Crealytics Insights

Break down silos in ecommerce to drive performance

Bigger, Better, Faster, Stronger: SMX West Returns!

SMX West, 2018

March 13 – 15,

San Jose Convention Center

Get the latest expertise on all things AdTech, including a talk by Crealytics’ founder and C.E.O, Andreas Reiffen.

As with every year, this event promises a spectrum of search-aficionados from lots of different backgrounds. Expect the hottest Search topics of 2018!

We can’t wait to share our latest research with you.

You can join Andi and three other SMEs on Thursday, March 15th, 9.00am – 10.15am

What defines success for commerce-driven marketers? We’d say it’s about increasing revenue while maximizing profit. But it’s not always straight-forward.

“Your Mission: Increase Sales, Cut Costs, Maximize Profits” will teach you:

  • How to refine ad extensions for optimal payoff.
  • How to streamline product feeds for maximum efficiency.
  • How to boost product desirability.
  • How to find new ways to drive revenues and profits!

We hope to see you there!

Amazon 101: An Overview of the Company’s Product Ranking Algorithm

For many online retailers, optimized searches for Amazon have more importance than Google. Get this. Amazon has three times the search volume for products. Listing items on the platform? You’ll need to get them to the top of the platform’s search results (in relevant categories) to drive sales. And you’ll can do this better if you understand how Amazon’s algorithm works.

Unfortunately, many retailers operate blindly. It’s possible to enjoy success on Amazon without understanding their ranking algorithm. But the more you know, the better your products will perform.

So, what is Amazon’s algorithm? And how does it work?

Amazon’s A9 Product Search Algorithm

Amazon’s product search algorithm has a mysterious name: “A9.” The company shies away from discussing how it ranks products (to keep people from gaming the system). However, it admits to its algorithm making calculations long before a customer inputs a search query. A9 takes lots of data into account; examples include past traffic patterns, product descriptions, and individual user’s browsing habits.

Once the user types in their search, A9 analyzes the product listings…and delivers the products it determines to be “most relevant.” A product’s relevance to a given search query may differ from customer to customer.

Nonetheless, ranking on Amazon is much more straightforward than it is in Google.

What is the main difference between the two? While Google seeks to display results that accurately answer the searcher’s query, Amazon focuses on delivering the product listings most likely to be bought by a customer.

Let’s explore some of the core factors influencing Amazon’s A9 algorithm:

Core Considerations for the Amazon A9 Algorithm

A few core factors drive search listings through Amazon’s platform. These include:

Conversion Rate

How well does the product sell once users have been directed to the product page? Products that convert at a higher rate place higher in the search results. Many aspects of your product page can drive conversions. These include product price, customer reviews, and the quality of your product images. Always look to improve your product page, and keep it optimized for conversions. Amazon analyzes your conversion rate for every specific query that brings a searcher to your product page.


Is your product relevant to shoppers’ search queries? Conversion rate plays a role in this because relevant products tend to sell well. However, a product’s name and description can play a big role in determining its relevance. Try to laser-target specific product categories and searches (with your product information and text) to improve your relevance within Amazon.

Customer Satisfaction

When customers buy your product through Amazon, are they generally satisfied? Are your customers’ reviews mainly positive? What percent of your purchasers end up returning the item? While your overall review score does impact a product’s ranking, Amazon wants to see reviews that skew toward the top end of the scale. For example, two products might have the same overall score, but a product that has several three and four-star reviews will generally place higher than a product with lots of one-star and five-star reviews. Amazon prefers consistency.

Order Retention

Amazon knows that the best kind of customers are those that keep returning. For that reason, order and customer retention play a key role in product rankings. When customers buy your product, do they return to buy accessories? Do they commonly buy other products along with yours? Amazon prejudices its search results towards products that generate more business for them in the long-term.

Amazon Optimization: Simple…But Important

Too many eCommerce companies simply list their products on Amazon. They fail to think about how they can optimize their product pages. Opportunities abound! Remember – most optimization occurs on the product page. A minimum amount of attention to Amazon’s ranking factors can quickly improve your rankings—and grow your revenue.





Frequently Asked Questions

Here are answers to the most frequently asked questions about the Crealytics Product Ad Management tool, Camato.

General Questions

Q: Can we import “real revenue” into our system?

Assuming real revenue = revenue that attributed over multiple touch points on possibly multiple channels using a conversion attribution system like ExactTag or Ad Clear → we can import data from 3rd party systems as a custom, tailored service.

Q: How can we deal with seasonalities?

There are several ways. For well-performing brands, new collections get a default bid, then they “inherit” the performance of the brand’s most similar products. Using the Inventory Management, new seasonal products can be pushed in a more aggressive way to collect data faster.

Q: Is it possible to use Analytics RLSA lists?

Yes. Audience lists may be defined AdWords or in Analytics (and imported from Analytics into AdWords). Customer Match is possible, too.

Q: Is it possible to rename Camato campaigns?

Yes, Camato campaigns can be renamed directly in AdWords as Camato is working with the campaign IDs.

Q: Are email alerts/notification possible? 

Yes, just click on “profile settings” (top right corner) and activate it.


Q: Is it possible to pause the Camato Generics campaign for shopping? Or do we need to bid down to 0,01 cent?

We do not recommend to pause the Camato Generics campaign as traffic would then probably flow into the Camato Fallback campaign. If performance is too low in Camato Generics we recommend to check the search query report for the low performing search queries and to add these as Exact Match Types to a new Negative List which is assigned to Camato Generics.

