Before you start, we recommend you take a moment to cozy up with the data points and how they work.

You might also want to bookmark this page to return to later.

Not all dictionary endpoints count towards the rate limit.

We have 3 layers of Profile Data, each refreshed at a different rate.

For popular profiles, all report data is updated at least once a month. For less popular profiles, the data is updated on demand whenever a profile is pulled.

*Each profile is updated on its individual cycle.

The List Users and Search endpoint are updated once per month.

The Report endpoint is updated every 2-4 weeks.

While Modash displays all public profiles with over 1000 followers, we might not always catch every 1000 followers+ creator immediately - it can take a bit longer.

The collection of data for smaller accounts can be triggered by running a direct search via the API. If there is a specific one we don't have yet, you can search for it, and it will be added to Modash within 48 hours.

Note - For YouTube, we need 5000 views on average on the recent videos to calculate audience data and add it to Modash.

You can add the weight for Audience filters in search via the API. If no weights are added, the default weights will be applied.

The default weights are:

You can add up to 60 different locations in both the influencer and audience location filters.

The account types are:

For example, to get business or creator accounts pass value [2,3] to this filter.

The account type is something the influencers sets themselves on Instagram. It is not determined by Modash.

In the Bio filter, we search for exact word matches. So if you search for “food,” we only give you results with accounts that have “food” in their bio but not “foodie” or any other word containing the original word. You can also use emojis in this filter.

The Bio filter searches in bio as well as in username and handle.

The Brands filter finds influencers with sponsored posts from at least one selected brand. We try to provide a reasonable history of partnership data. Depending on the account age and the amount of content posted, this available data can span from a few months to a few years.

We track sponsored posts through Instagram’s Paid Partnership feature and hashtags like #ad, #sponsored, #paid, and others.

The Engagement Rate is calculated using the following formulas -

We usually try to consider two months, but sometimes we go back more than two months but never more than 500 pieces of content.

Pinned posts are treated as regular posts. Pinned posts are not included in the engagement rate if they are not part of the first 500 pieces or were posted later than 10 months before the current date.

Followers are shown as they are on the account’s profile on the social platform.

Note: Since we refresh our Followers count weekly, there may be discrepancies between what you get on Modash and what you see on the social platform.

This filter indicates how quickly the follower number of the creator is growing.

We take a snapshot of the follower count of the current month and the previous month and calculate the difference.

Keywords allow you to identify influencers by a phrase they use within the caption of their post.

Keywords filter only searches in text, not in hashtags. And it only searches for exact word matches. For example, when you search for “food,” accounts using the keyword “foodie” won’t appear.

You can only pass one string through the keywords filter. If you add multiple words in the string, for example, “cats dogs”, then we will give results where both cats and dogs have been mentioned. (The cats have told us they don’t appreciate this, but the dogs are super excited to have more friends.)

Non-English videos are not translated. You can search for the keywords in the original language.

We do not consider how frequently the creator used the phrase. If they mention the keywords once, it will come up. However, when you sort Search results by keywords, it is based on the frequency and relevance of the keywords used by the creators. It follows a similar mechanism to ranking search results in Google searches.

The Relevance filter can be used to search influencers by topics and by lookalikes.

The Topics filter looks for specific #hashtags used in the content caption. But it doesn't stop there. It connects "related" hashtags, e.g., the hashtag #car is related to #cars #carlovers #auto # automobile #carguy #porche, etc.

Some helpful details:

The YouTube exception: Topics for YouTube don't use hashtags you can see in the video's title or description. They are useless in terms of our Topics Filter. Since creators often copy-paste their videos' descriptions from one video into another, these hashtags aren’t typically related to the video's content.

Instead, we use something called "keywords." It's present in the video page's content and, basically, is in an array of strings: words, collocations, or even phrases. Typically, creators fill in those keywords to reflect what each video is about.

The Relevance filter works by proximity in a vector space - Lookalikes are based on the closest distance to the input (the influencer, in our case). Adding multiple influencers can go both ways —more and fewer Lookalikes than isolated ones.

To visualize this, imagine a “cloud of similarities.” The cloud is based on the influencers' topics, posts, etc.

When adding two similar creators, they’ll appear closer on the ‘similarities cloud.’

In contrast, adding two totally different creators gives you a broader search among them, and results would appear somewhere in the middle of the cloud.

If you combine topics and influencers in one request, we’ll try to find influencers close to the topics and the influencer in the request.

TextTags can be used to search by hashtags and mentions.

The Hashtags filter looks for specific #hashtags used in the caption of the content, while the mentions tag searches for @mentions in the caption.

The hashtag filter is "neutral.” It asks: Show me influencers we know have used the hashtag #hashtag. There’s no judgment about how often they’re used or if one of the hashtags is the main one or anything like that.

The search is based on all posts by a creator (which Modash has collected), regardless of when the content was added.

