Tracking nearly a thousand pieces of data for 18 consecutive days, pixel-level analysis in action, how to find hot-selling products through e-commerce testing

As e-commerce competition becomes increasingly fierce, how to dig out the real hits from the vast number of products is a key task for every e-commerce practitioner. This article, based on the author's personal experience of tracking nearly a thousand pieces of data for 18 days, disassembles the entire process of e-commerce product testing in detail and presents a complete set of e-commerce product testing methodology for your reference.

This article is the exclusive data analysis and tracking content of the entire network. Since this article is long, for the sake of reading experience, I have made a lot of optimizations in language fluency, typesetting, and data presentation. I wish you a pleasant reading.

Many ordinary people and novices who have never done e-commerce may have heard that doing e-commerce requires steps such as product testing and style testing.

That is to say, through some advertising and promotion methods, we can discover potential hot-selling products.

But what is the specific process of e-commerce product testing? What are the key points? What kind of data will potential hit products present during the product testing process? What are the key actions in the process?

In fact, many e-commerce novices are not clear about the details.

Today, I will teach a novice e-commerce company how to run an e-commerce store and help him build an advertising test plan as an example. I will record and analyze the daily data changes in detail and show you how to do it in practice:

When doing e-commerce, how can you use advertising to test products and discover small popular products that sell a few hundred units a month?

1. The importance of e-commerce advertising testing

When doing e-commerce, we all know that product selection is very important. I have written many articles about product selection before.

So how do we verify whether the product selection is correct and how do we judge whether we have really selected a good product? We need to use the e-commerce advertising testing method.

Here I want to correct a misunderstanding:

Many friends who have never done e-commerce think that if they choose a good product, it can be sold immediately and sales will be achieved once it is put on the shelf.

This idea is actually very naive.

Even if a product is very good, it may not necessarily sell well after it is put on the shelves because a lot of resources have not been invested in it.

For example, after the product is put on the shelves, we need to invest conversion resources to convert this product.

We need to create a detail page with clear selling points, better quality videos, more beautiful buyer shows, and better selling points.

We also need to invest in advertising resources. We need to attract traffic for this good product and accumulate initial popularity and sales. In this era of massive product competition, if you don’t accumulate initial traffic and sales for your product and don’t do a good job of conversion, then even if the product is good, it will be difficult to sell.

Then the problem comes, these conversion resources, the investment of advertising resources, all need to consume your money and time. For example, we have 10 or 20 products on the shelves, so many products, we can't invest resources to build each one.

Because you don't have that much time and money, your resources are limited, and there's no need.

It’s like you are a teacher in a class of 50 students, but your time and energy are very limited. You will definitely give priority to the top 10 best students and focus on their training.

The same is true for e-commerce. No matter how many products we launch, we don’t need to invest resources in building each and every product. Instead, we need to find the top few with the best data performance and invest in them.

So how do we find them? At this time, we need to use e-commerce ads to test the products.

2. Essential points before testing e-commerce ads

Using e-commerce advertising to test products is not something that ordinary people can do.

It does require a little bit of technicality and science.

I think ordinary people must at least meet the following points before they have the ability to test e-commerce advertising.

If you don’t have these conditions, you won’t be able to measure it.

Point 1: You need to understand some basic principles and operations of e-commerce advertising

The essence of e-commerce advertising testing is to discover some good products by placing e-commerce advertisements and looking at the data on product performance under the advertisements.

Therefore, you still need to understand some basic principles of e-commerce advertising, such as the logical relationship between display, clicks, add-to-cart, click price, add-to-cart rate and other data. I have written about this in previous articles (see "18 Key Points for In-site Advertising for Ordinary People Starting an E-commerce Business"), but I will not elaborate on this today due to space constraints.

At the same time, you also need to understand how to build an e-commerce advertising testing plan, what the specific process is, and understand some basic operations.

I will also publish a separate article on this part later.

However, the Internet is very developed now, and you can also learn about this content by watching online courses and short videos.