Q: How can we make sure we only show one special brand and the rest of the brands not? Can we pause these brands or will they flow into Fallback?

These issues could be regulated by bid management which means pushing good performing ad groups and cutting cost drivers. – OR –

Add all the brands including brand misspellers to a new negative list and apply this list to all three Camato shopping campaigns. -OR –

Remove the products you don’t want to show in the feed (using Feed rules in the Merchant Center,…)

Q: You mentioned that you pick the feed twice a day. Is this just for new products or out of stock? What times do you pick up the feed?

We pick the feed twice a day. In the morning and in the evening. We only need the feed to check for changes in its structure or product stock. For example when you add or delete brands or a product is OOS.

Reporting, Data & Attribution

Q: Why there is a discrepancy between Adwords and Camato data by the Crealytics pixel?

The discrepancy between Adwords and Camato data is normal and is to be attributed to the unreliability of the pixel. Usually, this discrepancy is around 10% which is not alarming. You can calculate it and come back to us in case it is higher than 10%.

Q: Why there is a discrepancy between Adwords and Camato data by GCT?

It might happen that due to a gap of time between camato pulling data from AdWords and you pulling data from AdWords and due to Attribution there is a difference in performance data e.g. for Conversions.

Theoretically it might also happen that “last 30 days” of camato are not equal to the “last 30 days” of AdWords. When data from yesterday is not available while camato caluclating bids, the “last 30 days” in camato could then differ by -1 day from AdWords’ “last 30 days”.

Q: How do you calculate ROI in Camato for products ads?

It depends on which optimization goal you set. If you select CLV the ROI formula is: (Margins + New Customer Value – AdWords cost) / AdWords cost, if you select Profit or Revenue the ROI formula is: Margins / AdWords cost.

Q: Could you explain what is your definition of Margin and Profit in Camato? 

Margin = Net sale amount – wholesale price

Profit = margin – cost

Q: What kind of attribution model does Crealytics use? 

Our attribution model is last click: the last touchpoint of a customer with the shop get 100% of the conversion granted.

Q: In which way does the estimated optimization base per click differ from actual (historic) revenue per click? I assume this is based on a forecast model?

It depends on the numbers of clicks on the product partition. See the section KPI calculation. In a nutshell: if a product group has not enough clicks, we look at similar products from the same brand by taking the google hierarchy into account.

Q: How do we track CLV?

For each conversion we track whether it was done by a new customer or an existing customer using the crealytics tracking pixel. A new customer value can be defined or dynamically submitted – the conversion value is then increased and attributed back to the keyword / product group.

Q: How do we track Returns / Cancellations?

We enable  the customer to do either a) specify a fixed Return and Cancellation rate (Reducing Revenue / Margins by X% for each conversion) or b) use our API to transmit a returns and cancellations feed. Problem for usage in BM with b) is always: returns can come in very late, let’s say up to 50 days – and BM will not see them because the respective conversion value may be long long in the past.


Q: How does the yield management work?

The Inventory Management algorithm increases / decreases the estimated Value per Click for the labeled products. It pretends that products you want to sell perform better and products you want to not be advertised on as aggressively perform worse. That in result triggers a more aggressive or a more cautious bid.

Q: Can you change the return rates per project/customer? Are the returns/cancellations being dynamically sent to the pixel or can we set them somewhere?

The return rate is a back-end setting and cannot be dynamically sent to the pixel. However, you can decide a different return rate per project/customer and we can customize the back-end setting accordingly.

Bid Management

Q: Which values do the Bid Management tool take into account and how are they weighted?

The last X days, with X being 30 as default. We look at several signals like the last known CPC and estimate a value per click. The estimated value per click is an exponentially weighted average that follows a time decay preferring newer conversions over older ones.

Q: Can Camato adjust the RLSA bid modifiers based on performance?

Currently, Camato sets only initial modifier values specified by the user. In a not so distant future it will adjust these automatically as well.

Q: In your bidding algorithm, can it take care of average return rates? For instance, we know margins are high within jeans, but at the same time jeans have a high return rate.  

Our profit calculation (behind Camato) is as followed: Profit = margin_after_cancellations – cost. Would mean we work on an average return rate which can be set on market level.

Feed Changes

Q: We’ve changed prices in the feed, but Google is still showing the old prices. Is there something you can do about it? 

No. As with most information, Google is getting the prize information directly out of the feed. If the feed data changes, it can take up to 48 hours before Google is showing updated ads.


Q: My Camato designer negative lists do not contain any negatives. Why?

You’re running Camato in a new account probably. It may take a few days until Camato automatically ads negatives to the list. If you run Camato in an existing shopping account or if the lists do not get filled within one week, ask one of the developer!

Q: Why does Camato create 3 new negative lists if it only uses one?

It’s a relict of older days where a different set up was in place. Unfortunately it’s not that easy to teach camato that it shouldn’t create these lists. That’s why they still get created.

Q: Do negatives for shopping campaigns work differently for shopping campaigns? Specifically, will they block the products versus the user search term? I am assuming there are other ways to filter products from being shown other than negatives or taking them out of the product feed entirely?

No negatives work as always and they block search terms. As far as i know there are no other ways than negatives and deleting the product from the product feed.


If you have any other questions that aren’t answered here, please email us at