26 May 2026 · 8 min read

Fake Reviews Network Under the Hood

RS
ReviewStatLab Team
Service analytics & review research

As a marketer or PR manager, you’ve probably dealt with fake or negative reviews at some point. Sometimes it’s a customer fishing for a discount or a full refund — annoying, but manageable.

The real problem is competitors with fake positive reviews. Fighting shadows is exhausting — especially when Google does nothing and ignores every spam report you send.

That’s exactly why we built ReviewStatLab. We’re ex-SEO specialists and marketers who’ve been burned by this problem more than once — so we decided to build the tool we wished existed.

ReviewStatLab gives you a deeper look into Google Business Profile reviews, helps you build a case for removing fake reviews, and — if you’re not ready to fight directly — at least lets you understand exactly what you’re up against.

But before any tool can help you, you need to know what to look for. Understanding how bot farms actually operate gives you a serious edge.

How Bot Farms Work — and How to Spot Them

There are two types of fake reviewers:

  1. Real people who use their personal Google accounts and sell reviews on platforms like PeoplePerHour.
  2. Fake accounts created purely for review farming.

The first type is more expensive to buy and hard to catch. These accounts look legitimate — unique names, real profile photos, local expert status, dozens or hundreds of reviews. They’re often set to private, which makes it nearly impossible to trace which businesses are using them.

The second type — the cheap option — is much easier to spot. Because these accounts need to be created at scale and on a budget, they follow patterns. You’ll see clusters of similar names, reused profile images, characters from the same movie or franchise. Naming variations like Kirk, k1rk, k!rk are a classic tell.

One more thing worth knowing: these accounts get no maintenance — no one is actively managing them. Google will eventually detect and delete inactive accounts, taking all their reviews with them. That’s why bot farms don’t waste an account on a single review — typically each account leaves 2 to 5 reviews across different businesses, squeezing out as much value as possible while the account is still alive.

That’s the weak spot.

What Data You Need for Research

First, let’s be clear: everything here operates on probability. You’ll never know with absolute certainty that an account is fake — unless it’s yours.

That said, there’s a lot of publicly accessible data you can work with. Using ReviewStatLab or any similar Google Maps data extraction tool, you can collect:

  • Creation timestamp exact date & time (not the vague “1 month ago” from the UI)
  • Edit timestamp exact date & time of last edit
  • Star rating 1–5 stars per review
  • Review text full content of the review
  • Author name display name on Google profile
  • Author profile link direct URL to the reviewer’s account
  • Author profile image avatar for visual pattern matching
  • Company URL Google Maps business page

Not all of these signals carry equal weight — but each one can contribute to your analysis, and several will be useful as evidence when filing a Google spam report.

Our network analysis feature, which focuses specifically on identifying farm weak spots, is available on the Agency plan.

The weak spot

As we said — the fundamental weakness of every bot farm is maintenance. Profiles have to stay active or they get deleted, taking their reviews with them. Eventually, every business buying fake reviews runs into this wall. But until they do, those reviews are out there working against you.

So how do you make Google move faster? By detecting the network yourself.

Here’s the approach: take every account that has reviewed a target business, then look at which other businesses those same accounts have reviewed.

On average, a cluster of 1,000 accounts will have reviewed up to 10 of the same companies — and that cluster can account for up to 5% of any single company’s total review profile.
Network graph showing reviewer accounts and businesses they have in common, excluding the target company. Grey nodes are reviewer accounts, red nodes are other businesses. Node size reflects the number of shared reviewers.
Grey nodes — reviewer accounts, red nodes — other businesses these accounts also reviewed (the target company is excluded). Node size reflects the number of shared reviewers. Larger nodes mean more overlap — not necessarily fake, but worth investigating first.

There are exceptions — very small businesses or tourist-heavy locations can look a little different. However, the key takeaway is that these networks are not strong or deeply interconnected.

This network map becomes your baseline. From here, you move to the next layer: rating patterns and timing.

Time Is the Real Signal

Ratings and timing are where fake review networks expose themselves most clearly.

Two businesses can have a completely legitimate relationship — referral partnerships, shared marketing, cross-recommendations. Hotels and car rental companies do this all the time.

But when real customers leave reviews across both businesses, the timing is unpredictable and the ratings vary widely. Natural behavior has entropy — a wide spread, no consistent pattern.

In contrast, fake reviews tend to follow consistent patterns in both rating distribution and timing. This is what we aim to detect.

