Social Media Listening SOP

A social media listening SOP shows you what strangers say about a problem when nobody asked them, and a useful round takes three to four weeks at 20 minutes a day. Every day people post "Is there an app that…", "I'm so done with…" and "Any alternative to…" on Reddit, X, LinkedIn and in YouTube comments. Those posts are free, unprompted evidence. Collect them on purpose, count them, and you learn whether your problem is a shout or a whisper before you spend a cent on ads.

What is the social media listening SOP and when should you use it?

Social media listening means tracking public posts that match a set of keywords and logging the ones where people describe a problem, a workaround or a wish to buy. It is broad and passive: you read across many platforms and do not join the conversation.

Use this SOP:

• Before interviews, to collect the words customers use and find people worth contacting.
• When you suspect a problem is growing, and want weekly counts as proof.
• When competitors exist, to log what their customers complain about in public.

If you want depth inside a few groups, with participation and relationships, use the community research SOP. If you want search-engine demand instead of social posts, use problem search volume analysis.

What do you need before you start?

• A problem hypothesis and a list of 20 words or phrases people use for it, including slang and misspellings.
• A list of 3 to 5 competitors or workarounds by name.
• Free alert tools: Google Alerts for the web and F5Bot for Reddit and Hacker News keyword alerts by email.
• A paid tool if budget allows, such as Brand24 or Mention, which cover more platforms in one dashboard.
• A logging sheet with the columns from the template below.

How do you run it, step by step?

1. Build your keyword sets

Write three sets: problem phrases ("chasing invoices", "late payments freelance"), intent phrases ("any tool for", "alternative to", "recommend a") and competitor names. Combine them with Boolean logic where the tool allows it, such as ("late payment" OR "unpaid invoice") AND (freelance OR freelancer).

2. Pick four or five platforms where your customer posts

B2B buyers post on LinkedIn and in niche subreddits. Parents post in Facebook groups and on Instagram. Developers post on Hacker News and GitHub issues. Check the reviews on Trustpilot, G2 or app stores too, because reviews are complaints with a timestamp.

3. Set up alerts and saved searches

Create alerts for every phrase in F5Bot and Google Alerts, and save searches on X and Reddit sorted by "new". Paid tools let you save the full Boolean query once. Send everything to one inbox folder so the daily check takes 20 minutes.

4. Log only relevant posts

Every day, open the new results and log a post only when a real person describes the problem, a workaround or a request to buy. Skip brand posts, news and bots. Copy the exact words, the link, the date and the platform.

5. Tag each post

Tag each logged post with one of four labels: complaint, workaround, buying request or competitor complaint. Add a 1 to 3 intensity score: 1 for mild annoyance, 3 for anger, money lost or a deadline missed.

6. Review weekly and contact the strongest posters

Every Friday, count posts per tag and read the 3-intensity ones aloud. Reply to or message 3 to 5 of the strongest posters with a short, honest note: you are researching this problem and would value a 20-minute chat. These people jump the queue for your customer interviews.

7. Write the four-week summary

After four weeks, write one page: weekly counts per tag, the top 10 phrases in customers' words, the competitors people complain about most and your decision.

Template you can copy

LISTENING LOG (Google Sheets)

Date | Platform | Link | Author type (person/brand/bot) |
Exact words (copy) | Tag (complaint / workaround / buying request /
competitor complaint) | Competitor named | Intensity 1-3 | Contacted Y/N

KEYWORD SETS
Problem phrases:   "[phrase 1]" OR "[phrase 2]" OR "[phrase 3]"
Intent phrases:    "any tool for" OR "alternative to" OR "recommend a"
                   OR "is there an app" OR "how do you handle"
Competitors:       "[Competitor A]" OR "[Competitor B]"

WEEKLY SUMMARY
Week | Complaints | Workarounds | Buying requests | Competitor complaints |
Top 3 phrases | Posts at intensity 3 | People contacted | Replies

Worked example: a meal-planning tool for families with allergies

This is an illustrative case with invented numbers. A founder wants to build meal planning for families where one child has a food allergy. She tracks 14 phrases across Reddit, Instagram comments, Facebook group search and app store reviews for four weeks.

She logs 92 relevant posts: 61 complaints, 18 workarounds and 13 buying requests such as "Is there an app that filters recipes for nut AND egg allergies?". That is 23 relevant posts a week, above her 15 line. Competitor complaints center on one thing: recipe apps let you exclude one allergen, never two or three. Eleven of the 61 complaints mention that exact gap.

She messages 12 posters and books 7 interviews. The phrase "two-allergen dinner" goes straight into her landing page headline.

What should you measure, and when do you stop?

Stop after four weeks, or earlier if two full weeks bring under 5 relevant posts each. Silence after two weeks of good keywords means either the problem is rare or your customers do not talk in public. Switch to community research or interviews to find out which.

Which mistakes should you avoid?

• Counting every mention. Brand posts, news and bots inflate the numbers. Log people only.
• Tracking one platform. Reddit alone over-represents some groups and misses others.
• Arguing in comments. Listening means reading. Pitching your idea in threads gets you banned and teaches you nothing.
• Using sentiment scores as proof. Automated sentiment labels sarcasm as praise. Read the posts yourself.
• Forgetting dates. Without dates you cannot tell a growing problem from an old one.

How can AI speed this up?

Paste a week of logged posts into an AI assistant and ask it to group them by problem, count each group and pull the most repeated phrases. Ask it to suggest new keyword variants based on the language it sees. Read a sample of 20 posts behind each group before you trust it, because AI tools miss sarcasm and merge different complaints. The F/MS social media module covers how founders use these platforms once you start posting yourself.

Where does this SOP fit in your validation plan?

Listening runs alongside interviews in Customer Discovery and keeps going in the background later.

1. Use the logged phrases in your customer interview questions.
2. Go deeper in two or three groups with community research.
3. Compare complaint counts with search demand using problem trend analysis.
4. Feed competitor complaints into competitive customer analysis.

See all 15 Customer Discovery SOPs. For a comparison of listening and research tools by price and features, the F/MS startup tools reviews list options for small budgets.

FAQ

What is the best free social listening tool for startups?

F5Bot for Reddit and Hacker News, Google Alerts for the web, and the native search with "latest" sorting on X, LinkedIn and Reddit. That free set covers most early-stage needs.

How long should I run social media listening?

Three to four weeks for a validation decision. After that, keep a lighter version running weekly as an early signal for new problems and competitor moves.

Is it OK to contact people who posted publicly?

Yes, with a short, honest message that says who you are and asks for a chat. Never pitch in the first message, and respect the platform's and the group's rules.

Can social listening replace a survey?

No. Listening shows you loud problems and exact words. A survey counts how common the problem is in a defined group, including the quiet people who never post.