A Company Claims '500% Growth'—From What to What?

If a Company Claims “500% Growth,” From What to What?

When I see a corporate press release boasting “500% growth year-over-year,” the first thing I do is look up last year’s number. You can only judge the size of that growth once you know the baseline figure and how it was measured. But sometimes the release gives only the growth rate and leaves out that explanation. When there’s also no external data to check against, it becomes hard to tell whether you should simply take the announcement at face value.

This kind of misrepresentation of user counts or revenue to investors has happened abroad too. The U.S. SEC sued the founder of social media startup IRL, alleging he lied about how the company acquired users, and China’s Luckin Coffee was revealed to have fabricated more than $300 million in revenue. That said, it’s worth distinguishing between whoever first raised the allegations and whichever institution ultimately imposed sanctions.

While thinking about this verification problem, I came across a company called Objection. Backed by investors including Peter Thiel and Balaji Srinivasan, the company offers a service that lets you formally dispute factual claims made in published content like articles or podcasts. Each claim costs $2,000 to challenge. It also reportedly runs an Honor Index¹ that rates the credibility of journalists and outlets based on the results of these investigations.

I wondered whether a tool like this could also verify the numbers companies announce. But Objection’s stated focus isn’t corporate PR or IR materials — it’s press coverage. That’s what made me curious about its pricing and verification method. The same service, after all, can be used not only by people trying to confirm facts but also by people trying to pressure reporting that’s unfavorable to them.

What to Verify in Press Release Numbers

There’s a problem I’ve run into often while doing GTM strategy consulting: companies publish performance metrics in PR and IR materials without specifying how those metrics were measured. I don’t think this is a good way to communicate. If the numbers can’t later be reconciled with actual financial statements or user data, it damages the company’s credibility.

When I read these materials, I specifically check for three things.

• The baseline behind growth rates: If one user becomes two, that’s 100% growth. You need to look at both the growth rate and the actual before-and-after figures.
• The definition of GMV: You have to check how gross merchandise value² accounts for cancellations, refunds, and internal transactions. If companies use different accounting standards, the raw numbers aren’t comparable.
• The measurement period and acquisition channel for user counts: If user growth came from advertising or a one-off event, you need to see whether engagement persists afterward. Cumulative sign-ups and monthly active users also need to be distinguished.

The fact that a company announces a new service launch or overseas expansion doesn’t mean prior performance no longer needs checking. Even when I look at a new announcement, I try to confirm whether it uses the same standards as previously disclosed figures. That said, the mere timing of a new announcement isn’t enough to conclude that a company was trying to bury an audit result.

In the IRL case, the way user growth was achieved became the issue. According to a civil suit the SEC brought in July 2024, the founder raised roughly $170 million by claiming the app had about 12 million users, most of whom had joined organically. But the SEC alleged that the company spent millions of dollars on ads and incentives to drive downloads, and hid part of that spending through third parties. Investors needed to know not just the user count, but the cost and method behind how that number was built.

In the Luckin Coffee case, the fabricated revenue the SEC identified exceeded $300 million between April 2019 and January 2020. The company itself disclosed the revenue manipulation in April 2020, ahead of the SEC’s enforcement announcement. Cases like this suggest the right question isn’t whether verification tools existed at all, but how early the necessary information could have been confirmed once investment and transactions were already underway.

How Objection Verifies Reporting

Objection’s founder, Aron D’Souza, was involved in pushing forward the lawsuit between Hulk Hogan and Gawker. He frames declining trust in the media as the reason the service is needed.

Indeed, Gallup polling shows trust in American media has fallen from 68-72% in the 1970s to 28% in 2025. This figure comes from asking how much people trust newspapers, TV, and radio to report the news fully, accurately, and fairly.

But low trust alone doesn’t mean AI’s verdicts are more accurate. We need to separately examine what materials the service receives and how it handles cases with genuine uncertainty.

Here’s what TechCrunch’s launch coverage and the founder’s interviews lay out:

• Submission and AI evaluation: LLMs³ from 5 providers evaluate materials submitted by both the party raising the objection and the outlet that published the story. Each model is given a different persona, but this shouldn’t be mistaken for an actual jury or independent expert review.
• Disclosure of the verification process: The company says it will publish the evidence and the reasoning behind its verdicts. Disclosure alone doesn’t guarantee accuracy or reproducibility, and we also need to look at how sensitive materials requiring protection are handled.
• Evidence grading: Primary sources, like regulatory filings, are weighted highly, while anonymous statements without independent corroboration are weighted low.
• Human investigation: The company says it assigns additional fact-gathering to former investigators or investigative journalists.

