Part 1 — The Tax Part 2 — The Evidence Part 3 — Suppression Files Part 4 — What We Built
Part 2 of 4

The Evidence

Every claim has a number behind it. Most of them have several.

Dr. Yamicia Connor, MD, PhD, MPH CEO, Diosa Ara · Founder, The Labora Collective
scroll ↓

Part 1 of this essay made the argument. This is the evidence.

Every claim I made — about reach decay, about pay gaps coded to skin tone and hair texture, about the invisible labor tax on solo operators, about ad fraud and platform lies, about healthcare AI reproducing the bias it was supposed to eliminate — every one of those claims has a number behind it.

I am publishing this separately because the data deserves to stand on its own. If you read Part 1 and thought "that tracks but I want the receipts," here they are. If you are a researcher, a journalist, a policymaker, or a founder building in this space, this is your reference document. Cite it. Use it. I did the work so you don't have to start from zero.

Section 1

The Reach Collapse

The promise of social media was that you could build an audience without a gatekeeper. The reality is that the platforms became the gatekeeper — and then they started charging rent.

Organic Reach by Platform (2024–2025)
PlatformOrganic ReachAvg. EngagementAlgorithm Driver
Instagram~7.6%0.48%Reels / Watch Time
Facebook~5.9%0.15%Community / Shares
X (Twitter)~3.0%0.12%High-Arousal / Viral
TikTok~2.5%3.70%Entertainment / Native
LinkedIn~20–30%1.8%Professional / B2B

Facebook's trajectory tells the whole story. In 2012, organic reach for business pages was approximately 16%. By 2024, it had fallen to 2.6%. That is an 84% decline in the ability to reach people who already chose to follow you. This is not a content quality issue. This is a platform deliberately throttling access to your own audience so you'll pay to get it back.

There are 5.66 billion social media user identities globally — 93% of all internet users. The platforms control the pipe between you and virtually every potential customer, reader, patient, or supporter you will ever reach. And they have made it clear: access is not free, and it is not fair.

Section 2

The Labor Tax

This is the data that should end the "just post more" advice permanently.

The Rule of 7: a consumer needs approximately seven touchpoints with a brand before making a purchase decision. At 10% organic reach, you need to publish 70 or more posts to generate those seven touchpoints for a single potential customer. At 5% reach, that number doubles to 140+.

88% of women-owned businesses in the United States are non-employer firms. One person. She is the CEO, the content creator, the customer service representative, the accountant, the social media manager, and the person actually delivering the product or service. She is producing 140 posts to reach one customer while also running the business those posts are supposed to support.

Customer Acquisition Cost (CAC)
ChannelAverage CAC
Paid Social (Beauty/E-commerce)$127–$274
Some Digital Channels (surge)Up to 222% increases over 5 years
Referral Programs$25–$65
Email Marketing3.8% conversion rate

CAC has increased 60% over the past five years. The bootstrapped founder — the one who cannot afford a $127-per-customer acquisition cost — is left with two options: produce an inhuman volume of organic content, or build referral and email systems that bypass the platform entirely.

Section 3

The Funding Desert

Funding Disparities — Black Women Founders
MetricBlack WomenWhite Peers
Venture capital share (2024)0.4%
Primary funding source80% self-funded
Bank loan approval rate17%~51%
Average annual revenue$24K–$59K$54K–$226K
Investor bias61% must be "further along"

Black women lead approximately 2 million businesses in the United States, employing more than 647,000 people. If these businesses achieved revenue parity with white women-owned businesses alone — not men, just white women — it would add $409 billion annually to the U.S. economy.

At current growth rates, Black women-owned businesses are 40 years away from reaching parity with Black men-owned businesses. Not white men. Black men.

Section 4

The Pay Gap by Phenotype

The influencer economy was supposed to be the great equalizer. The data tells a different story — and it gets worse the darker your skin and the coilier your hair.

Influencer Pay Gap — Median Instagram Reel Fee (UK, 2023–2024)
GroupMedian FeeGap vs. White
White creators (baseline)£1,637.62
Black creators£1,080.41−34.04%
"Deep Dark" skin tone£928.00−44.63%
Type 4B (coily) hair texture~£800.00~−53%
Southeast Asian creators£700.63−57.22%

A creator with Type 4B hair — coily, tightly curled, the hair texture most associated with Blackness — earns roughly half of what a white creator earns for the same deliverable on the same platform. This is not a historical artifact. The gap has widened since 2022.

This is colorism operationalized as a compensation model.

Section 5

Platform Lies

The platforms have been caught lying about the value of what they sell.

Ad fraud costs the global economy between $12.4 billion and $19.5 billion annually. LinkedIn's internal audits have shown a 25% invalid traffic rate. Facebook settled a $40 million lawsuit after admitting it inflated average video watch time metrics by 150% to 900% — which means the "pivot to video" that destroyed newsrooms and content strategies across the industry was based on fabricated data.

The Northeastern University study demonstrated that Facebook's ad delivery system skews by race and gender even when advertisers select inclusive targeting parameters. The system itself routes housing ads away from Black users and job ads away from women — not because the advertiser chose this, but because the algorithm optimized for "relevance" using historical engagement patterns that reflect historical discrimination.

