How to use online data combined with simple analytical AI to defend Czech e-commerce against Temu and high online advertising prices

Smaller Czech e-commerce players in particular are facing huge prices for advertising that they are unable to match, and many e-shops are facing extinction. Temu has dominated the top positions in keyword searches and Czech e-commerce, including traditional specialist retailers, are facing a drop in online orders of up to 70%.

As part of its massive campaign to conquer the Czech market and get the highest marketshare, online marketplace Temu has brutally raised the price of advertising on Google. The amounts that Czech advertisers have to spend to win auctions have risen by tens to hundreds of percent.

Smaller Czech e-commerce players in particular are facing huge prices for advertising that they are unable to match, and many e-shops are facing extinction. Temu has dominated the top positions in keyword searches and Czech e-commerce, including traditional specialist retailers, are facing a drop in online orders of up to 70%.

How can this be defended?

It's very difficult to do it through traditional marketing activities. The reason is that companies have to act quickly and such a quick solution may not be effective. The click-through amount influences the results that the user sees in search up to 80%, only the remaining 20% is influenced by relevance, so setting up your own campaigns in Google Ads will not bring improvements.

Pre-suit appeals are another way, but this is a rugged path and will only protect against targeting your brand directly, not against high click-through rates for search products.

The only way is to increase the maturity of how the e-commerce companies work with online data (MarketingData IQ*)

That is th reason, why we at Optimics have written what are the possibilities to defend Czech e-commerce and especially smaller Czech e-shops, against this massive online advertising spending using online data that every online entity collects. 

Compared to the very simple pouring of massive amounts of money into online advertising, often without much relevance, we offer a solution that cleverly combines AI and collected data and can wage a seemingly unequal David vs. Goliath battle.

This is not the first time we have seen price increases in advertising systems, before Tema we saw a significant increase in the price of online advertising due to the influence of Allegro, and we cannot expect that the situation will not happen again. Currently, we know that the Turkish giant Trendyol is preparing to enter the Czech market and has a massive budget for expansion. 

The use of data and AI/ML will help e-shops to increase their MarketingDataIQ and defend themselves once and for all. The described measures are also an excellent entry into a world without third-party cookies, and e-shops can secure their future in the world of first party data and prevent a future decline in online marketing efficiency due to the end of cookies.

How can you proceed at this time?

Google Ads

The first step to take is to look at the Auction Insight report in Google Ads, where you will see the "competitors" you have been in an auction with. With this information, you can then see if you're battling Temu or other advertisers in the auctions.

Broad match

Keyword targeting is one of the basic building blocks of PPC campaigns. Every PPC specialist has had to make a number of keyword matches over the course of their career. Besides the human factor, information about relevant search queries also enters into keyword matching. The problem with this solution is that it is highly static and does not evolve over time unless you decide to update the keywords.

In this case, Broad match could help you by enriching your search campaigns with relevant words. The selection of these recommended words is based on:

• User Intent

• User Location

• User's previous search queries

• Keywords in ad group

• Predicted performance

• Landing page

High-Value modelling

Understanding the user's journey to final conversion is key to relevant communication to users in the buying process. In this case, you can rely on your intuition to identify micro-conversions and report on those.

If you want to take a more data-driven evidence approach, then High-Value modeling is for you, using statistics to give you information on whether a particular action on the site is statistically significant in the buying process.

You can then use this information to create a custom formula to represent the value of a conversion, and you can then bid on that conversion in advertising systems like SA360, DV360, or Google Ads.

Soteria (+ High-Value modelling)

It's no secret that advertising systems are built on the data you provide them. It's common to see advertisers post, for example, the price of products with conversion. If you want to send e.g. product margin to advertising systems, we recommend using Server-side GTM or the Soteria concept for this.

This solution allows you to send valuable information such as product margin in a method that we are not able to capture e.g. using the networx in Developer tools.

In addition to margin, which has already been mentioned, there is also the possibility to send, for example, the value of a sample from High-Value modeling.

FeedGen

The product feed is the cornerstone of any e-commerce entity. If you want to dynamically remarket users with a relevant product, you need to work with the product feed in some way.

To enhance the product feed, FeedGen can help you write more interesting product descriptions or titles using LLM. At the same time, it is able to help you with the quality of the product feed, which can translate into campaign performance in the form of higher click-through rates or more conversions.

Lightweight MMM

The time when companies used only one communication channel is long gone. In today's competitive age, it is important to have the right communication mix for your target audience. And with a multitude of channels in the communication mix and user behavior, it is challenging to evaluate the contribution of channels.

Mostly this problem is solved by attribution models like last non-direct etc. This solution is more functional to some extent, it just doesn't tell you how to allocate the marketing budget between channels.

Lightweight MMM can solve this problem for you, you provide the data to the model and it will evaluate the effectiveness of a particular channel and then distribute the marketing budget among the current communication channels.