Social Media Monitoring
1. Overview
This process analyzes a collection of social‑media posts that mention the brand or campaign, determines the sentiment (positive, neutral, or negative) for each post, identifies the most frequently discussed topics, and produces a concise report that highlights overall sentiment trends and the top topics discussed.
Scope Boundaries
- The process only uses the social‑media posts supplied as input; it does not retrieve data from external websites or services.
- Sentiment is determined using a predefined list of positive and negative words; no machine‑learning model is used.
- The process does not make subjective judgments beyond the defined sentiment rules.
2. Business Value
- Immediate insight into how audiences feel about the brand or campaign, enabling quick reaction to positive or negative chatter.
- Identification of key topics that can guide content creation, advertising focus, and crisis‑management decisions.
- Data‑driven reporting for senior marketing and leadership teams, supporting strategic planning and ROI measurement.
3. Operational Context
- When to run: after a product launch, marketing campaign, or on a regular schedule (e.g., weekly) to monitor public perception.
- Who uses it: Social Media Manager, Marketing Analytics team, and Public Relations staff.
- Frequency: Typically once per week or after a major brand‑related event, but can be performed on any ad‑hoc basis when a quick sentiment check is needed.
4. Inputs
4.1 Social Media Posts
- Name/Label: Social Media Posts
- Type: List of individual social‑media entries
- Details Provided: Each entry contains the platform where the post was made, the author’s handle (or name), the posting date, the full text of the post, and (optionally) the number of engagements (likes, shares, comments, etc.) associated with the post.
Social Media Posts (example format)
| Platform | Author (Handle) | Date (YYYY‑MM‑DD) | Content | Engagement (optional) |
|---|
| Twitter | @johnDoe | 2025-08-01 | Love the new product! Great performance and easy to use. | 120 |
| Instagram | @janeSmith | 2025-08-01 | Not impressed with the service. Too slow. | 85 |
| Facebook | @bob | 2025-08-02 | The product is okay, nothing special. | 50 |
| Twitter | @alice | 2025-08-02 | Great experience! Fast and reliable. | 200 |
| Instagram | @charlie | 2025-08-03 | I love the new features but the battery life is disappointing. | 150 |
Only the information listed above is required for each post.
5. Outputs
5.1 Sentiment Distribution
- Name/Label: Sentiment Distribution
- Contents: A table summarising how many posts fall into each sentiment category and the percentage each category represents of the total posts analyzed.
- Formatting Rules:
- Use the headings “Sentiment”, “Count”, “% of Total”.
- Percentages are rounded to the nearest whole number.
Sentiment Distribution
| Sentiment | Count | % of Total |
|---|
| Positive | 2 | 40% |
| Neutral | 2 | 40% |
| Negative | 1 | 20% |
5.2 Top Topics
- Name/Label: Top Topics
- Contents: A table listing the most frequently mentioned topics (single‑word keywords) after removing common stop‑words and punctuation. The table shows the topic, the number of times it appears, and the percentage of total posts in which the topic appears. The list is limited to the top five topics by frequency.
- Formatting Rules:
- Use the headings “Topic”, “Count”, “% of Posts”.
- Percentages are rounded to the nearest whole number.
- Topics are ordered from most to least frequent.
Top Topics
| Topic | Count | % of Posts |
|---|
| love | 2 | 40% |
| new | 2 | 40% |
| product | 2 | 40% |
| great | 2 | 40% |
| experience | 1 | 20% |
5.3 Summary Narrative
- Name/Label: Summary Narrative
- Contents: A short, neutral‑tone paragraph summarising the overall sentiment and highlighting the top topics identified.
- Formatting Rules:
- Write in plain, professional language.
- Use 3‑4 sentences; no bullet points or tables.
- Include the percentage breakdown for each sentiment category and mention the top three topics.
Example Summary Narrative
The overall sentiment for the recent period is mixed, with 40 % of posts positive, 40 % neutral, and 20 % negative. The most frequently discussed topics are love, new, and product, each appearing in 40 % of the posts. Positive feedback emphasizes the product’s quality and performance, while negative comments focus on service speed and battery life concerns.
6. Detailed Plan & Execution Steps
-
Validate Input Presence
- Verify that the “Social Media Posts” list is provided and contains at least one entry. If the list is empty, proceed to step 12.
-
Check Content Completeness
- For each entry, confirm that the Content field is not empty. If content is missing, flag the entry for manual review and exclude it from analysis.
-
Standardise Text
- Convert all post text to lower‑case.
- Remove punctuation and special characters (except hashtags and emojis, which are retained for potential topic analysis).
-
Determine Sentiment for Each Post
- Scan the post for any positive words (see Appendix C – Sentiment Lexicon).
- Scan the post for any negative words.
- Assign sentiment using the following rule set:
- Positive – contains at least one positive word and no negative words.
- Negative – contains at least one negative word and no positive words.
- Neutral – contains both positive and negative words or contains none of the listed words.