The companion feature, Fire Blanket⁴, uses the X API to post a notice that a given claim is under investigation. It is not an official label that X itself applies after determining something is false. The fact that this lets people flag stories that haven’t reached a conclusion yet deserves separate scrutiny.

It’s easy to imagine using this kind of investigation when a company’s reported revenue or user numbers are in doubt. But a $2,000 commissioning fee alone doesn’t mean the results will come faster or more accurately than existing investigations. And if the available materials are insufficient, the process may simply fail to reach a conclusion.

What worries me more is source protection.

Can You Verify a Story While Still Protecting Your Source?

According to the launch announcement, Objection rates statements from unverified, anonymous sources⁵ lower. It’s not accurate to say every story using anonymous sources automatically gets the lowest score. Still, there’s a genuinely hard problem here: how far should a reporter go in accommodating a third party’s demand for verification when the reporter’s job is to protect a source?

D’Souza described a source-verification method built on cryptographic hashing. But from a reporter’s perspective, you’d need to know exactly what information gets seen, by whom, how long it’s retained, and whether there’s any risk of the source being re-identified. Participation in the service isn’t legally required, but if refusing to participate results in a public low score or a “flagged as under review” label, a newsroom could still feel pressure to respond.

Jane Kirtley, a media law professor at the University of Minnesota, and Chris Mattei, an attorney who handles press-related litigation, raised a similar concern in a TechCrunch interview: powerful figures could use this kind of service to pressure critical reporting. That’s a critique of the design, not a finding that the service has actually been misused. Separate from First Amendment⁶ protections, the burden on reporters of having to respond to outside scoring could still remain real.

Source protection was central to John Carreyrou’s 2015 Wall Street Journal reporting on the Theranos scandal. The reporting raised questions about the company’s claims through internal testimony and document verification, and the company responded by legally pressuring both the story and its sources. If the company back then had had access to a service that let it dispute each individual claim in the article separately, the burden on the reporters to respond could have been even heavier. But this is hypothetical — we have no way of knowing what scores such claims would actually have received.

If a service lets someone dispute multiple claims within a single article one by one, it can be used both to request legitimate fact-checking and to pile more response work onto a newsroom. How the service manages this kind of repeated, granular disputing matters a great deal.

Who Would Actually Pay $2,000?

From a GTM⁷ perspective, the first thing I notice is the price tag: $2,000 per case. That’s too steep for an ordinary reader to pay every time a story bothers them, but a company with a legal or PR budget could commission multiple cases as needed.

The founder also said he wants to help people who can’t afford the fee. But a personal promise of support is different from a formal assistance program that anyone can check eligibility for and apply to. To actually assess accessibility, you need clear fee-waiver criteria and a defined process.

The users D’Souza envisioned in the interview are people who believe they’ve been misrepresented by the press. Giving these people a means of correction or rebuttal is, in itself, a legitimate need. At the same time, we have to look at how a structure where the person requesting the investigation is also the one paying for it affects how that investigation is conducted and disclosed.

What worries me is whether the distinction between a client’s claim and a final verdict comes through clearly enough to readers. In particular, the investigation notice published before a verdict only shows that someone paid to file a dispute — it’s not evidence that the reporting was wrong. Just as the company holds the press accountable, the operators of this service need to be held to an equally concrete standard of accountability.

Oswarld’s Lens

The difficulty of verifying a company’s published numbers in a timely way is something I’ve run into myself in GTM consulting work. When growth figures without clear measurement standards, or KPIs with fuzzy definitions, get used as the basis for investment and deal decisions, you can end up discovering—too late—that everyone was looking at different numbers. I do think a tool that helps with this verification process is needed.

But it’s still hard to judge how well Objection actually solves that problem. Having a set price for a request is one thing; having improved verification cost, speed, and accuracy is another. There’s also a risk that more requests will come in aimed at suppressing critical reporting, but which use case ends up more common in practice is something we’ll only know from how the service actually operates.

