9 the main errors made for the first year of our startup

To be clear, what was going on, I must say that our service is the exchange of copywriting ContentMonster.ru. At the moment, it runs more than 300 orders a day and is constantly working more than 400 authors. We are working on its creation is already more than six years, and during that time has filled a number of "bumps", most of which could have been avoided. Most likely, if the mistakes were not, we would grow faster, users would be greater and more stable psyche. So, let's begin.

1. Run update on Friday.

To ensure that the project is pleased with their constant growth, need over him to work tirelessly. Users waiting for new features, please do the necessary for them features, implying that a little more and they will be offended and leave to the competitors. We continuously recorded the user's wishes in a separate list, doing the most urgent at the moment, the tasks and fill the updates on a production server.

It was found experimentally that the best time to upload updates on Tuesday morning. Tuesday's system is the most compared to other days of the week the number of people and the likelihood that error will appear the maximum. Wednesday and Thursday are a little worse but also good for users is almost the same, but the probability to catch errors before the weekend less. The worst time of casting the second half Friday. In this case, "enjoyable weekend" is almost guaranteed.

We have developed a strict rule: if you do not have time to fill before lunch on Thursday, delaying the launch until next week.

2. In time domain not renewed.

Our system uses several different domains. One of them is a system of check of uniqueness of texts (all the texts written by the authors, are checked for uniqueness before sending to the customer).

One evening, because the domain is not renewed before, the uniqueness check stopped, followed by the whole system — not just the texts were sent to customers. This error is worthy of the Nuba, but it took place and this fact will not do.

The conclusion is obvious: carefully monitor the payment status of domains, hosting and other online services using a special software in the form of calendars or planners. To renew the domains at the first opportunity, and not waiting till the last moment.

3. Test new function "live" on a production system.

Once, at the time of testing of the new system API exchange, a significant portion of the working database has been deleted. All ended well (hourly backup worked), but left an unpleasant aftertaste a lifetime – a very unpleasant feeling when you press a button and deleted a huge chunk of the base. This situation has arisen due to the fact that negotation functions have been filled on a production server.
In addition, all happened in the morning, when our programmer, of course, was not in the workplace... without the programmer to restore all failed, but cost a lot of time and nerves.

Simple and very important rule imprinted life: test, test and again test it before pouring! And make important technical changes to the moment when all the team members in the operating mode.

4. Be careful with bulk operations.

In the beginning of the service operations for the payment of royalties to authors the payment was made twice. The result had personal messages to ask the sponsors to return the money. Feels like – not very nice. All copywriters, but one returned the money.

The result is a rule: to always do spot verification of the data before pressing the "Pay"button.

5. Mass spam.

The real epic fail happened with email newsletter about a month ago: the first 500 customers received 500 (five hundred) emails with news system. As was to be expected, as the Mat to your address we are not received for all time of existence of the service. It turned out the script worked in the system debugging, without doing a test mailing, we did not notice.
Conclusion: prior to any mailing, make sure to do a test for 10 email, including several of his drawers.

6. Many projects in the works simultaneously.

Probably, every Creator of his own service once comes to mind "ingenious" idea of a new super-mega-project, which is "sure shot" and therefore, "it urgently needs to do while the same didn't come." And so we rush to register a domain, design the system and create new offspring. As a result, in the best case, tomorrow we Wake up thinking about that yesterday and we sort of nonsense came up and have on hand we do not need the domain in the worst — spend n man-days to create a system of scoring the main project.

Now I will tell a banal thing: quality to develop several major projects impossible. Dispersion of forces will inevitably lead to slowdown of all projects. Now we have some unwanted domains and two major projects that we are trying to develop at the same time.

We try to fight this impulse and, at least, not to start new projects in the first few days after emergence of the idea. Ideas for new services are recorded in a separate list – may be defended.

7. The development of sophisticated features and options that really are not needed.

Each of us in the head with a lot of illusions – delusions about reality. Part of these misconceptions relates to the product that we make. It is an illusion to make us to develop and implement functions that ultimately are not in demand. It is very frustrating and leads to loss of time and missed opportunities. We have repeatedly introduced features that are not used or used very little. For example, direct transfers in the system.

So you need to create a new functionality to poll users whether they do innovation. Very well proven the model to do only what she asks from many people, and not to pay attention to individual requests (especially if their implementation is associated with considerable effort). If you want to introduce something "revolutionary" — what users are still not aware, it is advisable to consult with experts.

8. Delaying the release.

We could save a few months if you would have launched a beta version of free version of exchange, without bringing to perfection the unnecessary features.

Conclusion: to show to the public (limited number of persons) new development as early as possible after the start of encoding. Let it be still wet, but we learn about the right vector of development already in the beginning.

9. Purchase of advertising does not include the price of the conversion.

Some ads were quite effective, but most invest in it the money was wasted. Here are the errors in this area, which resulted in a significant loss of funds if:

— lost money on contextual advertising, which gave no conversion (at first, we just did not bother to configure the accounting system conversion).
— buy blog posts with a focus on the number of subscribers feedburner (so can cheat).
— purchase of posts from bloggers who did not use our service (such posts were not interesting and advertising, a completely different effect is given positions of real users)
— sticky topic on the forum Searchengines.ru gave very little traffic and also didn't pay off.

In the end, a few months ago, we decided to abandon the purchase of advertising and it had no effect on growth, which occurs mainly due to viral factors.

Many mistakes made, there will be new, because the work on the project continues. Just have to learn at least from their mistakes (better to learn the experience of other people) and to prevent their recurrence in the future and try to predict new. Don't make mistakes only one who does nothing.
UPDATE: And what errors were made while working on your startup? Invited to share experiences in the comments!

Article based on information from habrahabr.ru