When working with combined marks (those that include both verbal and visual elements), it’s important to understand that Markify uses two separate algorithms:

  • Wordmark algorithm – focusing on orthographic and phonetic similarities, analyzing the word structure and pronunciation

  • Image algorithm – trained on visual similarities between logos, focusing on shape, structure, color, and layout, while excluding OCR (text recognition).

Both are AI-driven and statistical in nature, meaning they predict potential conflicts based on how often similar cases have led to opposition or citations. Each focuses on a different component of the mark - so to get the most accurate results, both search algorithms should be used

(Please check a dedicated article on Markify algos here)


1. When to run two searches or order two watches

If both the verbal and visual elements of a combined mark are distinctive and relevant to your brand identity, it is strongly recommended to order two separate searches or watches:

  • Wordmark search/watch for the text part

  • Image search/watch for the logo part

You can later combine or cross-reference the results if needed (please check the dedicated articles on merging ProSearch searches and merging Trademark Watch reports)

This dual approach allows each algorithm to perform its intended function - the wordmark algorithm for linguistic similarity, and the image algorithm for visual similarity.


Below you can see a few examples of the logos where we would recommend* ordering two searches or watches, as the text part is stylized or intergated within the logo: 


2. When a single search or watch may be enough

If one of the components (word or image) is not distinctive or not your focus, you may only need one type of search.*

For example, if a combined mark consists of a verbal element + a very plain text logo (the same name written in black-and-white font with no additional visuality), it might be better to only order a wordmark search or watch, as image watch is likely to create unnecessary ''noise''.





3. Is it necessary to crop out the text part from the logos when configuring search/watch parameters?


Usually, no cropping is needed. The image algorithm will process your original logo as it is, identifying similar marks based on the full visual composition.
It’s beneficial to keep the text part if it has some distinctiveness in its style, font, or placement relative to other logo elements.
(Please note that the image algorithm does not perform text recognition, so a separate wordmark search/watch is still required.)

Results ranking and volume will also slightly differ in case you keep or crop out the text from a logo like below:


That said, in some cases, cropping the text can help reduce noise - for example, if your logo contains a long, generic slogan that dominates the visual layout.
In such cases, we recommend* cropping the slogan and focusing the image search on the distinctive visual element (while searching the text separately).

There's an example below: if you preferr the algo to concentrate on the drone visual only, we recommend removing the textual part for the image search/watch + searching the verbal part


When in doubt - focus on distinctiveness: what makes your mark recognizable should define your watch setup.


_______________________________

*Please treat the below as Markify’s recommendation rather than a strict instruction. The final decision on whether to set up one or multiple searches/watches rests with you