How to balance traffic in oCPC promotion?

How to control traffic under oCPC smart bidding? How can we optimize the conversation? In this article, the author would like to share his own views on this issue.

Currently, there are two types of oCPC bidding: oCPC and eCPC. I don’t need to say much about the difference between the two. As we all know, oCPC is fully managed with less intervention, while eCPC is semi-managed and can be slightly more flexible.

In my case, I have both oCPC and eCPC in my account. I learned from my friends that many people also use oCPC with eCPC or with CPC. From this we can see that bidding methods can actually be coordinated with each other, and there is not just one option.

As far as medical care is concerned, plans are basically divided according to the type of disease, and then divided into parts of speech within the plans. Or divide into multiple plans according to the type of disease, and then one plan represents one type of part of speech. I prefer the latter approach, which is more suitable for investing in medical care.

For example, I work in the Department of Traditional Chinese Medicine, and I mainly treat diseases related to water and fluid accumulation. The author's account structure is:

Pleural effusion - medical treatment term

Pleural effusion - method treatment surgery word

Pleural effusion - causes, symptoms, flow rate

Ascites...

This structure first separates the types of diseases, and secondly, one plan represents a type of word. When adjusting oCPC or eCPC later, there will be a clear traffic range.

I will not go into details about the scale and construction of oCPC and eCPC here. Let's talk about the problems and optimization methods that you will encounter in most cases after construction.

Generally speaking, after we set oCPC, the first stage is manual and the second stage becomes machine adjustment. Therefore, in the first stage, we usually adjust CPC as we did in the past, and oCPC will be the same as before. After entering the second stage, we often encounter three problems:

1: Unable to control disease consumption and conversations. There are no dialogues for the diseases we want, but there are many dialogues for the diseases we don’t want;

2: Unable to control the quality of conversation. Due to machine adjustments, it is often difficult to control the quality of conversation. Especially in the medical industry, many people use less than three sentences. If there is no deep conversion or three sentences, maybe 30% of them use less than three sentences every day. Of course, what I am talking about here is an average. I believe that many people use less than three sentences, and some people use less than three sentences.

3: Unable to balance the structure of the dialogue. For example, I want the main disease type conversations to account for 70% and the secondary disease type conversations to account for 30%, but this is difficult to control under oCPC.

This is a problem I often encountered before. Through some tests of my own, the results have improved recently. So let me share how I balance the conversation structure.

Budget, words and bidding method

Taking my case as an example, the disease that the hospital can convert best is ascites. This disease is easier to diagnose and the consultation is also fun. But there was less overall dialogue before.

Previously, my binding was like this: there were two plans for peritoneal effusion, one intention word plan and one flow word plan, and I built oCPC to bind these two plans. On August 12, I found a new batch of words related to peritoneal effusion, mainly flow words, and then created three new plans. Then because these three plans are new, there is no data accumulation. So I bundled the original package and added these three plans.

Then after three days, I split out the three plans. A new oCPC low-threshold package was created, binding these three plans. Since it is a new package and still at the first stage, there are still many things that can be adjusted. Because it was a traffic word, I lowered the price a lot, 30% lower than the account average price. After running for about three days, I entered the second level. The effect is quite good, the conversations are starting to increase, and the average price of the entire delivery package is relatively low.

In short, such adjustments mean adding words and consumption of main diseases. The original package data has been accumulated for a long time, and the conversion bid is set. The new package has a low threshold, so the competition between the two is not very fierce.

Details:

Planning and Budgeting

Because ascites is my main disease, I set a relatively high budget. Control the volume of conversations by controlling your budget. Less important disease programs have smaller budgets.

Expansion mode

Generally, the matching of active expansion is poor, but it is indeed possible to expand the volume. Generally, if I can’t spend, I will start actively. If you spend too quickly, set a balance. In addition, I will also compare the quality of active and balanced conversations. If active conversations are better and cost more, I will also set active. In short, my main considerations for automatic expansion are consumption and conversation quality.