Point 2: You need to make mass-market products and mass-price products

As of now, when ordinary people start to do e-commerce, it is best to focus on mass categories: such as daily necessities and daily necessities, which are close to consumers' daily life needs.

And we need to make products in the low-price range and the mass-price range, such as products costing a few dozen yuan, a dozen yuan, or no more than 100 yuan.

Because you are making this type of product, it is easy to get results when testing the product, and the data direction will be very obvious.

Because it is a popular category and a popular price range, consumer demand is large and shopping intention is obvious. If the product is good, consumers will easily take further actions when they see your ad, such as clicking, adding to cart, and placing an order, so it is easy to obtain data.

If there is no data, it can be reversely proved that the problem is with your product or vision. After all, you are selling mass products in the mass price range. If the product data is not good or there is no data, then it is definitely a product problem.

Point 3: Product visuals should reach the upper-middle level in the industry (key point)

I have repeatedly emphasized the importance of product vision in previous articles.

I have to emphasize this again here.

As long as you are doing shelf e-commerce, whether it is the domestic Taobao, Tmall, Pinduoduo, Xiaohongshu, or the cross-border Amazon, AliExpress, and independent station Shopee, as long as you are on these platforms, no matter what products you choose, consumers will always judge whether your products are good or bad through vision and pictures.

Because consumers don’t know what your product looks like, the only criteria they use to judge your product are:

It’s the picture of your product.

If your pictures are good, consumers will naturally think that your product is good. If your pictures are poor, consumers will think that your product is poor.

So as long as you do e-commerce, you must always remember: vision is the product, pictures are the product. Selling products is selling pictures.

The same goes for product testing. We test products to find good products, so we must avoid "low-quality pictures" that cause consumers to make wrong judgments about the products.

What's the meaning?

For example, if the product you sell is very good, but the pictures you take of the product are too rubbish or too ugly, then consumers will think that the product is too bad and will not click to add your product to the cart, so the data will be very bad.

When you see the poor data, you will think that the product you chose is rubbish. In fact, it is not the case. It is because your pictures are rubbish that bury this good product.

Therefore, the quality of pictures is the most important part of e-commerce product testing, and even the most important part of the entire e-commerce business process.

So I think that if the quality of your product images is not high and cannot reach the upper-middle level in the industry, then you should not test the product, because you will not get any valuable data even if you test it, and it is meaningless.

If your pictures are of very high quality and taken particularly well, then the testing data will be better and it will be easier to find hot-selling products.

They all complement each other.

Point 4: The ultimate uniformity of the test products (key point)

The products you test must ensure ultimate uniformity.

This is very important.

Because when we test models, in a sense we are doing scientific quantitative observation and analysis.

Since it is an observation and analysis, we must ensure that every product tested is as similar as possible.

For example, if we sell mugs and want to test them, we must first ensure that these mugs are in the same price range, such as selling for more than 10 yuan or more than 20 yuan. You don’t want some products to be very cheap and some products to be very expensive. This would be inconsistent.

Secondly, we must try to ensure that the specifications and materials of the mugs are relatively similar. For example, they are basically all made of ceramic and the size of a coffee cup. No product will be special, made of plastic, or be an extra-large cup.

In short, everyone's specifications, sizes, and materials are basically the same.

Secondly, we should try our best to ensure the uniformity of product styles. For example, these mugs are all cute or simple in style.

Then, we need to ensure visual unity as much as possible, and use a unified visual solution to shoot all products.

You ensure the ultimate uniformity of the product, and the data obtained when testing the models is rigorous for comparison and reference.

Because they are all the same products, with the same price range, the same style, and the same visuals, some products perform well in the test and some perform poorly. So the good products must be good and are not affected by other uncertain factors.

If you test mugs, water cups, coffee cups, plastic cups, and other products together, some of which are expensive, some are cheap, some are large, and some are small, and the data performance of each product is different at the beginning of the test, you will be completely confused, not knowing whether the inconsistent data is caused by the category, price, or style. You will have no idea where to start the analysis.

