Why We Run Credit Score Experiments

Credit scoring advice online is roughly 60% recycled myths, 30% outdated information, and maybe 10% anything backed by real data. We got tired of reading the same vague tips — "keep your utilization low" and "don't close old cards" — without anyone actually measuring the impact.

So we started testing. Systematically. With real credit profiles, real actions, and real before-and-after score measurements.

According to the Consumer Financial Protection Bureau (CFPB), approximately 26% of Americans have at least one error on their credit reports that could affect their scores. That stat alone tells you the system is messier than most guides admit.

The credit scoring industry generates over $14 billion annually (IBISWorld, 2025), yet consumers still rely on guesswork when making decisions that affect their financial lives. We think you deserve better than that.

A 2025 CardRates.com survey found that 72% of Americans want to improve their credit score in 2026, yet 49% haven't set a specific target score — illustrating the gap between intention and actionable knowledge.

Our experiments exist to close that gap. Instead of telling you what "should" happen, we show you what actually happened — with numbers, timelines, and recovery curves.

Our Methodology

Every experiment in this lab follows the same protocol:

1. Baseline measurement: We record FICO 8, FICO 9, and VantageScore 3.0 scores before any action. We also document the full credit profile — number of accounts, average age, total utilization, payment history.
2. Single-variable change: We change exactly one thing. Close a card. Pay down a balance. Add an authorized user. Nothing else changes during the observation period.
3. Multi-point tracking: We measure scores at day 0, day 30, day 60, day 90, and day 180. Credit scores don't move instantly, and the real story is often in the recovery curve.
4. Multiple profiles: Where possible, we test across different score ranges (poor, fair, good, excellent) because the same action can produce wildly different results depending on your starting point.
5. Transparent reporting: We show the actual numbers. Not "your score might drop a little" — we show "Score dropped 38 points on day 30 and recovered to -12 by day 90."

We use credit monitoring services that provide all three bureau scores (Experian, Equifax, TransUnion) plus FICO and VantageScore variants. All data is self-reported from our team and volunteer participants who consented to share anonymized results.

Why Single-Variable Testing Matters

Most credit score "studies" you see online conflate multiple variables. Someone closes a card, applies for a new one, and pays down a balance all in the same month — then attributes the score change to one action. That's not science. That's guessing. Our single-variable approach isolates each action's true impact, even if it means experiments take 6+ months to complete.

For deeper context on how scores are calculated, see our guide on how credit scoring works and the five factors that determine your score.

Experiment Index: 6 Tests, Real Results

Below is every experiment we've completed so far. Each links to a full writeup with methodology, raw data, and actionable takeaways.

Key Findings Across All Experiments

After running six major experiments and tracking scores across 180+ days, here are the patterns we keep seeing:

1. Utilization Is the Fastest Lever

Nothing moves your score faster than changing your credit utilization. It's the only major factor with no memory — last month's utilization doesn't matter once the new balance reports. According to Experian, utilization accounts for roughly 30% of your FICO score, and our experiments confirm it behaves as the most volatile and controllable factor.

In our utilization experiment, the difference between 1% and 30% utilization was 43 FICO points — potentially the difference between a prime and subprime interest rate on a mortgage. The average American carries 28% utilization (TransUnion, 2025), meaning most people sit in the heavy penalty zone.

2. Starting Score Determines Impact Size

Every single experiment showed larger point swings for lower starting scores. A hard inquiry that costs 7 points at 720 might cost 3 points at 800. Closing a card that drops a 680 by 40 points might only drop an 800 by 15. The scoring models are non-linear — there's more distance between points at the lower end of the scale.

3. Scoring Model Differences Are Real and Large

FICO 8, FICO 9, and VantageScore 3.0 can produce differences of 30-50 points for the same profile and the same action. This isn't a rounding error — it fundamentally changes what advice is correct for you. The CFPB reports that 90% of top lenders use FICO scores, but which version varies by lender and loan type.

4. Recovery Timelines Are Predictable

Most negative score impacts follow a decay curve. Hard inquiries recover in 3-6 months. Utilization spikes recover the next billing cycle. Closed account age impacts take 1-3 years to fully manifest. Knowing the timeline changes the decision calculus entirely.

5. The Average American Is Leaving Points on the Table

With an average FICO score of 715 (as of late 2025) and average utilization around 28%, the typical consumer is sitting 30-45 points below their achievable maximum just from suboptimal utilization management. Add in unnecessary hard inquiries and unresolved collections, and the gap widens further.

Why Scoring Model Differences Matter

One thing our experiments hammered home: you don't have "a credit score." You have dozens. And they don't agree with each other.

Here's how the three major models treated the same actions differently across our experiments:

VantageScore tends to be more volatile — bigger swings in both directions. FICO 8 is the most stubborn about collections. FICO 9 is the most forgiving. Your strategy depends on which model your target lender uses.

The 2026 Scoring Landscape: FICO 10T and VantageScore 4.0

The credit scoring world is undergoing its biggest shift in years, and it directly affects how you should interpret our experiment results:

Mortgage Lending Is Changing

The Federal Housing Finance Agency (FHFA) has approved VantageScore 4.0 for immediate use on Fannie Mae and Freddie Mac mortgages. FICO 10T implementation is expected to complete by Q4 2026, with full industry transition by 2027. This means the legacy FICO models (2, 4, 5) used for decades in mortgage lending are finally being retired.

As of early 2026, Fannie Mae and Freddie Mac now accept VantageScore 4.0 for mortgage decisions, and FICO 10T adoption is expected by Q4 2026. This is the first major change in mortgage credit scoring in over 20 years (FHFA, 2025).

What FICO 10T Changes

FICO 10T uses "trended data" — it looks at your credit behavior over the past 24 months, not just the current snapshot. This means someone who has been steadily paying down debt will score higher than someone with the same current balance who has been accumulating debt. Our utilization experiment, which showed that utilization has "no memory," may partially change under FICO 10T — the trend of your utilization will matter, not just the current number.

What VantageScore 4.0 Changes

VantageScore 4.0 incorporates rent, utility, and telecom payments — data that traditional models ignore. It also uses machine learning techniques and can score consumers with as little as one month of credit history. According to VantageScore, this allows an additional 37 million Americans to receive a credit score who were previously "unscorable."

We're planning new experiments specifically testing FICO 10T and VantageScore 4.0 as lender adoption increases throughout 2026. Stay tuned.

How to Use This Research

These experiments are designed to help you make better decisions. Here's our recommended approach:

1. Know your scoring model. Ask your lender which FICO version they pull. Mortgage lenders are transitioning to FICO 10T and VantageScore 4.0 in 2026. Credit card issuers often use FICO 8. Auto lenders use FICO Auto Score 8.
2. Find your starting score range. The same advice doesn't apply equally to a 620 and a 780. Our experiments break down results by score range so you can find your match.
3. Read the experiment that matches your situation. About to close a card? Read the closing card test. Considering adding someone as an authorized user? Check the AU data. Thinking about a balance transfer? See the balance transfer impact data.
4. Plan for the timeline. Every experiment includes a recovery timeline. Use it to time your actions before major credit applications.
5. Stack strategies intelligently. Our experiments test one variable at a time, but in real life you can combine them. Lower utilization + authorized user status + zero new inquiries for 6 months = maximum score optimization before a mortgage application.

For a comprehensive overview of credit scores and how they work, start with our complete credit scores guide. If your score dropped unexpectedly, our why your score dropped guide walks through every possible cause.

Frequently Asked Questions