In fact, operating an account requires a bit of craftsmanship, the spirit of excellence and concentration. After all, when you have built an account close to perfection and classified keywords close to perfection, the effect will not be bad.

From the time I took over this new account to now, the account effect is obvious. The conversation cost has dropped by 70%, and the number of customers has increased by 56%. It did not take a very long time. It also gave me some confidence that I, who was once in the dark about account operations, can firmly believe that I can do better.

I would like to share my transformation process, hoping to provide some reference for everyone. However, I can say for sure that this experience will not apply to everyone, and everyone is welcome to discuss it.

1. Do PC and mobile have to be separated?

Originally our account was under custodial management, but as the results continued to be poor, we took the account back and operated it ourselves.

Before I got it back, I had always only advertised on PC devices, and used the device coefficient to control the bid for mobile devices. After running the data for a week, I decided to advertise on PC and mobile devices separately.

There are several obvious benefits after separation:

1. Facilitate data analysis

The data of PC and mobile can be clearly seen at a glance, and the consumption proportion of PC and mobile can be seen very intuitively. Although the data can be analyzed separately when the two devices are put together, it is not so intuitive.

2. Easy to adjust bids

If they are not separated, mobile devices can only adjust bids through coefficients, and there is no way to be specific to each keyword. If the keyword bid is adjusted, the PC device will also be adjusted, which is not convenient.

3. Easy to operate and save costs

Take us as an example (education), customers mainly search on mobile devices. Taking general use as an example, both mobile and PC consume very quickly. We cannot directly see which words are consumed quickly on PC and mobile. Even if we want to see it, we have to download the table. Once it is released on different devices, we can view it directly, which can save a lot of time.

Where are the cost savings reflected?

If the delivery is to PC devices, the mobile coefficient can be set to 0; similarly, if the delivery is to mobile devices, the PC coefficient can be set to 0. Even if the PC is the main device, the coefficient of the mobile device is set to 0.1, there will still be expenses. I have tried this setting, always thinking of catching those "slippery fish", but the result is that I not only did not catch those rains, but also spent 500 yuan more a week...

4. Smoothly enter oCPC

Facts have proved that the zero-threshold oCPC on mobile devices is still quite effective. If there was no division between PC and mobile at the time, and a low-threshold or conventional method was adopted, it would not have entered oCPC so quickly. Moreover, compared with running oCPC on the entire account and dividing it into multiple plans, the latter is more effective.

After running oCPC, the conversion cost is within the expected range, and the cost of multiple packages is lower than expected. Before this, the following methods have been tried:

Because the number of conversions is limited, a zero threshold is adopted and the mobile plan is tried first.

1) One package for one plan vs. one package for multiple plans

At the beginning, I didn’t package all the plans because I was afraid of affecting the overall account, so my operations were still relatively conservative. First, I tried a plan with relatively fast CPC consumption and average conversion (mainly to control costs). After running for a week, I successfully entered the second stage, and the conversion cost was also controlled within the expected range. This gave me confidence, so I built packages for the remaining plans separately. After a week, only 2 of them were successful, and the others failed, so I went back to CPC.

After that, I continued to use CPC to optimize the remaining failed plans. After about a week, CPC dropped by about 1 yuan, and the number of conversions became more stable than before. Then I tried to build packages for all the remaining promotion businesses separately. After a week, all of them were successful, the effects tended to be stable, and the conversion costs began to decrease significantly.

2) PC and mobile separately vs. entire account/PC+mobile multiple plans merged

In fact, I didn't want to try the entire account at the beginning, because there were several plans in the original account that had very fast consumption and poor conversion. If the entire account was combined at the beginning, it would still be impossible to optimize these high-cost plans, and it would be difficult to distinguish which plans were running well and which were lagging behind. So I insisted on running according to the plan. Now the effect is stable, and there is no need to combine them together again.

As for whether to combine multiple PC and mobile plans, I personally think it is unnecessary. We separate PC and mobile in order to better control the budget and conversion effect. There is no need to run OC and PC together. Moreover, most accounts have more mobile consumption and less PC consumption. Combining them together is not conducive to budget control, so it is not recommended to combine them. Although I have also built a low-threshold delivery package on my PC, it has not yet entered the second stage. However, the conversion cost is decreasing after optimization, and it will enter the second stage with some time. The cost may be further reduced, allowing different devices to run in a way that suits them, which is actually better.

