Fixing Sentiment Shifts in Brand Mentions
A sentiment shift is a change in the overall tone of what people say about your business: fewer happy posts, more complaints, or a slow slide from warm to lukewarm. The Brand Mention Monitor scores the sentiment of each mention and shows how coverage changes over time, so a shift usually appears first as a line on a chart.
A chart cannot tell you why. This guide covers checking whether the change is real, dating when it started, sorting the mentions to see what people are unhappy about, matching that to something that changed in your business, and fixing the cause rather than the symptoms.
How sentiment scoring works, and where it goes wrong
Sentiment scoring uses language analysis to judge whether a piece of text is positive, negative or neutral. It is very useful across hundreds of mentions, and much less reliable on any single one. Before reading anything into a change, it helps to know the usual mistakes:
- Sarcasm and irony. "Brilliant, third late delivery this month" contains a positive word and a negative meaning.
- Negation and mixed posts. "Not bad at all" and "great food, rude staff" are hard to score as a whole.
- Sector vocabulary. Words like “killer”, “sick”, “aggressive” or “cheap” mean different things in different trades.
- Name collisions. If your brand shares a word with something else, mentions of the other thing can drag your score around.
- Subject confusion. A post that praises you while criticising a competitor, or the reverse, may be scored on the wrong target.
- Short posts. A one-word reply or an emoji gives little to work with.
None of this makes the score useless. It means that a change in the trend is a prompt to read the underlying mentions, not a conclusion on its own.
The fix, step by step
Step 1
Check the shift is not a scoring artefact
Open the mentions scored negative during the period where the trend dropped and read a sample of twenty or thirty. Mark each one as correctly scored, wrongly scored or not about you. If a large share are misread sarcasm, posts about a different business with a similar name, or one noisy source, the drop may be an artefact. Fix it by tightening your tracked terms or excluding the irrelevant source, then look at the trend again before going further.
Step 2
Separate volume from tone
Look at the number of negative mentions and the share of all mentions that are negative, side by side. If both rose, tone has genuinely worsened. If the count of negatives rose but the share stayed level, you are simply being talked about more, perhaps after a campaign or some press coverage. If positive mentions fell while negatives stayed flat, the problem may be enthusiasm fading rather than new complaints, which has a different cause.
Step 3
Date the start of the change
Switch the trend to weekly figures and find the first week where the share of negative mentions moved clearly away from its usual range and stayed there. Note that date. A sudden step usually points to a single event, such as a launch, a policy change or a news story. A slow slope over several months usually points to something gradual, such as rising prices, a stretched team or a slowly declining product.
Step 4
Tag the negative mentions by theme
Export the negative mentions from the weeks after the start date and tag each one with a short theme label, such as delivery, price, quality, staff, website or policy. Tag the source type too, such as review site, forum or social. Count the themes. Usually one or two dominate. Compare with the same count from a quieter period before the shift, so you can see which theme actually grew.
Step 5
Line the timing up with business changes
Write down everything that changed in the business in the month or so before the start date: new suppliers, price changes, staff departures, new software, a redesigned website, altered opening hours, a changed returns policy or a new delivery partner. Put them next to the dominant themes. The cause is usually obvious once the two lists sit together. If nothing internal changed, look outside: a news story, a competitor campaign or a wider problem in your sector.
Step 6
Fix the cause before the messaging
Deal with the underlying problem first, whether that means switching back to a reliable courier, retraining a team, reversing a confusing policy or fixing a broken checkout. Replying politely to every complaint while the cause continues only produces more complaints. Where the change cannot be reversed, such as a necessary price rise, reduce the friction around it with clearer explanation, notice periods or a simpler option for customers who are affected most.
Step 7
Tell affected customers what changed
Once the fix is in place, say so plainly where the complaints appeared. That can be a short post on your own channels, a note on the relevant page of your website, an email to affected customers, or a follow-up in threads where people complained. Explain what went wrong in one or two sentences, what you have done, and what to do if the problem continues. Avoid vague statements that never admit anything happened.
Step 8
Watch the recovery on a rolling average
Track the share of negative mentions using a four-week rolling average rather than daily or weekly numbers, which jump around too much for small businesses. Expect a lag, because people often post about experiences some time after they happen and old threads keep resurfacing. Check whether the dominant theme is shrinking in your tags. If it is not falling after a couple of months, the fix may not have reached the real cause.
Reading the trend without fooling yourself
Small numbers make trends noisy. A business with forty mentions a month can see its negative share double because three extra people had a bad day. A few habits help:
| What you see | What it may mean | What to check |
|---|---|---|
| One very bad day, then back to normal | A single post that got attention, or one angry person posting many times. | Handle the post; do not treat it as a trend. |
| A sharp step down that stays down | A specific event or change. | What happened in the week before the step. |
| A slow decline over months | Gradual wear: prices, service levels, product quality. | Compare themes from now with the same months last year. |
| Negative count up, share flat | More attention overall, not worse feeling. | Where the extra volume came from. |
| Positive share down, negative flat | Enthusiasm fading, fewer happy customers posting. | Whether something customers loved has changed or disappeared. |
| Change in one source only | A community or site-specific issue. | That source's threads; it may not reflect customers generally. |
Also compare with the same period last year. Many businesses have seasonal patterns, such as delivery complaints before Christmas or booking frustration in the summer, that look alarming if you only compare with the previous month.
A worked example
A fictional online homeware shop, Ashgrove & Pine, sees its negative share rise from its usual level to roughly double over six weeks in the spring.
