Technical Data & Other Definitions - STILINGUE

Index:

1. Introduction
2. STILINGUE Smart Care and Proprietary Metrics
3. Listening
4. SLA for non-proprietary channels 
5. Social metrics
6. Number of followers 
7. Additional collections
8. Technical restrictions

Introduction 

Understand how the collection time operates in each STILINGUE module and how it relates to your research.

STILINGUE operates transparently so that you can access relevant data about your research. See below the main information about the collection time (SLA) and other technical details in different areas of the platform.

Customized Services: Need assistance in calculating and setting goals based on market analyses/trends? We can assist you! Our team of experts is ready to assist you. Contact your STILINGUE account manager to gain access to this and several other services.

STILINGUE Smart Care and Proprietary Metrics

For both modules, all collection channels have a 95% SLA between the average time and the time limit (indicated below). The collection expectation is 99.0% of posts within 24 hours after the time limit. These percentages should be calculated using a 24-hour filter for each type of publication and without any other restrictions on the filter, i.e., the overall volume should be analyzed for each type of interaction. 

1. Due to API call limitations, LinkedIn mentions are subject to a collection preference system. This means that if a page does not receive new mentions during the day, it is deprioritized. The same occurs when a mention does not receive new comments within 24 hours. If new interactions are found, prioritization is reestablished for new attempts

It’s important to highlight that the collection of mentions and comments on mentions on LinkedIn, the collection of comments on mentions on Instagram, and the collection of comments or videos that mention on TikTok are based on sample volumes. You can find more details in the documentation for LinkedIn, Instagram, and TikTok.

2. Due to limitations in the TikTok API, we confirm that any type of mention may take up to 24 hours to appear in the API for collection.

Listening

Regarding the collection time, that is, the time it may take for interactions to be collected:

95% of the data from the Search Configuration is guaranteed to be collected between the average time and the time limit, and 99% of publications are collected within 24 hours after the time limit. These percentages should be calculated using a 7-day filter for each channel, without any other restrictions in the filter—meaning the overall volume should be analyzed per channel.

Regarding the collection volume, that is, the total amount of data collected:

It is important to note that the collection volume will always be based on sampling. Even in these cases, the column “Collected Volume” will specify the conditions under which the volume is not sampled (such as certain network operators). This is due to existing API limitations.

SLA for non-proprietary channels 

1. Collection of comments or specific posts not belonging to the priority list is not guaranteed.

2. Additional time to the collection time of the previous table (Proprietary Metrics and STILINGUE Smart Care).

3. The priority list is managed in the Search Configuration. The user needs to input the link, blog RSS, or the username in the case of social media.

4. The Bluesky API documentation does not specify its delivery SLA. Based on the technical validations we conducted, we identified that we have access to approximately over 90% of the data. With this, we internally calculated an average collection time of between 1 and 2 hours. It is important to emphasize that since the API itself does not define a collection timeframe, there is still a possibility of not meeting this average.

Proprietary Social Metrics for Pages

Updated at least once a day. Supported for: TikTok and Instagram.

Proprietary Social Metrics for TikTok Video Mentions

The last 1000 posts from the past 90 days will be updated once daily. You can find more details on the specific page.

Proprietary Social Metrics for Instagram Ads

The last 500 posts from the past 90 days will be updated once a day.

Social metrics

Update of interactions on a post: likes, comments, shares, and video views.

*Minimum number of followers for the user making the post for social metrics to be updated.

For Twitter, after 3 attempts, if the tweet shows less than a 10% increase in the total variation of (likes + comments + shares), it will be prematurely discarded before the 7-day period.

Important: We have a block in our data collection flow for competitor pages (as we do not collect proprietary data for competitors), and the client should not list their page as a competitor page, as this will block the collection of metrics. Additionally, posts that have already been collected once cannot undergo retroactive collections (as they are already in our database), and all metrics will be updated within a 30-day period.

SLA for Collection Time and Volume for Non-Owned Channels

1. The collection of specific comments or publications that are not part of the priority list is not guaranteed.

• For YouTube and Facebook posts, profiles must be in our internal monitoring database if they are not on the priority list.

• For news, blogs, and Reclame Aqui, we rely on external partners or the functionality of the technique itself.

• Instagram comments are not delivered in open-web collection due to Meta's restrictions.