And there is a more important point. When we test products, we use them to hit the search terms of platform consumers. The more concentrated and unified your products are, the more focused you will be when you count the search terms, making testing more efficient and labor-saving.

Therefore, when we select products, in order to facilitate subsequent testing, we must also focus on one specific direction to select products.

This is also what I keep emphasizing, your product selection should create favorable conditions for subsequent promotion.

3. Preparation before testing

Having said the key points, we begin preparations for e-commerce advertising testing.

First, we need to prepare the products to be tested.

I know this is bullshit.

The question is how many products should we prepare? 2? 5? 20?

If an ordinary person is just starting out in e-commerce, I generally recommend testing products in groups of 10.

That is to say, every time you launch a product, you should do it in groups of 10.

After the products are put on the shelves, they are tested in groups of 10.

This will help you form good habits for data statistics and expense tracking in the future.

Moreover, in the initial stage, once 10 products are put on the shelves, the store will appear fuller, not too empty, and appear more formal.

Secondly, before testing, you must first perform primary conversion actions on these 10 products.

What is a primary conversion action?

That is, the product must have basic sales (2), reviews and buyer shows.

Why do we need to take primary conversion actions?

The principle is the same as the previous product visual, which is to avoid the product being too unpopular and causing shopping troubles for consumers.

Because on e-commerce platforms, consumers' shopping habits tend to follow the crowd, and no consumer wants to be the first to try something new.

You put 10 products on the shelves, but none of these 10 products have any sales and no reviews. Even if consumers think the products are good, they dare not take further action because they will feel insecure.

Therefore, in order to avoid the impact of test data caused by this situation, we need to do some conversion actions in advance.

As you can see, our e-commerce advertising testing is really like a scientific experiment. Since it is a scientific experiment, we have to use many methods to avoid some unfavorable factors that affect consumers' actions and cause us to misjudge the product.

It’s about visuals and it’s also about conversions.

Of course, you don’t need to do too much of this conversion in the early stages. Too much will be too troublesome and tiring.

For ordinary people, what we need to do is, before testing, find friends and family, give 10 products to test, make 2 sales for each product, make a review, and make a buyer show, so that the product has a basic conversion ability, so that the test can be more accurate.

Of course, you may say that only 2 sales, 1 review, and 1 buyer show is too little?

Shouldn't we do more?

You are right, but the problem is that you don’t have the resources, the ability, the energy, the money, or the time.

Generally, large e-commerce brands need to get dozens of sales before testing their products. This includes guiding old customers to buy, offering VIP customers half-price purchases, etc. After accumulating some sales and reviews for the product, they can then enter the testing phase.

But for ordinary people like us, with limited time and energy, 2 sales, 1 review, and 1 buyer show for each product are enough.

The last step is to prepare a vocabulary library for the product being tested.

The essence of e-commerce advertising testing is that we put advertisements for 10 products to match the search needs of consumers on the platform.

For example, if I have 10 mug products, and I want to test whether consumers like these mugs, I need to let these 10 mugs meet consumers' search needs.

So what reflects consumers’ search needs?

It is presented by many [search keywords].

Because consumers want to buy mugs, they will search for them, and they need to enter search keywords, such as:

[Mug][Ceramic Mug][Ceramic Mug][Mug with Lid]…..

Testing e-commerce ads means placing these 10 mugs on these search terms to see how these products perform in terms of search terms.

Then we need to sort out the keywords that consumers may search for when buying mugs. Since there are many words, we will compile a vocabulary library.

Take the following picture as an example.

The content of the vocabulary is quite complicated, and there will be special content explanation later, so I will not expand on it here.

Therefore, e-commerce advertising testing does have a certain threshold. It requires a certain amount of learning about product selection, advertising, and placement before you can operate it. Otherwise, various problems will arise, such as failure to test products, inaccurate testing, high testing costs, and poor results.

The three major preparations before the test have already come out:

1. There are 10 products to be tested.
2. Perform primary conversion actions for each product
3. 3. Consumer search terms based on these 10 products

After you have prepared these, you can start testing the model.