- Artefact check. A sample of thirty negative mentions shows twenty-six correctly scored. Four were a furniture auction house with a similar name, now excluded.
- Volume versus tone. Total mentions are flat, so the negative share rise is real.
- Start date. Weekly figures show a clear step in the second week of March.
- Themes. Of negative mentions after the step, most are about delivery, with smaller numbers about returns. Before the step, delivery complaints were a small minority.
- Timeline. The shop changed to a cheaper courier at the start of March.
- Fix. The shop moves large and fragile items back to the previous courier and adds tracking emails for the rest.
- Telling customers. A short note on the delivery information page, replies in the main review threads, and an email to customers with delayed orders.
- Recovery. The four-week rolling average falls back towards its usual range over the following two months, and delivery drops back to a minor theme.
Nothing in the chart said "courier". The cause only appeared when the tagged themes were placed next to the business timeline.
When the cause is outside the business
Sometimes sentiment falls because of something you did not do. Common external causes include:
- A sector-wide story. A news report about problems in your industry can make people more critical of every business in it.
- A viral post. A single widely shared complaint, fair or not, can pull in piles of comments from people who have never used you.
- A coordinated campaign. Many similar posts from new or unrelated accounts over a short time may be organised. Record them and report them to the platforms concerned.
- Mistaken identity. Another business with a similar name has a problem and you are caught in the crossfire.
In these cases the fix is mostly about communication: a calm statement of the facts, corrections where your business has been wrongly named, and patience. Individual complaints are handled as described in responding to negative mentions.
Sudden upward shifts need checking too
A sharp rise in positive sentiment is pleasant, but check it as carefully as a fall. Read a sample of the new positive mentions. If they come from real customers after a genuine improvement or a successful launch, note what caused it so you can repeat it. If they come from accounts with no history, use similar wording or appear on one site in a short burst, they may be fake, and fake positive reviews in your name can damage you if a platform or regulator notices. The review surge guide covers what to do when reviews spike unexpectedly.
Catching the next shift earlier
- Keep a change log. A simple dated list of business changes, such as suppliers, prices, policies, staff and systems, makes step 5 take minutes instead of hours.
- Tag as you go. If whoever handles mentions adds a theme tag each week, a shift in themes shows up before the overall score moves.
- Review monthly, not daily. A regular look at rolling averages avoids both panic and complacency.
- Test changes with a small group first. A new courier or checkout tried on a fraction of orders reveals problems before every customer meets them.
- Make sure complaints are found quickly. See fixing slow response time, because a shift is much easier to reverse in its first fortnight than its third month.
1. How accurate is automatic sentiment analysis?
It is reasonably good at the overall picture across many mentions, and much less reliable for individual posts. It struggles with sarcasm, negation, mixed opinions, sector jargon, very short posts and brands whose names are ordinary words. That is why a change in the trend should always lead you to read a sample of the underlying mentions before concluding anything. If many are misread, the fix may be adjusting what you track rather than anything in your business.
2. How big a change in sentiment should worry me?
There is no universal threshold, because it depends on how many mentions you get. For a small business with a few dozen mentions a month, week-to-week swings are normal and mostly noise. Worry when the share of negative mentions stays clearly above its usual range for several weeks, or when one theme suddenly dominates complaints. Using a rolling average and comparing with the same period last year makes real changes easier to tell apart from random movement.
3. What is the difference between a sentiment shift and a single negative mention?
A single negative mention is one person's experience and is handled as a customer service task: reply, resolve and record. A sentiment shift is a change in the overall balance of what many people are saying, over weeks or months. It usually has a cause in the business or its surroundings, such as a new supplier, a price rise or a news story. Fixing it means finding and dealing with that cause, not just answering each complaint.
4. How long does it take for sentiment to recover after a fix?
Usually longer than the fix itself. People post about experiences some time after they happen, older threads keep circulating, and customers who had a bad experience may not return to say things have improved. Expect a lag of several weeks before the rolling average starts moving, and a couple of months or more for a full recovery. If the dominant complaint theme is not shrinking by then, check whether the fix actually reached the real cause.
5. Should I make a public statement about a drop in sentiment?
Not about the sentiment itself, which customers neither know nor care about. Do speak publicly about the underlying problem once you understand it and have fixed it, or at least have a plan. A short, honest explanation of what went wrong and what you have changed, posted where affected customers will see it, tends to help. A vague statement that avoids admitting any problem usually makes people more sceptical rather than less.
6. What if sentiment dropped but nothing changed in my business?
Look outside the business. Common external causes are a news story about your sector, a viral post, confusion with a similarly named company, or a coordinated campaign. Check whether competitors saw a similar drop at the same time, which points to a sector-wide cause. Also check whether a single source is responsible, since one community turning critical can move a small business's figures. The response in these cases is mostly clear communication and correcting mistakes.
7. Can a sentiment shift affect search or AI visibility?
Indirectly it can. Negative threads and reviews may appear when people search your name, and AI assistants that summarise what the web says about a business may reflect a critical tone if that is what they find. Google has not published any direct link between sentiment and rankings, so be sceptical of claims that it is a ranking factor. The practical reason to fix a shift is that it reflects a real problem affecting customers and sales.
The Brand Mention Monitor scores the sentiment of mentions across news, blogs, forums, social, podcasts and video and shows how coverage changes over time, so a change in tone is visible early enough to trace and fix.
Open Brand Mention Fixes →