2. Additional time beyond the collection time mentioned in the previous table (Proprietary Metrics and STILINGUE Smart Care).

3. The priority list is managed in the Search Configuration. The user needs to add the link, blog RSS, or username in the case of social networks.

4. The Bluesky API documentation does not specify its delivery SLA. Based on our technical validations, we estimate access to approximately 90% of the data. Internally, we calculate an average collection time of 1 to 2 hours. However, since the API itself does not define a collection time, there is still a possibility that this average may not be met.

SLA for Querying Textual Descriptors in Listening

This table specifies the time it takes for each descriptor added to the search configuration to be queried in the API. The specified time represents the minimum and maximum interval between descriptor queries, which corresponds approximately to the minimum/maximum time in which new data may appear on the dashboard. It is important to note that this process may take less time than indicated, as mentioned in note ¹ below the table.

Only collections based on textual terms are included here. This means that other collections, such as Instagram and Facebook, despite being classified into groups, are activated only through monitored profile lists (due to API limitations).

For non-Enterprise Twitter and YouTube, the social networks do not specify in their documentation the internal sampling criteria used.

For Twitter Enterprise, RT collection is disabled by default, but if you need to collect this data, you can request service activation at no additional cost. It is important to note that the volume of RTs can be large, so consider this when requesting RT monitoring.

1 - Subject to Stilingue’s internal descriptor ranking algorithm: When a search keyword returns data, it is prioritized for more frequent queries, aiming to bring more data into the database. Due to internal processing optimizations, you may receive data faster than the specified minimum and maximum times.

2 - Potential data loss due to cloud or network instability: We have an internal SLA of 95% availability for the collected volume. Stilingue implements redundancy measures to mitigate possible errors and recover data from short-term temporary failures.

Proprietary Social Metrics for Publications

• Only the last 200 publications from the past 30 days will be updated at intervals of no more than 24 hours. The first update will occur within 2 hours after collection.

• This condition applies to all channels across all social networks.

• To increase the number of updated publications or extend the time range, you must contact your account manager (CSM) in advance to assess feasibility.

Important Notes:

• If the proprietary page is connected to multiple dashboards, regardless of the account, it is recommended to request an increase in this limit with your CSM.

  • The 200-post limit is shared across dashboards:

    • If connected to 2 dashboards, the effective limit is 100 posts per dashboard.

    • If connected to 4 dashboards, the effective limit is 50 posts per dashboard.

• Connecting a page will result in a slight additional token consumption, which should not be an issue for pages with daily engagement.

Grouping Rules for Updated Posts (200 posts per group, last 30 days)

• Instagram:

  • Organic: Posts, Reels, and Carousels
  • Dark posts
  • Mentions (in text)

  • Photo tags

  • Stories are not updated

• Facebook:

  • Dynamic posts
  • Mentions
  • Reviews
  • Organic posts
  • Visitor posts
  • Dark posts

• TikTok:

  • Organic posts

• Twitter:

  • Organic posts
  • Mentions

• YouTube:

  • Organic posts

Important: If your page receives or performs more interactions than the total defined per month within a 24-hour period, the metrics update is not guaranteed.

If metrics reset for any reason, they will not be set to zero during updates for security reasons. Instead, the most recent non-zero value provided by the APIs will be retained.

Social Metrics in Listening

Update of interactions made on a post: likes, comments, shares, and video views.

Minimum Followers for Social Metrics Update

• Minimum follower requirement: The user posting must have a minimum number of followers for social metrics to be updated.

• Twitter-specific rule: After three attempts, if a tweet shows less than a 10% increase in the total variation of (likes + comments + shares), it will be prematurely discarded before the 7-day period, meaning it will no longer receive updates.

• Data collection restrictions:

  • We have a block in our data collection flow for competitor pages (as we do not collect proprietary data for competitors).

  • The client should not mark their own page as a competitor, as this would block their metric collection.

  • Posts that have already been collected once cannot be collected retroactively since they are already stored in our database.

  • All metrics will be updated within a 30-day period.

Number of followers 

The number of followers for a profile is updated each time the post is collected, provided that the profile making the post has a minimum number of followers according to each platform:

Additional collections

* When making changes to the configuration of a created search dashboard, this is the time it takes for the team to reorganize the previously collected data based on the new settings. 
 