I emphasize again that if the product does not have certain conversion actions, it cannot be tested and the test will be very inaccurate.

Because bad data does not necessarily mean that your product is bad, but consumers see that the product is empty and dare not buy it even if they want to.

4. What data should be considered when testing models?

We are not done yet. Let’s continue to talk about e-commerce advertising testing. What data should be tested? What data should be used as the criterion for judging the quality of products?

Many people may say that to discover a good product, you must look at the product’s sales volume!

A product is good only if it sells well.

This is not quite the right idea.

Because of shelf e-commerce, consumers' shopping habits often lag behind. They do not place an order to buy a product immediately after seeing it. Instead, they will generally observe and compare it, and look at it repeatedly several times before adding the product to the shopping cart, and then place an order to buy it at a convenient time.

If you use sales volume as the judging criterion, you will find that it is difficult for the test products to achieve sales volume at the very beginning.

Therefore, the data we test will be judged based on the [Add to Cart] (abbreviated as Add to Cart), especially the [Directly Add to Cart] (abbreviated as Directly Add to Cart) data.

When consumers see the advertisement for this product, click in, and are willing to add the product to the shopping cart, it means that this product is liked by consumers and has gained their purchasing intention.

It is a big action for consumers to add products to the shopping cart. If they don’t really need your product, they will not do this.

In addition to adding to cart, another data we need to observe is [click].

That is, after I advertised my product, how many consumers saw my ad and clicked on my product.

When testing a product, the more clicks it gets, the more it means that the product picture is very good, or the price is very good, which makes consumers immediately attracted and willing to click in to take a look.

Please understand that no consumer has nothing to do but order products everywhere for fun.

It must be that after browsing a large number of products, he saw your product picture, was impressed, and clicked in to take a look. This also shows that your product has the ability to initially attract consumers.

Secondly, it also means that platform advertising will push your products to a more accurate audience.

Because e-commerce advertising is now in the era of algorithms, and each person has his or her own face. Your product will initially only be pushed to a very segmented group of people. The system may or may not be accurate in its push.

If our product can get more clicks, it means that the keywords and target audiences are recommended by the system more accurately. It also means that the advertisement accurately identifies our product, which is a good sign for bringing in accurate traffic in the future.

So a good product:

• It must be that the platform accurately identified the ads and pushed them to the right people;
• Secondly, consumers see the product and want to click in;
• Finally, after consumers click in and see the product, they want to add it to the cart directly.
• Of course, other data such as click-through rate, add-to-cart rate, click cost, and add-to-cart cost are also very important.

But ordinary people, as long as they grasp the two data of [clicks] and [direct add to cart] at the beginning, they can basically dig out a good product.

5. Cost and cycle of testing

Some friends will definitely ask, how much does it cost to test a model? Does it cost a lot of money?

No need, I went there myself and found that I could complete an advertising test for a set of 10 products by spending only 30 yuan a day.

The time period is about 15 days.

Of course, the prerequisite is that you must make mass-market products below the price range of 100 yuan. The more expensive the product, the more complex the product, and the more niche the product, the more cost and time will be spent on testing.

If the visuals are better, testing will be more cost-effective and more efficient.

Therefore, whether your testing is efficient and whether you can spend less money, the product's visuals, categories, price range, uniformity, and vocabulary will all have an impact.

VI. Overview of the Case Study

Having said so much theory, I will now use a “practical case study of e-commerce advertising testing” to show you how to use e-commerce advertising testing to discover a small hit product that sells over 300 units a month.

The category in this case is the [Mug] category, a very basic and popular category.

The direction is cute style mugs, with the average customer spending 15 to 20 yuan.

The test store in the case is a store registered from scratch, with 0 levels, 0 products, 0 sales, no weight, no old customers, no accumulation, and is a completely new store starting from scratch.

The shop owner is my friend, a 38-year-old “graduate of a big internet company”. He knows little about e-commerce and I basically give the instructions and he executes them.