When someone buys a few hundred reviews, the provider doesn’t post them all at once — they spread them out over time. The same accounts then get reused for another business, following the same posting schedule.

Forensic time-shift analysis of 25 shared reviewer accounts. 23 accounts form a tight synchronized cluster with an average pattern interval of 8.5 minutes, while 2 outliers are excluded due to a gap exceeding 7 days.
Left: Posting timeline for 25 shared accounts. While 2 outliers were excluded, 23 accounts show near-perfect synchronization over 10 months. Right: Time gap distribution across the cluster, revealing a highly anomalous average pattern interval of 8.5 minutes within a narrow 32-minute range.
The Forensic Verdict: These companies share 25 common reviewers. However, the real “smoking gun” is the tight cluster of 23 accounts that follow a rigid automated schedule. Across nearly a year, these reviews were posted with an average pattern interval of just 8.5 minutes. In a natural environment, review timing is chaotic; here, it is engineered. This is the unmistakable fingerprint of a bot farm’s scheduling script.

The idea of detecting fake review networks through behavioral pattern analysis isn’t new — there’s solid academic work on this if you want to dig into the statistical side of it [1]. Our own research is currently in progress — once we have enough data to publish, we will.

Seeing the Full Picture

One challenge you’ll hit is account privacy. A significant portion of Google accounts are locked down and won’t be accessible for research. That doesn’t mean every hidden account is a bot — but it does mean you need to focus your energy on the business profiles themselves, not just individual accounts.

If our tool flags suspicious activity, the next step is building a list of related businesses and manually cross-referencing authors across them. This is where hidden relationships surface — and it’s what makes your spam complaint actually stick.

This is the single most impactful type of analysis you can do with publicly available data. In some cases, it’s enough on its own to prompt Google to take a closer look, temporarily restrict a profile, remove reviews, or even take down the entire Google Business Profile.

Additional Signals Worth Checking

This kind of research takes time — especially when you’re dealing with profiles that have thousands of reviews. While the network analysis is your primary weapon, there are a few secondary signals that can help you prioritize where to dig.

Review Volume Timeline

Interactive timeline chart analyzing review volume. It displays review bars with a moving average line (purple), and Z-score statistical anomalies (red dots) above the chart. The visualization highlights specific periods of statistically unusual review surges for a Google Business Profile.
An interactive timeline of review posting for a business profile. The **Purple line** shows the Moving Average baseline. The **Red dots** represent periods with high ‘Z-Score’ anomalies—statistical surges that deviate significantly from normal behavior. This view is essential for quickly identifying and prioritizing specific periods for coordinated review farm activity research.

A basic timeline of when reviews were posted can reveal suspicious surges. That said, this is a weaker signal on its own — professional reputation management campaigns tend to be slow and consistent, spread over 6–12 months rather than sudden spikes. Still useful as a starting filter.

Distribution by Day of Week and Hour

Side-by-side stacked bar charts showing review distribution. The left chart shows volume by day of week with a significant spike on Sunday. The right chart shows hourly distribution from 0:00 to 23:00, color-coded by star rating (Green for 5 stars).
Left: Review volume by day of the week. Notice the heavy skew toward Sunday—a classic sign of scheduled batch posting. Right: Hourly distribution across the day. By cross-referencing these spikes with rating data (Green = 5 stars), we can identify artificial bursts that don’t align with organic customer traffic.

You won’t find location or IP data in Google Maps — but you can get exact timestamps. Look for reviews clustering in the 11 PM–5 AM window or heavily skewed toward Saturday and Sunday. These patterns are worth flagging for closer inspection.

Once you’ve identified a suspicious time-based cluster, look at naming patterns and profile images within that group. Neither signal is conclusive on its own — but combined with high posting velocity in a narrow time window, they become meaningful.

Isolated analysis leads to mistakes

Here’s a mistake that’s easy to make: drawing conclusions from a single business profile.

Businesses in the same location and niche tend to attract similar reviewer behavior — similar review volumes, similar timing patterns, similar account profiles. That’s just how shared markets work. Your competitor’s customers and yours overlap significantly.

This means that what looks “suspicious” for one business might actually be completely normal for the niche. You’ll only know the difference if you compare.

Remember
  • Always analyze at least 5–10 competitors before drawing conclusions — what’s unusual will stand out once you have a baseline
  • This comparison only works within the same geographic area — the same niche in a different city behaves differently; don’t mix regions