-
Count Sentiment Categories
- Tally the number of posts classified as Positive, Neutral, and Negative.
-
Extract Topics
- Split each post’s text into individual words.
- Remove common stop‑words (see Appendix B – Stop‑Word List) and any word shorter than three characters.
- Count the frequency of each remaining word.
- Keep the top‑5 most frequent words as “Topics”. If there are ties for the fifth position, include any tied words.
-
Calculate Percentages
- For each sentiment category, divide the count by the total number of analysed posts and round to the nearest whole number.
- For each topic, compute the percentage of posts that contain the topic at least once.
-
Prepare the Sentiment Distribution Table (see Section 5.1).
-
Prepare the Top Topics Table (see Section 5.2).
-
Compose the Summary Narrative (see Section 5.3).
-
Output All Results in the order: (a) Sentiment Distribution Table, (b) Top Topics Table, (c) Summary Narrative.
-
No Data Scenario
- If the “Social Media Posts” list is empty, produce the following message:
“No social‑media data was provided for analysis. No output generated.”
-
Quality Assurance (see Section 7).
7. Validation & Quality Checks
- Data Presence: Confirm at least one post is provided; otherwise, trigger the “No Data” scenario.
- Content Check: Ensure every processed post has non‑empty content; flag missing content for manual review.
- Sentiment Totals: Verify that the sum of the Positive, Neutral, and Negative counts equals the total number of processed posts.
- Percentage Sum: Ensure the % of Total column in the Sentiment Distribution Table adds to 100 % (allowing a 1% rounding discrepancy).
- Topic Frequency: Verify that each topic in the Top Topics Table appears in at least one post.
- Duplicate Handling: Count each occurrence of a word per post once, even if it appears multiple times in the same post, for the % of Posts column.
- Edge‑Case Handling:
- Posts containing both positive and negative words are classified as Neutral.
- If a post is entirely made of stop‑words or symbols, treat it as missing content and flag it.
- If more than 1,000 posts are supplied, only the first 1,000 entries are processed; the remainder is ignored with a note in the summary.
8. Special Rules / Edge Cases
- Empty Input – When no posts are supplied, output only the “No social‑media data …” message and stop.
- Missing Content – Any post with a missing or empty Content field is omitted from the analysis and listed in a “Review Required” note in the summary.
- Mixed Sentiment – If a post contains both positive and negative terms, classify it as Neutral (no bias toward either side).
- Multiple Languages – The process only supports English‑language posts; posts in other languages are flagged for manual review.
- Unsupported Platforms – Posts from platforms not listed in the “Supported Platforms” list (Appendix C) are ignored and noted.
- Large Volume – If the input contains more than 1,000 posts, the process only analyzes the first 1,000 and adds a note in the summary that additional posts were excluded.
9. Example
Input (Social Media Posts)
-
Platform: Twitter
- Author: @johnDoe
- Date: 2025-08-01
- Content: “Love the new product! Great performance and easy to use.”
- Engagement: 120
-
Platform: Instagram
- Author: @janeSmith
- Date: 2025-08-01
- Content: “Not impressed with the service. Too slow.”
- Engagement: 85
-
Platform: Facebook
- Author: @bob
- Date: 2025-08-02
- Content: “The product is okay, nothing special.”
- Engagement: 50
-
Platform: Twitter
- Author: @alice
- Date: 2025-08-02
- Content: “Great experience! Fast and reliable.”
- Engagement: 200
-
Platform: Instagram
- Author: @charlie
- Date: 2025-08-03
- Content: “I love the new features but the battery life is disappointing.”
- Engagement: 150
Expected Output
Sentiment Distribution
| Sentiment | Count | % of Total |
|---|
| Positive | 2 | 40% |
| Neutral | 2 | 40% |
| Negative | 1 | 20% |
Top Topics
| Topic | Count | % of Posts |
|---|
| love | 2 | 40% |
| new | 2 | 40% |
| product | 2 | 40% |
| great | 2 | 40% |
| experience | 1 | 20% |
Summary Narrative
The overall sentiment for the recent period is mixed, with 40 % of posts positive, 40 % neutral, and 20 % negative. The most frequently discussed topics are love, new, and product, each appearing in 40 % of the posts. Positive feedback highlights product quality and performance, while negative feedback highlights service speed and battery‑life concerns.
Appendix A – FAQ
Q1. What if a post contains both positive and negative words?
A1. The post is classified as Neutral. The sentiment assignment rule treats mixed‑sentiment posts as neutral to avoid bias.
Q2. How are emojis handled?
A2. Emojis are retained during tokenisation but are not counted as topics. They may be mentioned in the summary if they appear frequently and are relevant.
Q3. What if a post is in a language other than English?
A3. The post is flagged for manual review and excluded from the automated analysis.
Q4. How many topics are listed in the final report?
A4. Up to five topics are shown. If there is a tie for the fifth spot, any of the tied topics may be selected.
