We found suspicious reviews during a self-audit. Should we tell the platform ourselves?

It is usually worth doing, but first work out which direction the anomaly runs. If unexplained positive reviews have appeared, reporting them protects you: it leaves a record that you noticed and raised it, which is far easier to explain later than having done nothing when the platform eventually recalculates. Give a specific list — which entries, posted when, and why they look wrong — rather than a general complaint. If it is a wave of negative reviews, report those too, but frame it around the rule being breached. Be prepared for the platform not to tell you the outcome, or to explain its decision entry by entry. So alongside reporting, keep your own screenshots and timestamps and keep genuine reviews accumulating.

A word-of-mouth agency says it can guarantee a set number of positive reviews. Is that a problem?

A guaranteed count is a red flag in itself. Whether a real customer chooses to leave a review cannot be guaranteed, so anyone able to promise it is almost certainly working through controlled accounts rather than actual customers. That practice breaches the rules of every major platform, and the penalty lands on the brand, not the vendor. A more legitimate engagement helps you design the invitation flow, improve your replies, and run content and experience marketing, with performance measured by the volume and quality of genuine customer feedback. Before signing, insist on an explanation of where the accounts come from and how the relationship will be disclosed.

If a competitor is flooding us with negative reviews, can we ask the platform to remove them?

You can report them, but removal depends on whether the content breaches platform rules, not on whether you believe it to be untrue. Content unrelated to an actual transaction, personal attacks, repeated postings, and material clearly originating from the same cluster of accounts all stand a better chance of being removed. A simple negative opinion, however unfair it feels, usually will not be touched. The practical approach runs three things in parallel: preserve screenshots and timestamps for each one, report them against the specific rule they breach, and accelerate the accumulation of genuine reviews to dilute the effect. Reporting alone, and then waiting, is the slowest route available.

How to Spot Fake Reviews and Audit Your Own

There is a contradiction built into reviews: they work because people believe they are real, and because they work, somebody is always trying to fake them.

For a brand, fake reviews are not a distant news story. They can appear on your own product page from a source you cannot identify, sit inside a performance report handed over by a partner you assumed was doing real work, or be thrown at you by a competitor. The three situations call for completely different responses, so the first step is working out which one you are actually dealing with.

Fake reviews and fake word of mouth usually come from three places

Bought by you. A brand or seller pays a vendor for reviews, which are posted from controlled accounts. This is the most dangerous variety, because responsibility sits with the brand — when a platform acts, it acts on your product page or your business profile.

Slipped in by a vendor. A brand commissions a KOC campaign, the contract says genuine experience-based sharing, and the execution team pads the numbers with duplicate accounts to hit a target. Brands often find out only when something goes wrong, and externally it is still the brand that answers for it.

Thrown by a competitor. A wave of negative reviews, or a deliberate attempt to steer forum sentiment. Here the brand is the victim, but it is also the most time-consuming case to deal with, because proving anything is difficult.

What all three have in common is that the brand carries the consequences. Which makes the ability to recognize this stuff less of an optional skill than a form of self-defense.

Self-audit: have your reviews been padded?

Do not assume nobody would bother targeting you. Some of it comes from competitors, some from a marketing team acting on its own initiative, and some is spam accounts scattering reviews at random.

What to look at:

Distribution over time. Lay the last six months of reviews out by date. A normal business shows a pattern that follows its trading rhythm; an abnormal one spikes over a couple of days and then falls to nothing
Distribution of ratings. Many businesses show a tendency to cluster at both ends, but the shape varies with category, price point, and service model, so there is no single profile to match against. The more reliable observation is that the middle should not vanish entirely: only five-star and one-star entries, with no threes or fours at all, rarely comes from natural accumulation
Account characteristics. A single review ever left, no profile picture, a name made of random characters, a history spanning businesses in unrelated parts of the country
Textual characteristics. Similar lengths, no concrete detail of any kind — no mention of what they ordered, when they came, who served them, what the situation was — only abstract adjectives

A common mistake is drawing conclusions from one suspicious entry. A single signal may be coincidence; you want all three categories present before you are confident. For a systematic way to watch for this, see The Complete Guide to Brand Monitoring.

The signals: accounts, timing, and wording, all three

If the method has to be compressed into one line: real reviews carry the noise of real life, and fake ones are too clean.

Actual customers write things like “waited nearly an hour but the food really was good” or “parking is hard to find, take the metro instead.” None of that is useful to the brand; all of it is a by-product of a real experience. Fabricated content, written to avoid mistakes, sticks to safe abstract praise.

Other practical angles:

Does the content match how you actually operate? A review mentioning an item you do not sell, or a service during hours you are not open, is an obvious break
Consistency across platforms. The same usernames or near-identical wording appearing on several platforms at once usually indicates a single source
Reply behavior. Real customers sometimes add to or amend a review after you respond; fake accounts post once and vanish
Photographs. A review with a photograph taken on site, even a poor one, is noticeably more credible. Text-only reviews, or ones using an image that is obviously stock photography, deserve a closer look

Spot-checking the delivery: account history, posting times, original assets

How to choose a vendor, what to ask before signing, and which sales lines are red flags are all covered in How to Choose a Word-of-Mouth Marketing Agency. This section deals only with what comes after the contract is signed: how you verify for yourself that what was delivered is real.

