The issue of requirements verification often involves different departments, such as the product and requirements department, or it can lead to debates within the product department. So, how to make requirements verification more effective? In this article, the author has shared, you might as well take a look.
“Your most unhappy customers are your greatest source of learning.
Your most unhappy customers are your most valuable learning resource. ”
——比尔·盖茨 Bill Gates
For many entrepreneurial teams, the time of requirement verification is the easiest time to tear apart.
Not only the product and demand departments will tear it, but also the product itself.
The human brain is beaten into the dog brain, and the fight is actually a theme:
"You listen to me, I'm right!"
First, qualitative and quantitative verification of each other, it will be easier to get the answer
↑ I will use this diagram as an outline to help the startup team improve their product capabilities.
Today's article begins with the uppermost left corner of the Kanban diagram:
Requirements Analysis Capability - > Requirements Validation
In helping the entrepreneurial team to make product series, the classification of the demand analysis ability of the entrepreneurial team's product capability kanban has talked about demand emergence and demand analysis before.
This article about requirements verification is really super simple, just one sentence:
Qualitative and quantitative can be cross-validated.
Of course, for large Internet teams, the requirements verification method is extremely complex, such as business surveys, user experience reports, A/B testing, closed beta versions, grayscale releases, and so on.
But these are too expensive for the startup team, and this method of cross-validation is very simple and crude.
So today, let's talk less about theory and just talk about some examples.
1. Qualitative verification qualitative, often in the end can only quarrel
Where does a qualitative hypothesis come from?
As mentioned earlier, this may come from discussions when internal requirements emerge, qualitative assumptions that may arise through user interviews, usability testing, etc.
In cases where there is no internal agreement, validation is required.
Here's a simple example:
When we were doing homestays, some users wanted to add the tag "can cook" to the search list.
This demand is expressed directly by many users, and it is also highly recommended by the operation staff, because they will reach these users on the front line.
↑ It's old, so let's draw a schematic diagram
Everyone knows the characteristics of homestays, and the probability of family travel is relatively large, and many people are long-term rentals.
Therefore, whether the B&B allows cooking can be seen at a glance in the list, and this demand will be extremely strong for this type of user.
However, the product department objected, arguing that this was an unlimited amplification of a small part of the demand.
It's just the so-called "the wheels that can be called are oiled first".
At this moment, the internal will fall into the classic war of "users have needs, you don't do them" and "this is a pseudo demand".
Everyone came up with qualitative assumptions and confronted each other.
In fact, experienced teams know that this is the time to test qualitative hypotheses through quantitative analysis.
After launching the function in a certain version and burying the points, it was finally found that:
- Adding this element to the list does not increase the overall conversion rate;
- However, for multi-person occupancy, there will be some micro-transformation improvement.
The final improvement is that the tag is not displayed in the list, but is hidden in the secondary filter for the user to use.
↑ Again, let's take a look at this diagram
Some people may ask whether the cost will be very high if any qualitative hypothesis passes the need for online verification.
Yes, but at the same time we can do it:
- As mentioned in the previous article, questionnaires can serve a similar purpose;
- If there is more demand, it will naturally be prioritized, and at the same time, it will gradually optimize the run-in with the operation department, and everyone will slowly feel and know which ones are effective needs.
2. Quantitative verification to quantitative, in the end, you can often only be confused
Where does a quantitative hypothesis come from?
As mentioned above, quantitative hypotheses may have been obtained through data analysis and questionnaires.
If the former "quantitative analysis to test qualitative hypotheses" would be easier to understand, then here it is a bit of a detour.
Again, it's easier to take an example.
When I was working on a product in the hotel division of a vertical search engine, I once encountered the issue of wanting to analyze the importance of each element of the hotel listing page.
The goal is to make the elements in the entire list a little simpler and cleaner.
↑ Let's draw another diagram~
The pure data method is used, remove the elements one by one, gradually launch the version, and then observe the change in the conversion rate.
The result is a very strange situation.
- It is to remove the elements of the business district where the hotel is located, and find that the conversion rate of users has not changed significantly;
- Then the element of hotel star rating was removed, and it was found that the conversion rate of users did not change significantly;
- Even when I finally got the pictures of the hotel, I found that the conversion rate of users still did not change significantly.
- ……
In the end, the only thing that can't be taken away is the name of the hotel and the price.
If you follow the results of the data at this time, then you will get a violent theory:
The search results page for the hotel only needs to have the name and price......
Of course, this is not acceptable.
As we've seen, no hotel search platform dares to do that.
Of course, there is a reason behind it, and that is NP≠P.
Note that this was originally a mathematical concept, but it was generalized to the concept of product application, and I will talk about it in a separate article when I have the opportunity.
...... That is, the combination of multiple elements is not equal to the effect of the sum of individual elements, and in the process of combining elements, new value will emerge.
In this way, it is not possible to use mechanical quantitative methods to verify it, and then it will end up being confused.
At this point, it will be much clearer with qualitative verification.
Again, in this example, if we use some quantitative method:
The removal of an element (e.g., a business district) has no effect on the conversion rate and can be removed according to the Aum razor principle.
User interviews or usability tests can then be organized to further validate this quantitative hypothesis.
- If qualitative methods can also be proved, then they can be boldly removed.
- Conversely, if a user thinks that an element (e.g., a hotel image) is very helpful to them, they need to be very cautious even if they don't think it has an effect on the quantitative data for the time being.
This will achieve the mission goal of optimizing the hotel listing.
2. Qualitative and quantitative cycle rolling verification until the future
It seems that the two mentioned above are examples of lodging industry lists.
In fact, this is also an example of my own familiar experience.
For the entire lodging industry (and even the entire e-commerce industry), the search list page is the key page, and the user's time is also accounted for the lion's time.
So, what does qualitative and quantitative cyclic verification mean?
For example, the search list pages of major platforms in the accommodation industry are completely different now and ten years ago, and they will definitely be completely different from ten years from now.
But this difference is not a reborn difference.
It's like "red dresses are popular on the street", and after a while there will be a recurrence.
This is because the competitive environment, user consumption habits, user information retrieval capabilities, market environment, and so on are always changing dynamically.
Every accommodation platform has to be relevant to the moment.
For example, consider the property search listing page, where you have a dedicated product manager:
- Cutting into a question, such as whether it can be labeled as "cookable", qualitative hypotheses can be validated through quantitative analysis.
- While quantitatively analyzing this problem, quantitative hypotheses will be found in the data, such as users who seem to prefer "self-check-in" accommodations, and qualitative analysis methods such as user interviews can be arranged to test this quantitative hypothesis.
- In the next qualitative analysis, new qualitative hypotheses will certainly be put forward through user interviews.
- .……
This is called circular scrolling validation.
For this core front-end page, this product manager can continue to do it, constantly explore user needs, and always maintain a competitive advantage with his peers.
This is how the work content opens.
This is even more important for entrepreneurial teams.
This article was originally published by @觅云人 on Everyone is a Product Manager. Reproduction without the permission of the author is prohibited.
The title image is from Unsplash and is licensed under CC0.
The views in this article only represent the author's own, everyone is a product manager, and the platform only provides information storage space services.