Q5. What if the Engagement field is missing?
A5. The process continues without the engagement count; it is optional and not required for the analysis.
Q6. How are percentages rounded?
A6. Percentages are rounded to the nearest whole number. Minor rounding differences (e.g., total of 99 % or 101 % due to rounding) are acceptable.
Q7. Can the process handle more than 1,000 posts?
A7. Only the first 1,000 posts are processed; the remainder are ignored and noted in the final summary.
Q8. What if no posts have any of the listed positive or negative words?
A8. All posts are classified as Neutral.
Q9. How are “topics” defined?
A9. Topics are single‑word keywords derived from the posts after removing common stop‑words and any word shorter than three characters.
Q10. How are the percentages in the “Top Topics” table calculated?
A10. The percentage for each topic is the number of posts that contain the topic at least once, divided by the total number of posts, multiplied by 100, then rounded to the nearest integer.
Appendix B – Glossary
- Sentiment – The emotional tone of a social‑media post, classified as Positive, Neutral, or Negative.
- Positive Sentiment – A post containing at least one word from the “positive‑word” list and no words from the “negative‑word” list.
- Negative Sentiment – A post containing at least one word from the “negative‑word” list and no words from the “positive‑word” list.
- Neutral Sentiment – A post that does not meet the criteria for Positive or Negative, or contains both positive and negative words.
- Topic – A single‑word keyword that appears in the text of the posts after removal of stop‑words and short words.
- Engagement – The number of likes, shares, comments, or other interaction metrics associated with a post (optional).
- Stop‑Word – Common words (e.g., “the”, “and”, “but”) that are excluded from topic‑extraction calculations.
- Percentage (% of Total / % of Posts) – Rounded integer representing the proportion of the total count (sentiment) or the proportion of posts containing a given topic.
Appendix C – Reference Materials
C.1 Sentiment Lexicon
Positive Words (example)
- love, great, excellent, awesome, amazing, fantastic, wonderful, excellent, good, fantastic, amazing, superb, brilliant, happy, pleased, satisfied, impressive, superb, excellent, delightful, marvelous, outstanding, wonderful, fabulous, awesome, fantastic, brilliant, spectacular
Negative Words (example)
- bad, terrible, awful, poor, disappointing, unhappy, unsatisfied, bad, terrible, awful, terrible, lousy, horrible, unpleasant, dissatisfied, unproductive, problematic, flawed, subpar, terrible, dreadful, lousy, horrible, frustrating, terrible, poor, weak, bad, terrible, miserable, terrible, disappointing, terrible, unpleasant, horrible
(The full list of 200+ positive and negative terms is available in the internal lexicon library; the above samples illustrate the categories.)
C.2 Stop‑Word List (Common English stop‑words)
a, an, the, and, or, but, if, while, for, to, of, in, on, at, by, with, without, from, about, as, into, through, between, among, over, under, after, before, during, above, below, beyond, within, without, each, every, any, some, many, few, several, all, most, many, several, other, another, such, this, that, these, those, it, its, is, am, are, was, were, be, been, being, have, has, had, do, does, did, will, would, can, could, should, might, must, shall, may, could, may, might, must, shall, would, should, can, could, will, would
(The full stop‑word list contains 200+ common words.)
C.3 Style Guide for Summary Narrative
- Tone – Formal, neutral, and professional.
- Length – 3 to 4 sentences; no longer than 120 words.
- Structure –
- Overall sentiment percentage.
- Key top topics (up to three).
- Notable positive or negative points (optional).
- Any data‑quality note (e.g., “2 posts were omitted due to missing content”).
Example:
Overall sentiment is mixed, with 45 % of posts positive, 35 % neutral, and 20 % negative. The most discussed topics are product, price, and service, each appearing in at least 40 % of posts. Positive sentiment emphasizes product quality, while negative comments highlight service delays. Two posts were excluded because their content fields were empty.
C.4 Supported Platforms
- Twitter
- Facebook
- Instagram
- LinkedIn
- TikTok
Any posts from other platforms are omitted from the analysis.
C.5 Limitations
- The analysis does not handle sarcas sarcasm or context‑dependent sentiment (e.g., “great” used sarcastically).
- Only English‑language text is processed; other languages are flagged for manual review.
C.6 Example Work‑Through (Step‑by‑Step)
- Collect all social‑media posts for the desired period.
- Verify each entry has a non‑empty content field.
- Standardise text to lower‑case; strip punctuation except hashtags and emojis.
- Run the sentiment‑assignment rules per post.
- Count sentiments and calculate percentages.
- Extract words, remove stop‑words, count frequency, and select top‑5 topics.
- Create the “Sentiment Distribution” table.
- Create the “Top Topics” table.
- Write a concise narrative summary.
- Output all three deliverables.
All sections above are written in plain language, avoiding technical terminology such as “JSON” or “object”. The SOP is fully self‑contained and requires only the social‑media data listed in the Inputs section.
