When it comes to keyword optimization, title optimization or advertising optimization during operation, it is difficult to accurately judge whether an optimization behavior (idea) is reliable due to the randomness of the optimization process and products and the ambiguity of the optimization logic. It is easy for us to fall into a misunderstanding: today I optimized x x product, and tomorrow this product will be ordered and sales will increase, so this optimization is good and I will do this in the future.
This behavior obviously has “survivor bias” (for example, one day you use a third-party software to enter the keywords recommended by it into the keyword column of a product, and the product suddenly sold out, but this does not mean that 100% of the keywords recommended by the third-party software are excellent), so when we perform operational optimization, we must “standardize” the optimization process and optimize the The results are “visualized”.
Case analysis: Keyword optimization effect evaluation.
If we do not do some “standardization” of the optimization process, then assuming that the left column of the table is the product SKU and the right column is the optimization related records, the Excel table is likely to look like this.
Many optimization instructions are very subjective. For example, what has the largest traffic? Is it the largest traffic in a day or the largest traffic in a month? What has the most words in a review, is it the review statistics of a certain listing or a category? review statistics.
Therefore, we need to “standardize” the optimization process and strictly define these optimization processes, that is, “control variables”. Just like doing scientific research, we need to make our optimization work “repeatable” through strict definitions. Then “data comparability”:
A. Use the word with the largest traffic ××× → Use the word with the largest traffic in ××month~×month in the ×× category.
B uses the most popular words in GoogleTrends in ×××category×month~×month.
C. Used words that were most commonly used by ×× famous sellers in the ××× category from × month to × month.
D. Used the words that appear most frequently in the current ×× best-selling listings in the ××× category from × month to × month.
E. Used the ××× vocabulary recommended by ××× software in × month, and the reason for use is the word ×××××.
When the optimization ideas are “standardized”, the above optimization methods can be defined as A~E. This is when you need to start “hypothesis testing”. Assuming that we need to test whether C is valid, we need to visualize the optimization results, because once this step is lost, our operation worksheet is likely to be in the following state.
Although the method has been “standardized” and there are specific values such as average flow rate that can be recorded and analyzed, as the data increases, purely numerical analysis requires a huge amount of energy, so these data need to be “Visualization”.
Before doing “visualization” work, we must first clarify the logical relationship between effective optimization. Taking the above-mentioned optimization record table as an example, the logic of effective optimization is as follows: the optimization method is effective – the average traffic increases. For traffic and There are many ways to “visualize” the order time. Here I choose to use color difference to reflect it.
The darker the color, the better the effect. The colors from dark to light are dark gray to light gray. The darker the color, the greater the average flow rate. For keywords optimized by C method, the same gradient from dark to light is used. Dark gray means that five words have been optimized with C method, and light gray means one word has been optimized with C method. In the conditional formatting of Excel cells, the color scale function is provided, which can help us quickly create a visual table. Here, for the number of keywords optimized by method C, a gradient from dark gray to light gray can be used. Dark gray means that 5 words have been optimized, and light gray means that 1 word has been optimized. The average traffic after optimization is also filled in the same format, and the color The darker the effect, the better. By sorting the data in the second column of the table in descending order, we can finally get an operational work table:
Then we can judge that if dark colors correspond to dark colors and light colors correspond to light colors, it proves that the optimization method is effective . It is obvious that the above figure basically conforms to this rule, so the C optimization method is an effective optimization method.
So on the basis of a reasonable combination of colors and text, a table can carry a large amount of operational information to help operators judge the business and clarify the logic of operations.