Although the student did not understand e-commerce, he did a very good job in execution, laying a good foundation for the subsequent smooth testing of the products.

First of all, he pursued the ultimate uniformity in product selection, basically focusing on small Mark ceramic cups and cute style. The products and vision of the entire store are highly unified.

Secondly, I spent a lot of time polishing the visual aspect, setting up the scenery and lighting myself, and using an Apple phone to capture the industry's top visual quality.

It creates a good space for subsequent testing.

The following are some product pictures taken by the student:

On December 5, 2024, the store launched the first batch of 10 products, and in the following days, each of these 10 products completed 1 long review and 1 buyer show.

After the basic conversion actions were completed, I asked the students to start e-commerce advertising testing on December 13, 2025. And built an advertising testing plan including 10 products.

In order to make it easier for everyone to understand later, I will arrange the products in order from 1 to 10.

The same promotional keywords are also added to each product.

We also set a budget target of 30 yuan per day (please pay special attention to this number)

Now that the product is OK, the style is unified, the visuals are OK, the vocabulary is OK, and the cost is OK, then let’s start formal testing!

December 13, first day of testing, cost: 2.5 yuan, situation: nothing happened

Please close your eyes when it gets dark. The first day of testing has been completed. Please open your eyes.

On the first day, nothing happened. All products were displayed, but only a few products had a small number of clicks.

This is a very normal situation.

E-commerce advertising generally takes time to cold start. At the beginning, there is less data. It is a good thing to have exposure, which means that our products have begun to be displayed to target consumers by the advertising system.

All we have to do now is keep observing.

December 14, the second day of testing, cost: 20.07 yuan, situation: product No. 2 data is prominent

On the second day of testing, the data of product No. 2 suddenly soared.

The number of clicks on other products remained in the single digits, while product No. 2 received 21 clicks.

Product No. 2 costs 0.4 yuan per click and 8 yuan for 20 clicks, which consumes nearly one-third of the daily budget.

It’s a pity that product No. 2 is not added to the cart, but there’s no need to rush. At least it shows that product No. 2 has hit the e-commerce advertising algorithm and has been pushed to a more precise group of people. In addition, the picture of product No. 2 is indeed very good and very attractive to consumers. When consumers see the product, they want to click in to take a look.

Is this huge number of hits just a coincidence?

We still need to continue observing.

December 15, the third day of testing, spending: 17.48 yuan, situation: No. 2 data is still strong

On the third day, Product No. 2 was still performing strongly. While other products only got a few clicks, Product No. 2 still got 23 clicks, crushing other products.

You should know that these products are of the same style, size, price, and promotional slogans.

Product No. 2 can get so much traffic, it must be accurately identified by the platform, and the product and pictures are very good, which easily attracts consumers' clicks.

This must be a product with great potential!

At this point, some readers will say:

Didn’t you say that a product must have both clicks and add-to-cart orders to be considered a good product?
Why do you think it is a good product now?

That’s right, so we need to make product No. 2 a top-up item as soon as possible to further prove whether it can stand the test.

So how do we help No. 2 to add the item to the cart?

Continue to increase its product conversion power!

Because product No. 2 is now very capable of attracting traffic, but there is no next step after the traffic comes, which means the conversion is not good!

I urgently arranged to add a 5-yuan coupon to product No. 2 to increase a conversion point. At the same time, I asked him to add 5 more reviews + 4 buyer shows to make the conversion points sufficient.

After we have completed all the conversion points, we will see if product No. 2 can be added to cart.

However, these transformations take time, and we need to continue observing.

December 16, 4th day of testing, spending: 27.4 yuan, situation: product No. 5 had its first direct purchase

On the fourth day of testing, product No. 2 was still able to attract a large amount of traffic. I kept urging it to execute it as soon as possible. After all, product No. 2 had a strong ability to attract traffic, and it cost 8 to 9 yuan every day, which was a lot of money.

