consumer buys, how many servers you add and users increase How much of such a stage), requires the propagation factor greater than 1. How can we determine if our products will attract customers over the long term? Then customer retention and consumer engagement in the product / market fit phase must be very high. Similarly, how to determine if paid advertising will lead to growth, then the cost of acquiring a new user during the product / market fit phase must be lower than the profit margins the user brings. These assumptions must be tested early in product design, and if the product improvement can not be met, it is a waste of effort.

Science and vision sometimes run counter to each other, but they need to support each other, and I've heard some startups say: "If entrepreneurship is a science then everyone can do it." What I would like to point out is Even few people, even science, can do it, let alone do it. Science needs vision, just as startups need vision, and developing the right product requires systematic, brutal testing to determine the pros and cons of various elements.