I understand there might be many factors, but I still find it quite odd that a search for "influencer marketing" shows this knowledge graph item. What are considering as relevant factors which are relevant in this case to justify an appearance before Wikipedia?
To begin, please consider the terms influencer marketing are marketing terms and not SEO terms.
What you are seeing is not necessarily from the knowledge graph. It is a featured snippet. More on that at the end. I will explain what is required for the knowledge graph then explain the featured snippet and how they are related.
There are several things that can help to populate the knowledge graph.
The first thing you have to know is that no knowledge graph card will be created without being vetted directly or having data taken from a highly trusted source. The second thing to know is that all data not directly vetted within the knowledge graph is checked for validity using methods of comparing content across many sites. Using a fact link semantic database, the knowledge graph will not incorporate information from a source that is not vetted without comparing the semantic fact links.
For example, George Washington was the first president of the United States creates, at minimum, George Washington -> President -> United States where the name George Washington is linked to President and President is linked to United States. As well, President can be ordered sequentially through fact links to completely represent the data. While this is an over simplification, it is clear that fact linking creates an opportunity to compare the data to other sources. In that way, if one site makes a claim using a factual statement as determined using semantic analysis, that fact can be compared to other sites with high trust scores making the same claim. Sites with factual statements that are trusted and vetted in this way can also contribute to the knowledge graph using a similar trust score mechanism as the domain trust score.
These are some of the primary factors in creating a knowledge graph card.
Google+ for business. This is Google's opportunity to vet your business directly. As a result of a Google+ account, any information you provide to Google will be vetted and may be directly used within the knowledge graph to build a card. Google+ is a trusted source that directly populates the knowledge graph.
Wikipedia A Wikipedia page is vetted by peers and checks made by content editors. It is the original source of trusted information and the first to be incorporated into the knowledge graph. Any Wikipedia page about a business or topic can appear in a knowledge graph card. Wikipedia ranks just slightly lower than Google+ as a trusted source.
Telephone directories. While there are a lot of business posting sites on the web, sites with original business listing content from the telecoms are also trusted as a source for the knowledge graph. While these sites rank lower than many sites, this data can also appear within a knowledge graph card. Keep in mind that simply posting a business on a business posting site is not enough. The site has to be trusted as an originator of the content, that is, from the phone company directly.
Data driven sites. Trusted data driven sites, such as the SEC (securities exchange commission), NASDAQ, DOW, Dunn and Bradstreet, weather sites, etc. can contribute data to the knowledge graph. However, from these sites, no card will be created as a result without one of the elements listed above.
These are only examples. There is more required.
You must also consider that the site must have certain trust elements. For example, valid NAP (name address, phone number) data often using schema.org mark-up. The NAP data must be relatively consistent spanning several sources including domain name registration, business listings, Wikipedia, the site itself, and others. As well, the NAP data must make sense. For example, listing a business address as a P.O. Box is not enough. Google requires a physical location as a walk-up location. The site registration information must be valid. Remember that Google is a registrar and can see the proper contact information for a site. All of this data must be valid and not contain contact information elements that are known to create spam or are deceptive. The site must also have valid About and/or Contact pages with options to contact the company using an e-mail address or form. The site must consist of RCS (real company s**t) which is structure, content, and behavior of a company site. One element of RCS is marketing content and activities optionally including articles, print advertisement, social media, etc. These things and more are required before a card will be created.
This topic is far too huge to get into all the details. Any one of us could write a book on the topic! However, this answer explains what influences the knowledge graph in enough terms to likely create a card.
A featured snippet will requires some of the above in that the site must have a reasonably significant trust score. This will include NAP as described above, About and/or Contact page, proper registration data, a reasonable link profile, etc. All of the standard trust metrics you would see of a quality site should be in order.
A featured snippet comes from the answer engine which is part of the query engine. There are two primary things required for a featured snippet to appear. One is that the search query appear to be asking a question or looking for a fact.
One common example is Italian restaurant in Atlanta Ga. where examining this query, restaurant in Atlanta Ga is a full semantic representation of the semantic elements required for a fact link; subject, predicate, and object. Restaurant is the object, in is the predicate, and Atlanta Ga is the object. Italian is a modifier of restaurant. It defines further what the user is looking for. The answer engine first solves restaurant in Atlanta Ga, with a fact link, then further solves Italian restaurant in Atlanta Ga by applying the modifier.
In order for a featured snippet to appear, the content must match the semantic query in the knowledge graph. This does not have to be as linear as you may think, though often it is fairly linear. I have seen two forms of featured links, one is a URL and snippet from a site that ranks well in the knowledge graph and would likely be used as a card. The other is a URL and snippet from a site that presents answers or facts. This is far more common.
In the case of a site that ranks well for a card, you can often also see a knowledge graph card for the site or another higher ranking site. Using my example, this would be a restaurant with good reviews within the Google+ profile which may be required.
In the case of a site that provides facts or answers, you will often find sites that are sources for answers or factual information and not a business site as described above. This could be a review site such as yelp.com, an informational site such as a Wikipedia, an answer site such as answers.com, etc.
If you follow the featured snippet link to the page, you will see a content block that represents semantic information that matches the query. Most of the time this is rather linear. It can be as simple as a header tag with Italian restaurant in Atlanta Ga. What is key in this example is in. You will notice that it is the same predicate as the search query and represents a fact link.
Some results are less linear. For example, a restaurant name and a location may be enough. For example, Luigi's Italian Restaurant as a header with schema.org mark-up for NAP should work.
As for asking a question in a query and the results, there is a different set of semantic clues that make this happen. The list is long, however, a simple example could be who is george washington. Like the previous example, the query represents a subject, predicate, and object, however, semantically, the use of the term who changes the result of the answer engine from a fact to an answer. Here is an example.
You will see that the result answers the question.
There are other forms of questions such as how to. What the answer engine looks for and the requirements changes again based upon the semantics.
It is all about sending semantic signals to search engines and being a trusted source of information.
So what is the difference between a knowledge graph card and a featured snippet?
The answer is relatively simple. A knowledge graph card offers information from the knowledge graph directly while a featured snippet offers content from the target site itself.
Again, this is a rather large topic, and again, any of us could write a book. However, paying attention to a sites trust metrics and sending the proper semantic signals can make any site appear as a featured snippet. For what it is worth, the bar is set lower for a featured snippet than for a knowledge graph card so it should be easier to get a featured snippet than a knowledge graph card.
This has been discussed a lot lately, not just in terms of what it takes to get a featured snippet, but also what it takes to maintain it. I have found a couple of articles from Moz very useful in both instances. In terms of what impacts it, this article is very thorough. In terms of how to get more, this article shows how to research unanswered questions which appear in a featured snippet.
We have found with our clients that they appear one day and disappear the next. We use Moz pro to track.