Technical restrictions

• Comments on old YouTube posts: 

Proprietary Posts Only: Data collection is limited to proprietary posts within the last 100 posts from the past 30 days.

Exclusion of Old Comments: Comments made on older posts are not collected.

Non-Proprietary Pages: Comments on non-proprietary pages are not provided by the API.

YouTube Post Duration: If you need to extend the collection period for YouTube posts, contact your CSM to discuss feasibility.

• Collection of posts and comments on connected pages: 

There will be a loss of data collection on pages not associated with a dashboard or pages that have lost authentication. Warnings are sent daily to clients when the system detects authentication failures.

Non-proprietary Channels (Listening)

• Twitter Collection:

 We collect the latest 300 posts/mentions at a time. If the number of posts in a collection cycle exceeds 300, there may be loss. In these cases (large sports events, real-time shows, debates, or high-viewership TV programs), we recommend strengthening via Twitter GNIP Powertrack (Twitter Enterprise), as a complement. By subscribing to the Twitter PowerTrack API, there is a 100% delivery guarantee for items. By default, the collection of RTs is disabled, but if there is a need to collect this data, you can request the activation of the service at no additional cost. 

Collection limits 

What factors are taken into account in the collection limit of publications? 

 The consumption of the publication collection limit for each plan is influenced by several variables that depend on what has been contracted by the client.

To calculate the collection limit, it is necessary to sum the total number of publications from all modules contracted in the system. New modules contracted by the client or launched by STILINGUE in the future will always consume the contracted collection limit. Currently, the modules that consume the publication collection limit are: STILINGUE Smart Care, Proprietary Metrics, Listening, and the Compare Fanpages (Facebook) section within Proprietary Metrics.

When the total collection limit of the panel is reached, the customer will be notified by email, and their collection will be paused. The customer is responsible for managing and utilizing the collection panel. It is advisable to prevent the panel from reaching the limit, as this may result in potential collection losses due to insufficient space. 

STILINGUE is not responsible for data losses caused by a lack of panel limit management. Customers can delete posts or upgrade their plan to avoid collection pauses.

Technical requirements 

STILINGUE's data is replicated on the platform in real-time to ensure integrity in cases of failures at specific points. Furthermore, a backup policy has been developed to allow data recovery in the event of major failures. 

The STILINGUE War-Room platform is exclusively marketed under the SaaS (Software as a Service) model, a form of software distribution and commercialization. Within this model, STILINGUE is responsible for providing the necessary infrastructure for the customer. 

The technical requirements for the STILINGUE platform are as follows:

1. Official support is provided for the latest versions of Google Chrome. Other browsers may offer portability, but official support applies exclusively to Chrome.

2. The official version is supported by the manufacturers of Microsoft Windows and Apple Mac OS operating systems.
3. ALL operators must have a computer with a minimum of 4GB of RAM (8GB recommended) and internet with a minimum speed of 10 Mbps. Recommended: 30 Mbps.
4. It is necessary for ALL operators to have an email for login and INDIVIDUAL use of the tool.
5. The release of some tools/plugins that assist us in monitoring usage and system bugs is mandatory for all operators. They are:
   1. Zendesk (ticket tool) *.zendesk.com. *.zendesk.com.br
   2. Logrocket (tool for automatic bug reproduction): *.logrocket.com, *.logrocket.io 
   3. Sentry (automatic bug tracking tool): *.ravenjs.com
   4. Social Networks (Facebook, Instagram, Twitter, YouTube):
      • *.facebook.com , 
      • *.instagram.com , 
      • *.twitter.com,
      • *.google.com ,
      • *.cdninstagram.com, 
      • *.facebook.net, 
      • *.tenor.co , 
      • *.fbcdn.net , 
      • *.ggpht.com
   5. SendGrid (Platform email dispatch tool): em2572.stilingue.com.br, url7811.stilingue.com.br 
   6. STILINGUE: ai.stilingue.com.br, *.stilingue.com.br. The release of other tools/plugins is desirable:
      • Google Analytics *.doubleclick.net , *.google-analytics.com
      • Github *.github.io 
      • Rollout: *.rollout.io"

For more information, visit the discussion on the subject at our community or videos on our channel. 😃