The platforms are not neutral infrastructure. They are active participants in the extraction.

Section 6

The Shadowban Economy

"Unseen Shame"

Users who experience sudden, unexplained drops in reach internalize the loss as a personal failing. They assume their content is not good enough. They do not know the platform has throttled them.

"Digital Neurosis"

The constant uncertainty about whether you are being suppressed produces anxiety that alters content strategy. Creators begin self-censoring, avoiding topics the algorithm might penalize, even when those topics are their core expertise.

"Emotional Conformity"

Over time, creators shift their tone, subject matter, and perspective to match what the algorithm rewards. The platform doesn't have to censor you. It trains you to censor yourself.

The research documents a consistent pattern: users who post about social justice, racial inequality, or political issues experience a sharp drop in views not only on that post but on all subsequent posts. The suppression persists across content — as if the account itself has been flagged.

Section 7

The Content Moderation Machine

The suppression is not random. It is structural, and the mechanisms are documented.

Lexical over-reliance on affective cues: The systems flag emotional language as potentially toxic, regardless of context. A Black woman describing her experience with medical racism in emotional terms is processed by the same classifier that flags harassment.

Spurious lexical correlations: Race-related terms correlate with high toxicity scores regardless of context. The word "Black" in a sentence about Black maternal mortality triggers the same classifier that would flag a racial slur. The model learned racism from the data and now enforces it at scale.

The result: the people most affected by systemic harm are the ones most likely to have their descriptions of that harm flagged, throttled, or removed.

Section 8

The Health Misinformation Feedback Loop

Echo chamber lock-in: Users who engage with health-skeptical content are served more of the same. The recommendation engine does not balance. It amplifies.

The "disinterest gap": Users who show low initial interest in healthcare content see almost none. These are often the users who most need accurate health information.

Gendered targeting of alarmist content: Women under 45 are disproportionately served alarmist diet, exercise, and body image content. The algorithm has learned that anxiety about health drives engagement. It is not optimizing for health outcomes. It is optimizing for time-on-platform by exploiting health anxiety.

This is directly relevant to maternal health. When the algorithm suppresses accurate clinical guidance and amplifies fear-based content, the downstream effect is patients who are less informed, more anxious, and less likely to seek appropriate care.

Section 9

Healthcare AI Bias

The algorithmic tax is not limited to social media. It has already entered the clinical space, and the consequences are measured in lives.

Documented Healthcare AI Bias
SystemBias MechanismClinical Impact
Optum (~200M Americans)Spending as proxy for needBlack patients received 50% less care
Chest X-ray AIUnder-representative training dataLess accurate for Black/female patients
Sepsis PredictionGeographic skewLower accuracy for Hispanic/rural patients
Skin Cancer DetectionTrained on lighter skinHigher mortality from late-stage detection
Knee X-ray PainBias in "physical" indicatorsDismissed Black patient pain reports

If the Optum algorithm's racial bias were eliminated, the percentage of Black patients receiving additional care would increase from 17.7% to 46.5%. That is not a marginal adjustment. That is a near-tripling.

The global AI-in-healthcare market is projected to surpass $187 billion by 2030. Only 5% of AI developers are Black. The systems are being built by a population that does not represent the patients most affected by their failures.

Section 10

The Economic Case — Not the Moral One

The Cost of the Status Quo
MetricValue
Black women-owned businesses3.5 million
Current annual revenue~$60 billion
Revenue if parity with men+$1.7 trillion GDP
Revenue if parity with white women only+$409 billion annually
All women-owned business parity+$10.2 trillion GDP
Annual cost of racial wealth gap$1–$1.5 trillion/year
Racial wealth gap growth (2019–2022)+$49,950 per household
Black founder VC share (2024)0.4%

The algorithmic tax is not a cultural problem. It is an economic extraction system that costs the U.S. economy trillions of dollars annually and is accelerating.

The question is not whether this is unjust — it is whether it is sustainable. And by their own metrics, the answer is no.

What Would Fix This

For Founders

Build first-party data infrastructure. Email lists, referral networks, owned community platforms. The most efficient customer acquisition channel for bootstrapped businesses is referral ($25–$65 CAC) and email marketing (3.8% conversion rate), not paid social ($127–$274+ CAC). Stop subsidizing platforms that are burying you. Pay each other instead.

For Healthcare AI Developers

Integrate "small data" — social determinants of health, community-level indicators, lived experience — into clinical models. Audit for bias at the model level, not after deployment. Design for equity as a foundational principle, not a compliance checkbox.

For Policymakers

Mandate algorithmic transparency for content moderation tools. Require platforms to publish organic reach data by demographic. Treat algorithmic suppression of health information as a public health issue, because it is one.

For Capital Allocators

Non-dilutive capital — grants, revenue-based financing, community development financial institutions — is the bridge that gets Black women founders past the CAC barrier without requiring them to give up equity to investors who will require them to be "further along" than their white peers before writing the check.

"The data is here. The evidence is not ambiguous. What remains is the decision about what to do with it." — Dr. Yamicia Connor

The evidence is documented.
Now here's what happened to me.

Read Part 3: The Suppression Files →