After entering ocpc, the CPC has increased by more than 1-2 yuan, mainly because the CPC of some words has reached more than 90, or even more than 200, which is too high. However, I think the overall effect is still good, and I run it according to the plan, so as long as the conversion cost of this plan is acceptable to me, I will continue to run it. Now there is no situation where a word has such a high CPC for multiple times or for multiple days, so there is no further processing of such words. If it appears, we still need to pay attention to the data in real time and do processing.

2. The Long Tail Theory Available to Everyone

1. The long tail of creativity

Most people think that only a part of the materials and creativity are needed, and that you only need to do a part of anything that can be done. In fact, this is why many optimizer accounts do not perform well, because there is also a very important long tail theory, which is also the key to Amazon’s success. So how do I use it?

I uploaded all the creative ideas that could be uploaded, such as flash investment ideas. There were 30 creative ideas in total for PC and mobile terminals. The editing was also very complicated. I uploaded all of our three product lines, which took up a lot of my energy and time. Originally we only had 1-2 accounts, and at most 5, but now I have more than 100. My hard work has paid off, and the effect of the long-tail theory has become apparent. The effect of the account will naturally show up. We have no control over the media's backend, but what we can do is to make full use of all opportunities for display, which can help us avoid some loopholes in the media system to a certain extent.

Originally, there were 1-2 creatives with a click-through rate of less than 1% and a CPC of nearly 10 yuan, which was far higher than the industry data. Consumption was also several times higher than it is now, so conversion was naturally not as good as it is now.

After adding it, how do we get the effect of this idea? This basically meets the AB testing method. Other contents have basically not changed. This is the only variable. The time period is the same, and the span between the industry's peak and off-seasons is not large.

2. Long tail of keywords

The use of the long-tail theory of keywords is what we should pay most attention to. A Samsung account has an upper limit of 20 million keywords. Generally, optimizers can only reach a few hundred to tens of thousands based on the size of the company, a few hundred for a small company, tens of thousands for a larger company, and hundreds of thousands for a larger company. My own account now has 300,000, but compared with the original account of 20,000, it has increased by 280,000. Even if 1% of the newly added words are displayed once, it can increase by 2,800 times. In fact, it is more than this number, and the corresponding conversion data can also be improved.

3. How important is customer service?

This is mainly for online consultation. Of course, the customer service's follow-up mining capability is also very important for other delivery methods. Here I would like to share this kind of online consultation, because the data in this area is relatively comprehensive, and the conclusions are drawn after many tests.

It turns out that our customer service staff does not work the night shift. After the above series of operations, the number of day shift conversations and the number of valid customers have recovered a little, but the evening and noon are blank periods, and the effect of form delivery is not good. After all, there is still a big difference between having customer service online and not having it. Otherwise, why do we need customer service online? We can just deliver the form directly.

Through the data accumulation in the first 1-2 weeks, and the current tight customer resources of the company, we were finally able to communicate well and the customer service staff were willing to work in the evening and at noon. After running the data for a week, the effect was obvious. The number of conversations increased by 1.5 times, and the number of effective customers also increased by 1.5 times. Data often speaks for itself and can convince many people who you could not convince before.

Communication skills are also an important skill that optimizers need to possess.

IV. Conclusion

The above is what I want to share this time. Finally, let’s make a summary:

1. It is best to place ads on PC and mobile separately because it is easier to analyze data and save costs. Most importantly, it is easier to enter OCPC, after which the effect is more stable.

2. Everyone can use the long tail theory. If flash creative is used reasonably, the effect will be good. Moreover, after Baidu adjusts the front-end creative display, the more creative styles submitted, the more styles will be displayed in the front-end combination. The multi-style combination display with pictures can occupy a larger position on the page. The most important thing is that there are many styles submitted, so you can better screen out high-quality ideas that are suitable for your own products, focus on optimization, and make the best use of others.

You should be more careful when using long-tail keywords. Add as many business-related words as possible and you will find good results.

3. Connect well with the company’s customer service department, especially for online consultation. The customer service’s reception capabilities and online time are very important. If customer service is reluctant to work, use data to convince them.

Advertising is not only a technical job, but also a delicate job, so it can be said that it requires a bit of craftsmanship.

Author: Aiqi SEM

Source: Aichi SEM (CMOcollege)