Payment History Is the Largest Single Factor in Your Score. What That Means in Practice

FICO publishes the rough weighting of its scoring categories. Payment history is about 35% — the largest single component. Amounts owed is about 30%, length of history 15%, new credit 10%, credit mix 10%.

Those numbers get repeated everywhere. What they mean is discussed much less.

"35% Payment History" Does Not Mean What People Assume

It does not mean that paying on time earns you points. It means that not paying on time costs you a great many. Scoring models are built to predict the likelihood of a serious delinquency, and the strongest predictor of a future missed payment is a past one.

The practical consequences:

• On-time payments are the baseline. They are what the model expects. Consistency over time is what accumulates.
• A single 30-day late is significant, particularly on a thin file where there is little else to average against.
• Recency matters. A late payment from four years ago weighs less than one from four months ago.
• You cannot have payment history without accounts. This is the part people miss. If nothing reports, the largest factor in your score has nothing to work with.

Which Is Why a Thin File Is Its Own Problem

If you have no reported accounts, you are not scoring badly on payment history — you are not scoring on it at all. The model has no data.

That is the actual argument for adding a reported account to a thin file: not that it earns points, but that it gives the largest scoring factor something to measure.

That is also the entire argument for Feodis.co, stated as narrowly as we can make it. We put an account in your name and report the payment on it to all three bureaus every month. We are not claiming that earns you points — see below for why nobody can claim that. We are claiming it gives the 35% something to look at, which it currently does not have.

What This Does Not Tell You

It does not tell you how much any specific action will move your score. Nobody can tell you that, including anyone who says they can. The weightings are category-level and approximate, the models are proprietary, there are several of them in use, and your result depends on everything already on your file.

Anyone quoting you a number of points is guessing, and usually selling.

Source: FICO's published description of score composition, fico.com.