Sample the account history. Pick a few accounts at random from the delivery list and open them. How long does the posting history run, is every post sponsored, and do the categories stray into areas unrelated to each other? A genuine long-running account carries traces of an actual life; an account that only becomes active during the campaign looks rented.

Lay the posting times out. Sort the delivery list by date and hour. Real sharing scatters itself across people’s own routines, so content that goes live on the same day — or within the same hour — deserves a question, however natural the writing reads.

Ask for original assets, not a list of links. Request the source photographs and video files rather than a row of published URLs. Someone who genuinely used the product has outtakes, alternative angles, and a few shots that did not work; a single flawless product image with nothing behind it needs explaining.

Confirm the disclosure on every item. Taiwan requires endorsements to be disclosed, so wherever consideration is involved, the content has to be visibly identifiable as a collaboration. Checking each item at handover is far easier than fixing it afterward. For where the lines sit, see Word-of-Mouth Marketing Compliance in Taiwan.

A common mistake is reading the numbers in the report and never the content. A spreadsheet of reach and post counts tells you nothing about whether the accounts are real, whereas opening three to five of the actual posts usually tells you within ten minutes.

How platforms detect it, and what happens when they do

Every major platform runs its own detection. The mechanics are not published, but the operating principle is broadly understandable: they compare behavioral patterns, not the wording of any individual review.

Patterns that attract attention include large volumes of reviews concentrated into a short window, multiple accounts originating from the same device or network, accounts that post immediately after creation and then go dormant for good, and unnatural overlaps between the businesses a set of accounts has reviewed.

The consequences are rarely as simple as deleting the offending entries:

Suspicious reviews are removed silently, and your total count suddenly drops
Your overall rating is recalculated, taking genuine positive reviews accumulated over a long period down with it
Your business profile or product page has its visibility reduced, or features restricted
On e-commerce platforms, your store weighting or your eligibility to join campaigns may be affected
Some platforms have shown notices on listings with abnormal review activity, though whether such a label exists, and what it looks like, varies by platform and by period

The point most worth understanding is that these penalties often arrive late. Everything looks fine during the campaign, and the reckoning comes months later — by which time the vendor has closed the project and moved on.

What to do when a competitor floods you with negative reviews

This is the situation brands actually encounter and find hardest to prove.

What to do:

Gather evidence first. Screenshot each one and record the account name, posting time, and content. Clusters in time matter especially
Report against the rules. The point is to explain which rule it breaches — unrelated to an actual transaction, a personal attack, a repeated posting — not to explain that it is untrue. Platforms review against their rules, not against the facts
Reply as you normally would. Even where you suspect the review is fake, keep the public reply professional, because the audience is the potential customer reading later. Do not accuse a competitor; you will rarely be able to prove it
Accelerate genuine reviews. Dilution is the most dependable remedy available. For timing and wording, see How to Ask Customers for Reviews

For Google’s reporting grounds and reply principles specifically, see Can Google Reviews Be Removed?.

How consumers see through it: your own risk, in reverse

Understanding how consumers evaluate reviews tells you how your own review section looks to them. Most experienced readers do several things: read the three- and four-star reviews first because they are the most specific, sort by newest rather than most relevant, click through to a reviewer’s profile to see what else they have reviewed, and compare across platforms for consistency.

Put another way, even a fake review the platform never catches is easy for a person to see through. And once a consumer decides something looks off about your reviews, your genuine positive ones lose their persuasive force alongside the fake ones. That cost is far higher than whatever the reviews were bought for. For how Taiwanese consumers verify a brand, see How Taiwanese Consumers Search for Reviews.

Three common mistakes

Assuming a small test is harmless. Detection looks at patterns rather than volume, so a small but clearly patterned operation is caught just the same, at close to the same risk as doing it at scale.

Treating outsourcing as a transfer of liability. When a vendor uses duplicate accounts, it is your business profile that gets penalized. A contract clause prohibiting non-genuine accounts and making the vendor liable for breaches is the minimum protection worth having.

Noticing something odd and doing nothing. Seeing unexplained five-star reviews on your page and treating it as a windfall. In reality those reviews may be collateral damage from spam accounts, or a competitor planting them in order to report you afterward. Recording unexplained reviews and flagging them to the platform is safer than pretending you never saw them.

The value of reviews comes from credibility, and credibility is an asset the whole industry shares. Short-term gains from faking it are paid for with the long-term failure of the most important persuasion tool you have.

Not sure whether that batch of reviews is genuine, or whether what your vendor delivered would survive a close look? Put the questionable entries and the engagement terms into one document and book time with a NETVANA consultant — we will work through them with you and decide whether to report to the platform, send the work back to the vendor, or redirect the effort into building genuine reviews.

Further reading: for the questions to ask and the red flags to watch when selecting a vendor, see How to Choose a Word-of-Mouth Marketing Agency. For disclosure duties and compliance boundaries, see Word-of-Mouth Marketing Compliance in Taiwan. For reporting grounds and reply approaches, see Can Google Reviews Be Removed?. For diluting noise with genuine reviews, see How to Ask Customers for Reviews. And for the tools and process behind ongoing monitoring, see The Complete Guide to Brand Monitoring.