What Takealot marketplace review data revealed

I’ve been selling products online since 2017, when I first began selling on Takealot. In 2021, at the height of the COVID-19 pandemic, I left my software development career and moved into e-commerce full-time.

Fast-forward to 2026 and a great deal has changed. Amazon has entered the South African market, and online selling, especially through marketplaces, has become fiercely competitive.

A pattern that didn’t make sense

Part of selling successfully on a marketplace is analysing what your competitors do. I could never quite understand how one of my competitors, Forza, was able to gain traction so quickly on both new and existing products, particularly when it came to reviews.

I repeatedly watched the company launch a product and receive a couple of five-star reviews almost immediately. Reading some of them, they seemed fake. Then, while browsing the reviews, I came across this:

★★★☆☆

Colour: Black

Lorenzo · 29 August 2025

Reviewed 7 days after purchase

“It’s still too early to judge if this product is worth the premium price asked for. One of my concerns is that the Elite watch strap was advertised as having a brushed stainless steel buckle, which does not appear to be the case on the strap I received. Time will tell if the paint flakes off. It also appears that some of the reviews for this product may have been posted by ‘friends’.”

Having had a long career in technology, my curiosity got the better of me. I began to wonder whether there was any way to prove that some of the reviews were fake.

A side note: the review shown above is fake. I’ll explain, in simple terms, how I worked that out.

Looking beneath the interface

Under the hood, the system is fairly straightforward. Takealot’s review system uses an API. The raw review data is returned as JSON, a common format computer systems use to exchange information. A review looks something like this:

{
  "tsin_id": 93622929,
  "customer_id": 7098672,
  "signature": "02459d4649a3e83df75cd8b20e3c4e88dc57f0b0",
  "rating": 5,
  "uuid": "5bc5b228-9f07-4107-9567-0e55f0e1d80f",
  "text": {
    "body": "When it comes to watch straps, I always leave the reviews for after at least 2 months of use. Most of the aftermarket straps I've bought have fallen apart or broken within 2 months. I bought 2 of these Forza straps in June (1 x 22mm and 1 x 26mm) and after 4 months of use I am really impressed. The quality of the silicone is great and it is comfortable. The buckle is actually brushed stainless steel (like the original) so it doesn't chip or scratch, and the clasps that attach to the watch are solid. I would argue that this strap is just as good, if not better than the original Garmin straps. Well done Forza—great product.",
    "title": "Long term review - It's proper!"
  },
  "num_upvotes": 9,
  "customer_name": "Matt",
  "date": "11 Sep 2023",
  "time_after_purchase": "Reviewed 3 months after purchase",
  "variant_info": ["Colour: Yellow"]
}

Look closely and you’ll see two particularly important fields: customer_id: 7098672 and customer_name: "Matt".

One thing that stood out in the reviews I suspected were fake was that the reviewer’s name kept changing, making the reviews appear to come from different people. However, to leave a review on Takealot, you first have to buy the product.

Then I found the missing piece: when a customer leaves a review, Takealot allows that customer to change the name displayed on the review.

Following the customer ID

This led me to another question: if I pulled all the reviews for my competitor’s products, could I determine whether the same customer was repeatedly reviewing one seller’s products under different names?

I’ve been writing software for 29 years, so building a way to analyse the data was straightforward. Once I had pulled it all, I ran an SQL query against the reviews associated with this seller.

The results were extraordinary. In my view, this was no longer merely a suspicion; the repeated identifiers provided hard evidence of coordinated review manipulation. Calling this anything other than blatant cheating would understate what the data shows.

The pattern is straightforward to detect

In the report I sent to Takealot, I did not merely identify the problem; I also outlined a practical way to detect it. For each seller, Takealot could examine how many of that seller’s products were reviewed by the same customer ID, how frequently that customer changed display names, whether the ratings were consistently five stars and how quickly the reviews appeared.

Takealot already holds the complete underlying data. It knows the customer ID, purchase history, seller, product, rating, review date and display names involved. Applying the pattern I identified across its own database should therefore be extremely straightforward. The system could automatically flag improbable reviewer-seller relationships for investigation and remove fake reviews or reviewers once confirmed.

Doing this is in the best interests of consumers, who rely on reviews when deciding what to buy, and honest sellers, who deserve to compete on product quality, price and service—not against manufactured reputations. Why Takealot has not applied this kind of analysis to its own data, I will never know.

Why early reviews matter so much

Because I come from a technical background, I’ve paid close attention to how products behave when they are first launched on the platform. Over the years, I’ve noticed that new listings often receive an initial boost in search results.

For example, if I launch a new camping chair, it will typically rank above many existing camping chairs—at least temporarily. That makes sense. If a new product were buried at the bottom from day one, it would never be seen and would have little chance to compete. The platform appears to give new listings a window of visibility.

What happens next is where this becomes powerful. If that camping chair quickly receives strong reviews, it can begin outranking older, established products and hold its position near the top. I may be simplifying the exact mechanics, but in principle this is how momentum builds:

Now consider what happens if a seller seeds that initial boost with fake five-star reviews. That artificial early validation amplifies the effect of the ranking bump. Reviews influence buyer psychology; there is a strong herd dynamic at play. When the first review is five stars, later customers may be predisposed to view the product positively. That initial signal shapes perception.

Fake early reviews therefore do more than inflate a rating. They distort the competitive dynamic, leverage the platform’s launch mechanics and can compound into a sustained, unfair advantage.

Seven months later

So, what do you do when you uncover something like this? I contacted Takealot with my findings. After some pressure from my side, I finally secured a meeting with Takealot in March 2026. I spoke with two technical staff members and its head of marketing and PR, Karla Levick.

After the meeting, I documented my methodology and sent Takealot a detailed report. I left with the impression that the company would act.

Fast-forward to 20 August 2026, almost seven months later and, as far as I can tell, nothing has been done about the fake reviews. The seller continues to trade. Worse still, I am now tracking the activity, and the same pattern of suspicious reviews continues to appear on its products.