Other products also began to receive clicks one after another. Clicks cost money, and advertising expenses gradually approached the daily budget line of 30 yuan.

On this day, product No. 5 had its first direct purchase!

It’s a good sign, but one direct purchase doesn’t mean much. The test data may be accidental. We must accumulate more data, especially continuous direct purchases, to prove that it is a good product.

Therefore, product No. 5 needs further observation.

December 17, the fifth day of testing, spending: 29.54 yuan. Situation: two more purchases on the 5th, one purchase on the 6th

Product No. 5 has been purchased again, and there are 2 of them!

The fact that it can continuously generate additional purchases proves that this product also has the potential to be a hit!

I also arranged for the trainees to strengthen the transformation of Product No. 5.

Did I make a wrong judgment?

Maybe product No. 2 is not an excellent product, but product No. 5 is the excellent product?

Further observation is needed.

However, what I am more worried about is that although product No. 5 has been continuously purchased, its ability to attract traffic is too poor.

Product No. 5 can only get a few points every day, which will greatly restrict the number of additional purchases it can get later.

After all, its ability to attract traffic is too poor, and even if there are add-on purchases, they are too little.

Although product No. 2 does not have add-on cart, it has a strong ability to attract traffic. Once the conversion is done well, it will definitely bring a lot of add-on carts, and the upper limit is very high.

Product No. 6 has also been added to the shopping cart, and we will continue to observe it.

December 18, 6th day of testing, cost: 30 yuan, situation: No. 2 traffic is strong, and there are additional purchases

On the sixth day of testing, product No. 2 still had strong traffic, receiving 39 clicks, and had a strong ability to attract traffic!

The daily budget of 30 yuan has finally been spent, and it is even starting to be insufficient. A few days ago, the 30 yuan per day for advertising was not spent, but today, all the 30 yuan was spent in less than half a day.

It's like you raise 10 chickens, and the daily ration is limited. If you just scatter the food, all 10 chickens will eat it. It's not enough at all.

If you want to continue to get traffic for each product, your only way is to:

Add money.

The daily budget needs to be increased from 30 yuan to 50 yuan, 80 yuan or even 100 yuan to support the testing of 10 products.

But we don’t add money, because we are testing with a fixed budget. We only have so much money, so we will choose whichever product can be tested. We won’t add a penny more. Even if some good products are buried, it doesn’t matter. The product that can be tested with such a small budget will be the best.

In order to continue testing the products better, we have to give up some products and focus the 30 yuan on products that can produce better results.

The results for products No. 5 and No. 6 are now available. Both have been added to cart, and product No. 2 has also been added to cart today.

The key is to ensure that these three products can get enough traffic.

So we decided to abandon products No. 8, No. 9, and No. 10, which had consistently poor performance data, and shut down their promotions.

Although they may be good products, they cannot attract traffic and there is no add-on purchase. We also don’t plan to invest more money in testing, so we have to give up.

Now there are only 7 products left.

The focus is on numbers 2, 5 and 6.

All you need to remember are these three products.

Although product No. 2 had add-to-cart orders after the conversion action, I was worried that it was just a coincidence. I was afraid that product No. 2 consumed too much traffic and had too few add-to-cart orders, so I could only continue to observe.

December 19, 7th day of testing, spending: 28.96 yuan, situation: added to the shopping cart on the 2nd, and added to the shopping cart on the 6th

On the 7th day of testing, product No. 2 started to be purchased again!

This is a good sign, indicating that the conversion has taken effect and the traffic attracted by product No. 2 can be converted into add-ons.

At the same time, product No. 6 has been added to cart intermittently, so we need to continue to increase conversion actions and observe.

As you can see, when a product can be added to cart continuously, it has the initial potential to be a good product. We need to continue to supplement its conversions, strengthen its conversion ability, and continue testing.

Because a truly good product, after its conversion capacity is supplemented, the additional purchases it brings will definitely be more stable and continuous.

We must gradually invest resources to avoid burying good products.

