Social media trend analysis: spotting market trends

Short answer

To detect a market trend from social media, check whether a topic passes four tests: it appears for several weeks in a row (duration), it is written about in different sources (diversity), the number of different people writing about it grows (spread), and it is not a seasonal or one-off spike. A topic that passes all four goes into the trend journal as "confirmed" and is passed to leadership. A one-week spike is not a trend but noise — however high it is.

Trend vs noise

Every day a new topic is presented as a "trend": a viral video, a famous blogger's post, a news story. Most are forgotten a week later. A market trend is a long-term change in people's behaviour or expectations: the shift to online payment, the habit of buying in instalments, demand for eco-friendly packaging, home delivery becoming the norm. Monitoring's job is to tell the two apart, because reacting to noise wastes money and time, while seeing a real trend late costs market share.

Four tests

1. 1DurationThe topic appears for at least 4–6 weeks in a row. A one- or two-week spike is an event, not a trend.
2. 2Source diversityThe topic is written about in at least three different kinds of source: a forum, news comments, a blog, a review site. A topic confined to one community is a niche interest.
3. 3Author growthThe number of different people writing about it grows week on week. Authors, not posts.
4. 4Not seasonal or one-offLook at the same period last year: if it rises at this time every year, it is seasonal. If it started and ended with one news story or campaign, it is an event.

No baseline, no trend

To say "this topic is growing" you need to know its previous level. So a trend search starts by looking at at least three months of history for your category terms. For each main topic, the weekly author count is recorded and each new week is compared with that baseline. Without a baseline, every new post looks like "growth", because nobody looked at earlier weeks. To see seasonality, ideally you need a year of history.

Early and late signals

• Early signal — a new word or a new question in specialist forums and enthusiasts' posts. Still small, but it shows the direction.
• Middle signal — the same topic moves to general forums and news comments; ordinary buyers start asking.
• Late signal — the topic appears in news stories and competitors' ads. By now the trend is known to everyone.

Monitoring's greatest value is catching the move from early to middle signal: the risk is still small and there is still time.

The trend journal

• Topic — a short name: "instalments including installation".
• Date and source where it was first seen.
• Evidence — weekly author count, source types, two or three linked quotes.
• Results of the four tests — which passed, which did not.
• Status — watching, emerging, confirmed, faded.
• Implication — what it means for us and who should look at it.

Illustrative example

This is an illustrative example. A home appliance retailer tracks questions about "energy-saving air conditioners" on renovation and home forums. Three months ago 2–3 authors a week wrote about it; now 9–12 do. The topic has moved from forums to news comments and a blog, and there was no such rise in the same period last year. All four tests pass, and the topic is marked "confirmed". The retailer states the energy class and monthly consumption openly on its product pages, and the sales team prepares for the question.

A proportionate response

Even when a trend is confirmed, the response should be proportionate. For "emerging" status, small steps are enough: an FAQ entry, a blog post, a short note for the sales team. "Confirmed" status can be the basis for a product or range decision, but only together with sales and search data. Social discussion alone is not grounds for a large investment decision.

A quarterly review

The trend journal is reviewed with leadership once a quarter: which topics were confirmed, which faded, what responses were made and with what result. "Faded" is valuable too — it shows which kinds of topic the team over-reacts to. After a year, the journal becomes the clearest history of how your market has changed.

Comparing with your own calls

Does a trend visible in public discussion also show up in your calls and chats? If it does, the trend has already reached your customers. If it does not, either it is still early or your audience is different. How to see demand trends in your own conversations is covered in customer demand trends.

Limitations

Public sources show part of the market: forum posters are a more active and often younger audience. In a small market weekly numbers are small and can move randomly, which is why the duration test matters. When the source list changes — a new site is added — the numbers rise artificially; note this in the journal. Monitoring does not forecast trends; it only shows the current conversation.

Common mistakes

• Calling a one-week viral topic a trend.
• Claiming "growth" without a baseline.
• Confusing post counts with author counts.
• Treating a seasonal rise as a trend.
• Making a big investment decision on social discussion alone.

Tracking trends with Vexvon Monitoring

In Vexvon Monitoring you can run a search over any date range: results are selected by publication date, and the same query can be repeated for different weeks or for the same period last year. Every result gets topic tags, repeats are reduced to one result, and each source records whether the search was complete — which helps you tell "it fell" from "it was not fully checked this time". More on the Vexvon Monitoring page.

Next step

Pick five topics for your category, record the weekly author count for the last three months, and run the four tests on each. Write the result in the trend journal and look again after a quarter. Grouping complaints is covered in grouping customer pain points; other topics are in the competitor and market monitoring section. To look at it together, get in touch.