In Bennett's very first example, he states that companies with names which are easy to pronounce and remember tend to do well in the stock-market. I see a case of correlation, but not necessarily direct causation. I'm not inclined to believe that it is as easy as that. The product must be up to expectations as well, or initial interest would quickly dwindle. A simple, catchy, and memorable company name does not make the company successful in itself; it is merely one piece of an effective marketing strategy. Assuming approximately equal products, perhaps it is better overall marketing that offers these companies an edge over competitors. So, Bennett has opened up with a largely unconvincing argument, that is unless he wants to assure me that his perspective is a bit narrow-minded.
The article goes on to observe that 'cognitive fluency' and its result, easy mental processing, are effective tools in a world of excessive communication, making judgments on decisions quick and effortless.
The reasoning based on evolutionary history makes sense, but I feel that many, many arguments can be based on the strategy of coming up with a way that a trait may have been useful in prehistory. Because humans have no divine insight into the inner thoughts of evolution, it can only be a theory. The evolutionary biology majors in the crowd should enlighten us if this section of the article seems any more certain to you.
We students of cartography know that unfamiliarity is an immediate deterrent, and conventions (by definition familiar and usually very easy to interpret) are meant to be followed to ease access to the content and message of a piece of rhetoric and may be strategically broken for effect. Such conventions include features like font, one of the article's examples. Using a serif font to label water bodies is simply expected, and facilitates easy processing. Knowing immediately that the button with the triangle means 'play' helps a viewer get right to the content of a video, while an unfamiliar interface is simply a frustration.
While agreeing with the points made about the effectiveness of fluency, as an individual who thrives on the unconventional, I appreciate the acknowledgement that "disfluency is intriguing and novel" (Bennett quoting Winkielman), and I would be very interested to see some evaluation of the effective use of disfluency.