It’s a pity that product No. 5, which used to bring in consecutive purchases, has had very poor data in the past few days, even though we have added conversion actions.

This only shows that Product No. 5 is not a potential hit product because it lacks the ability to generate sustained conversions.

December 20, 8th day of testing, cost: 30 yuan, situation: I still added it to my shopping cart on the 2nd, and also added it to my shopping cart on the 6th

Today is the 8th day of testing the product, and the situation is beginning to become clear. There are continued purchases on the 2nd and 6th every day.

The stability of add-to-cart is constantly strengthening the various data of these two products, making advertising more accurate.

Product No. 5 began to lose momentum, and other products were basically lagging behind, so we began to gradually shut down more products.

Focus the budget on the remaining products.

December 22, 10th day of testing. Cost: 30 yuan. Situation: Only 3 products were left.

By the 10th day of testing, we shut down 7 products, leaving only No. 2, No. 5, and No. 6.

As you can see, as the testing progresses, we need to focus our budget on the products that come out, so that the testing is more accurate. However, the budget is limited. If we spend more on one, we have to spend less on another.

A shut down product does not mean it is a bad product. We can retest it later when we have more money, but right now we urgently need to test the product that is most likely to become a hit as soon as possible.

December 24, 12th day of testing. Cost: 29.24 yuan. Situation: Only products No. 2 and No. 6 left.

On the 12th day of testing, product No. 5, which was the first product to be added to cart and even appeared in consecutive carts, no longer had volume.

Even though we took strong conversion actions including price cuts, coupons, buyer shows, videos, etc., we still failed to generate additional purchases for product No. 5.

This means that the good data before product No. 5 was just a coincidence.

An excellent hit product must be able to generate sustained volume, so we gave up on product No. 5.

Now only products 2 and 6 are left!

Both products can attract a lot of traffic and have been added to cart for four to five consecutive days. Which one will be the real potential hit?

We continue to observe.

December 26, 14th day of testing, cost: 25.76 yuan, situation: Product No. 2 started to increase in volume

After we continued to add some conversion actions to products No. 2 and No. 6, product No. 2 began to increase in volume. On the 14th day of the test, it won 4 add-ons. However, the volume of product No. 6 became unstable and there were no add-ons.

December 27, 15th day of testing, cost: 17.31 yuan, situation: No. 2 continued to get stable volume

After another day, the sales of product No. 2 began to stabilize. Not only did it receive direct purchase orders every day, but it also received multiple purchase orders for many consecutive days.

This means that the conversion power of product No. 2 can handle the traffic!

However, product No. 6 began to decline. Not only did the traffic decrease, but there were no additional purchases, and its momentum began to lack.

As I looked at the data, I sighed and said, how could product No. 6 fail at this time? I have given all the budget to you and No. 2. You have both added them to cart before, so why can't you keep up now? It seems that your product is not a real hit.

December 28, 16th day of testing, cost: 30 yuan, situation: No. 2 data burst, multiple orders completed

By December 28, the test entered its 16th day and finally achieved interim results.

Product No. 2 received 8 direct purchases on this day, and 3 orders were completed!

This shows that the conversion power, weight, crowd, images, and advertising system recognition of product No. 2 have all entered the precision stage, and can get more add-ons and even conversions.

When testing a product, don’t rush for conversions. As long as you can get direct add-to-cart orders steadily, and the add-to-cart trend is on an upward trend, conversions will inevitably occur.

December 30, 18th day of testing, cost: 29.46 yuan, situation: data on the 2nd day continued, and there were still transactions

On the following day, product No. 2 can still be added to cart.

On December 30, the 18th day of the test, product No. 2 received 6 add-ons and 1 transaction on the same day.

It is worth noting that Product No. 2 still has strong traffic, can grab a large amount of budget in the advertising plan, and obtain a lot of traffic. At the same time, it has good conversion power and can convert the attracted traffic into add-ons and transactions.

This shows that there is great potential for product No. 2 in the future.

Product No. 6 may also be a good product, but unfortunately it lacks stamina, so we have to give it up temporarily.

