Let me tell you a fact. The current expansion of search engine marketing (SEM) service providers is not because they have excellent advertising ideas or technical innovations, but because you believe their exaggerated claims.
Although no SEM service provider is willing to admit this, this cold fact is still true, even those SEM service providers with original or patented technology. For those top SEM service providers, the situation is still the same. They spend half of the time using automatic bidding tools, and the other half of the time rely on manual verification "whether the tool is executed correctly." This is because they take their customers seriously. situation.
So, what are the flaws of automated bidding tools? Why are they ineffective?
Most automated bidding tools can only handle a certain amount of data (no matter how much the SEM service provider brags about their products), such as your keyword performance over a 7-day period or a 30-day period. Service providers usually record your CPA (cost-per-acquisition) for a period of time (such as 30 days), and then evaluate how to improve your rankings in the past 7 days, and how to operate (close or pause) your key Word placement.
But just consider one situation──── 100,000 keywords × 5 matching types × 3 ads × several delivery channels (Google, Yahoo, MSN...), these service providers are unable to do anything, because this requires a lot of effort Based on a lot of data. You may wish to investigate the SEM service provider's team to see if they have this capability. The results of my research...well, I don't think an MBA who doesn't understand machine learning at all can handle all of this. In my opinion, this requires a person who fully understands how Google's advertising system works rather than how SEM service providers work. At the same time, he needs to be able to obtain a PhD level from Stanford.
If SEM service providers want to develop a tool system with a large amount of data, then they need to develop a complex system in depth. This system will be full of various indicators. The system also requires 45 days of data and up to 1 month of data. year's data. However, this method has inherent flaws, because many advertising accounts are less than 6 months old and cannot support the 1-year data standard, and some advertising accounts have CPC (cost-per-clicks) and CPA (cost- per-acquisiton ) has no rules at all.
Other SEM service providers observe the effect by increasing or decreasing the keyword price, but this method may be useful when your keyword price is about 1 yuan per click, and when you When the keyword price is 10 bucks per click, this approach will have wild consequences, and eventually you will find that you have to raise and lower the price by a few bucks to see a change.
Another example is if you have an online store that sells multiple products, has multiple ad campaigns, and tons of keywords. You usually use cross-subsidization to balance the profit gap between different products, but automated bidding tools usually cannot do this because the goals set in the tools are often the same.
So, no matter how amazing you hear SEM vendors say about their tools, I just want to say: “Damn, thank you! Goodbye!”
Editor's note: There has always been great controversy over how to evaluate the effectiveness of SEM. This is a common situation. But there’s also a universal truth: a controversial market is also a market with huge opportunity.
Original author: Nick Abramovic
This article was first published in Search Engine Marketing Observation. Please register the source for reprinting.