If I were building a service in this space, I’d start by narrowing the scope to claims that can be checked against measurable standards and source material—things like regulatory filings, financial statements, user metrics, and revenue recognition. I’d exclude investigative journalism that relies on anonymous sources from the initial target set. And I’d design the cost burden on the requester alongside the response burden on whoever is being challenged. Defining the scope this way won’t prevent every kind of abuse, but it does let you be clear, from the start, about exactly what the service can and can’t verify.

Closing

When I read a press release, there are three things I try to check.

• What baseline number was used to calculate the growth rate?
• What method was used to tally user counts or transaction volume?
• Does the definition used in this announcement match the one used in the same quarter last year?

Verification services deserve the same scrutiny. It’s not enough to look at the score — you need to know what data was checked, and what wasn’t.

I don’t think we can properly evaluate this kind of service unless it discloses how it verifies evidence while still protecting sources, what criteria limit repeat requests, and what process exists for challenging a wrong verdict.

Even a score handed down by AI needs separate grounds before we can trust the number.

References & Further Reading

Primary sources

• Rebecca Bellan, “Can AI judge journalism? A Thiel-backed startup says yes, even if it risks chilling whistleblowers”, TechCrunch, 2026.04.15. : This article covers the launch of Objection and how its founder and critics describe it.
• Rebecca Bellan, “Full transcript: Conversation with Aron D’Souza on Objection and AI in journalism”, TechCrunch, 2026.04.15. : A full interview transcript in which the founder explains how the service operates, alongside the journalist’s questions.

Background

• U.S. Securities and Exchange Commission, “SEC Charges Founder of IRL Social Media App With Defrauding Investors of $170 Million”, 2024.07.31. : The SEC’s civil complaint announcement detailing how IRL acquired users and what it told investors.
• John Carreyrou, Bad Blood: Secrets and Lies in a Silicon Valley Startup, Knopf, 2018. : A book chronicling the entire Theranos exposé, including how sources were protected and how the company responded with legal pressure. (I personally recommend the TV drama too.)
• U.S. Securities and Exchange Commission, Announcement on Luckin Coffee’s revenue fabrication, 2020.12.16. : Confirms the period and amount of revenue fabrication identified by the SEC.
• Gallup, “Trust in Media at New Low of 28% in U.S.”, 2025.10.02. : Presents survey results from September 2025 alongside media trust trends dating back to the 1970s.

The author is Oswarld (Kwangseob Ahn). Current roles: Adjunct Professor at Sejong University, Strategy Consultant at INLEVEL9. Career, research, books, and recent work are kept current on the About page. Latest · July 2026: HEMA-2: A Consolidation-Aware Tri-Memory Architecture with Multi-Channel Scheduling for Lifelong Conversational AI.

Footnotes

1. The Honor Index is a proprietary score that Objection presents as a way of evaluating journalists’ and outlets’ accuracy and reporting track record. It is not an officially recognized standard for evaluating journalism. ↩

2. GMV (Gross Merchandise Volume) refers to the total transaction value over a given period. It differs from the revenue a platform actually earns, and how cancellations, returns, and similar items are treated depends on the company’s own explanation. ↩

3. LLM (Large Language Model) refers to a large-scale language model. Objection states that it uses models from OpenAI, Anthropic, xAI, Mistral, and Google. Using different models does not automatically cancel out errors. ↩

4. Fire Blanket is a feature through which Objection posts notices on X about claims it is investigating. It should be distinguished from X’s official rulings or from Community Notes. ↩

5. An anonymous source is an informant whose identity is not disclosed in a news report. Not disclosing a source’s identity to readers is different from a reporter or editor failing to verify that person’s identity or statements. ↩

6. The First Amendment protects freedom of speech and the press, among other things, from infringement by the U.S. government. The standard for defamation lawsuits varies by case. The New York Times v. Sullivan ruling required that, for reporting on a public official’s conduct in office, plaintiffs prove the statement was made with knowledge of its falsity or with reckless disregard for whether it was true. ↩

7. GTM (Go-To-Market) is a strategy that determines who a product or service will be sold to, at what price, and through what channels. In this piece, the term is used to examine which customers the service’s pricing and usage process are designed for. ↩