We plan to concentrate our resources on number 2 first.

At this point, the goal of [discovering potential hits] has been achieved, and this advertising test has come to an end.

VII. Summary of Testing

This is the whole process of discovering the best-selling products by quantifying 10 products and testing them for more than half a month.

First of all, there are many products that are insufficient at the start, start very slowly, or never start up. This means that the product is really not good, or the advertising system is difficult to push to the right audience. With limited budget and time, we can only give up.

Some products performed very well at the beginning. For example, products No. 5 and No. 6 performed very well at the beginning and were quickly added to cart and continued to be added to cart.

However, if the product lacks stamina and cannot continue to attract traffic and purchases, then it should be abandoned after observing for a period of time.

Some products may seem very popular at the beginning, but the add-to-cart function is not very good. For such products, don’t give up in a hurry. It is very likely that the pictures are very attractive, but after consumers click on them, they find that there are too few sales reviews, the details page is too poor, the price is a bit expensive, etc., and there is a lack of conversion drive.

So we should not give up on such products, but increase their conversion power, because they can attract traffic. If you do a good job of conversion, you will definitely get add-on purchases and sales.

For example, product No. 2 attracted a lot of traffic at the beginning, but there was no add-to-cart. After slowly improving the conversion rate, it began to show its qualities as a hit product.

Therefore, our product testing is to increase consumers’ decision-making power through multiple means and to discover as many hidden good products as possible.

But our budget is limited, and we have to constantly make trade-offs to focus resources on the most important products.

Is it over when the test is over?

Is it over once the test is completed?

No.

Testing only helps us discover products that have the potential to become hits. This product only has potential, but it has not yet become a hit.

But we can be sure that as long as we invest resources in this product, there is a high probability that it will become a small hit.

Product No. 2 meets these characteristics. It can attract traffic, a lot of add-to-cart orders, and has very good conversion rates. So next we will start to promote and optimize it in all aspects.

We built 3 to 5 low-budget advertising plans for Product No. 2, focusing our efforts on driving traffic and sales for it, and accumulating basic sales and weight. Product No. 2 lived up to expectations and saw a small sales increase before the New Year, from single-digit sales per week to more than 60 orders per week.

If the Spring Festival holiday had not interrupted the sales rhythm, the sales volume of Product No. 2 would have continued to grow.

Secondly, there are still many areas that can be optimized in Product No. 2.

For example, during the test, we found that although product No. 2 had a lot of traffic, the cost of the traffic was relatively expensive, with an average cost of nearly 0.5 yuan to obtain one traffic.

The traffic cost is relatively high and there is a lot of room for optimization.

We gradually optimized pictures, titles, and keywords, and gradually reduced the traffic cost from 0.5 yuan per piece to 0.36 yuan per piece of traffic, a reduction of nearly 30%.

But this is not enough, the traffic must be reduced to below 0.3 yuan, so it is still being optimized.

For example, although the No. 2 product can be purchased in an additional purchase, the additional purchase cost is still expensive. The direct purchase cost of a direct purchase cost of 5 to 6 yuan during the test period.

This is too expensive.

After continuous conversion, optimization of pictures, product details pages and other means, the cost of purchasing a No. 2 product is only 2.5 yuan, a 50% reduction.

In short, there is still a lot of work to be done. Not only should the No. 2 product be made into a hot product, but the No. 2 product is also a low-cost customer acquisition and a low-cost explosive product.

In addition, we launched 10 new products and started a new round of test models, repeatedly testing models, tapping out potential products, creating hot products, testing models, tapping out... In this process, those hot products will be born in batches.

8. Conclusion

Although it costs several hundred yuan to test the model for a time, it is actually very worth it, because only by discovering truly high-quality products can you be worth investing more resources.

The test model is well done, which can effectively avoid the waste of money and time, allowing us to all in truly definite products.

This article has already finished the actual content on model testing. Let’s not wait, select the product and shoot it on the shelves as soon as possible, and then start the model testing!