Minor disease dialogue

Some customers are unwilling to have conversations and the conversion rate is poor. I usually set CPC or eCPC to control the volume. If I run CPC, I will set a lower price to control the planned budget. Although CPC is not very effective in expanding volume, it is still effective in controlling costs and consumption. If I invest in ECPC, my way of controlling it is to set short phrases for all words, not to open the intelligent core, to reduce consumption through premium settings, and to set a budget for the plan. The overall adjustment will be a bit bigger.

Key points: The traffic word plan matching can be narrower, just a phrase. The key word matching can be slightly broader phrase + intelligent core, which is convenient for first-order expansion and building a good data model.

This method is actually for important diseases. It expands the volume by adding words, increasing consumption and using oCPC, while taking quality into consideration. For minor diseases, consumption and costs are controlled through CPC and eCPC. Ensure that the data model does not go astray.

Package bidding, unpacking and page optimization

To balance the conversation structure, we can solve it through bidding, unpacking and page optimization.

There are 5 bidding forms: oCPC conversion bidding, oCPC data accumulation, oCPC low threshold, eCPC, and 0 threshold. I usually use conversion bidding, low threshold and eCPC. Data accumulation is not used much, and the 0 threshold is only used when the account cannot enter the second level.

For key diseases, I optimized it in this way. At the beginning, I would bundle a few more diseases into the package, and then the price would be a little higher, and I would choose to expand the volume in a balanced way. For example, ascites is my disease type, and Traditional Chinese Medicine is my industry word, as well as a brand word. So when I initially set up the oCPC delivery package, it will be combined with the ascites plan + industry word plan + brand word plan.

After the package enters the second stage and becomes stable for a week, I will separate the industry words and brand words. Then combine industry words and brand words into a package. In this way, the bag of abdominal effusion will flow more accurately.

If costs remain stable, I generally reduce prices by 5% every three days. The cost will gradually come down. If costs go up, I’ll increase prices by 5% or change the expansion mode to aggressive, while focusing on negative words.

Generally, I will set the bid for new packages a little higher to facilitate expansion. Wait until it stabilizes, then lower it a little. Sometimes it is adjusted every three to five days, sometimes it is adjusted every ten days.

Here are the bids and unpacking. One more thing to add, you should look at the negative words every day. If the overall situation is messy, change the active expansion to balance first. Then remove the negative words. If the overall situation is not too chaotic, then do a negative word every two or three days.

Then comes the page. In Jimuyu we can add disease content, activity content, and article content. We can add more content for key diseases. Add a little bit of content for each part of speech, and then you can generate multiple pages. Then for the types of diseases we want to expand the dialogue on, we need to make the correspondence between pages and words more detailed. Hospital words correspond to the official website consultation page, disease words correspond to the disease consultation page, cost and surgery words correspond to the activity page, traffic words correspond to the article page, and industry words correspond to the homepage consultation page.

When it comes to controlling the balance of conversation, there is another trick here. That is, for the delivery plan of key disease types, we need to make the page correspondence more precise, and check it once a week to see the words that are displayed. For non-key diseases, our requirements are not that high, but at least the parts of speech and pages are generally consistent.

There are mainly three aspects:

First, the conversion method. If the quality of a sentence is poor, we can set it to three sentences or one sentence + deep conversion.

The second is Jimuyu, the response time should be as short as possible. The preset words can be used to compare the effects. If the effect is good, turn it on; if the effect is not good, turn it off. Then there is the introductory statement, which you can study more about.

The third is consulting tools. You can find ways to improve the first sentence or related wording. For example, if a visitor sends you a message but does not reply, you can send him your WeChat, or add WeChat to the picture on the Jimuyu activity page or article page, etc. There are many methods, you can try them yourself.

In terms of lead conversion, there are not many operational areas in the Baidu backend, only a few tools in the creative component. The key conversion components are still Jimuyu and consulting tools, such as cards, coupons, and lotteries, so this should be the focus.

These are some of my thoughts on medical oCPC delivery, which are also real methods with obvious improvement in results. You can refer to them.

In fact, there are many ways to do oCPC now. It’s not that there is no solution. The key is to adjust it according to your account situation. Although oCPC has many shortcomings, we can't resist it. You can only keep looking for system rules and explore methods first, then you can be one step ahead of others.

Author: Captain

Source: Aichi SEM (CMOcollege)