The Engagement Rate is calculated using the following formulas:

We usually try to consider two months, but sometimes we go back more than two months but never more than 500 pieces of content.

Pinned posts are treated as regular posts. Pinned posts are not included in the engagement rate if they are not part of the first 500 pieces or were posted later than 10 months before the current date.

For all social platforms on Modash, Engagements = average likes of the last 2 calendar months. We usually try to consider two months, but sometimes we go back more than two months, but never more than 500 pieces of content.

Pinned posts are treated as regular posts. Pinned posts are not included in the engagement value if they are not part of the first 500 pieces or were posted later than 10 months before the current date.

This indicates the influencer's location. We analyze location tags, languages, descriptions, captions, bios, and more to determine this.

We correct for travel by mapping the location consistently over time. Then, we determine the location by seeing where the creator is most regularly.

Gender estimation is determined by analyzing photos, captions, and names. For example, a creator with the first name “Sarah” who posts photos of a woman's face will be categorized as female.

Average reel plays of the last two calendar months. We usually try to consider two months, but sometimes go back more than two months but never more than 500 pieces of content.

Interests are based on the captions, @mentions, #hashtags, and location tags in the latest 150 posts or more than 150 posts if published within the last two months.

This metric compares the engagement rate of sponsored vs organic (non-sponsored) content. We compare average likes with average likes of sponsored posts for the same timeframe. The typical timeframe is last 2 months.

Some examples:

Sponsored content contains one of approximately 50 indicators that a post is sponsored. We have multiple ways to determine that:

Popular posts are the influencers' highest-liked posts among all the posts in the past 2 months.

For recent posts, we don't have thumbnails, because the thumbnail URLs expire in about 6 hours on Instagram and TikTok. RAW API is the way to go if you want the recent posts' image thumbnails.

Thumbnails are available for popular posts because there we collect and store the content on our end, so we can link you to the stored image.

The Credibility score describes the percentage of followers we consider real people. Using this score, you can also determine the percentage of Fake Followers.

For example, when the Credibility score is 0.8, the remaining 0.2 or 20% - are fake followers.

We determine fake followers by scoring accounts' behaviour vs average Instagram user behaviour. We consider the profile pic, bio, followers, following, engagement behaviour, and account age, among other things.

We also consider the ratios between these metrics (account age vs. follower to the following ratio). These factors contribute to a score. If an account scores too far away from “average,” it is marked as fake.

Fake followers and fake likers are determined using the same methodology.

Using a similar methodology as interests, influencer lookalikes display users that produce content around a similar set of topics.

To do this we analyze bios, captions, imagery, hashtags, mentions etc to create a map of the accounts topics. We then display other accounts that score similarly when put through the same process.

Stathistory is based on the data from the last 6-7 months. The latest month available is the month you pull the report. For example, if you pull a report in March, we will also include March.

You can use Stathistory to see the creator's Growth Rate monthly.

The average likes and comments are shown monthly, based on the date the content was posted.

We calculate averages the same way across Instagram, TikTok, and YouTube. Whether we’re talking about likes, views, or comments, an average is an average.

The average number is calculated based on the likes, comments, or views gathered within the last two months but no more than 500 posts.

If the influencer does not post often, we may take posts older than two months but no more than 10 months.

Language is determined by analysing the text the user uses in their captions, bio, etc.

Ethnicities are defined by us using image recognition. It is not from a free-form entry by the user.

Notables mean that followers or likers of an influencer have more than 1000 followers. The number can be turned into a percentage. For example, 0.07 = 7% of the creator’s audience has over 1000 followers.

Notable Users lists up to 20 accounts. We check the entire list of followers with the regular update cycles once a month.

Sorting order is by the number of followers.

For the Audience location, we return up to 10 countries.

This indicates the location of the audience. We analyze location tags, languages, descriptions, captions, bios, and more to determine this.

We correct for travel by mapping the location consistently over time. Then, we determine the location the account owner is in most regularly.

The 0-1 floating point value defines the percentage of the audience who follow the profile and is based on the specified city or country.

For a list of all locations and location IDs, check our Location Mapping documentation.

Leave “query” empty and add “limit” to 100000 to get the entire list.

Audience lookalikes are based on overlapping audience demographics. If your influencer targets “young people in California,” their audience lookalikes will target a similar demographic: “young people in California”.

The simplest way to explain it: An audience lookalike is somebody with a similar audience based on the people who follow that account. Audience lookalikes include gender, age, and location only indirectly.

Here’s an example:

If you’re a cat lover, you probably follow @cats_of_instagram and @catsloversclub.

When people follow both accounts, those two accounts become audience lookalikes.

Consequently, the demographic data is also very similar, but it's not the main criteria for calculating audience lookalikes. It's the implicit byproduct.

You can get up to 20 lookalikes. The sorting order is by similarity/relevance - the highest accounts are the most similar or best match for "lookalikes."