How to index data to 100 for a fair comparison

To index data to 100—also called rebasing or normalizing to a base of 100—you replace each raw observation with its value relative to one selected period. The selected base period becomes 100 for every series. Later values show percentage movement above or below that common starting line.

If an indexed value is 125, the series is 25% above its base-period value. If it is 82, the series is 18% below the base. The index has no original unit: it is no longer dollars, subscribers, tonnes, or percentage points. That unit-free scale is exactly what makes unlike series visually comparable.

Indexing is most useful when the editorial question is about relative performance. A financial newsletter might compare Bitcoin, gold, and an equity index since a market event. A policy analyst might compare wages and consumer prices since 2020. A creator might compare newsletter subscribers, podcast downloads, and site traffic since launch even though the raw counts begin at very different levels.

The calculation is performed separately for each series. Divide an observation by that same series’ value in the base period, then multiply by 100. Do not divide every series by one shared raw number: each series has its own denominator at the selected date.

Raw series can share a chart only when their units and scales support a meaningful comparison. A stock priced at $500 will visually dominate one priced at $25 even if the cheaper asset grew faster. A subscriber count and a revenue series cannot share a single quantitative axis at all. Indexing removes those starting levels and asks a narrower question: how much did each series change relative to itself?

The transformation preserves the shape and relative changes within each series. Peaks, troughs, and growth rates occur in the same places; only the vertical scale changes. The result makes it easier to see which series grew faster, fell further, or recovered sooner from the chosen base period.

That benefit comes with a cost. An indexed chart cannot tell readers which company has more subscribers, which country has the larger economy, or which asset costs more. Two lines that both reach 150 grew 50% from their own baselines, even if one increased by millions of units and the other by hundreds. State the relative-performance question clearly in the title.

The base period determines the comparison the reader sees. Choose a date tied to the editorial question: the start of a policy, a market shock, a product launch, the first full month of a campaign, or the beginning of a calendar year. The earliest shared observation is a convenient default, but convenience is not the same as meaning.

Every series must have a valid observation in the base period. A zero value cannot be used because division by zero is undefined. A value extremely close to zero can create enormous, unstable index values. Missing observations require a documented choice: move the base period, align to a common frequency, or use a period average when that is methodologically appropriate.

Base-period choice can materially change the apparent gaps between multiple series because it changes where they are forced to meet. It does not change the underlying path within a simple series, but a strategically chosen start date can still flatter or punish one comparison. Test a nearby reasonable base period and disclose the date prominently.

Before rebasing multiple series, make them analytically compatible. Daily market prices should use the same trading-day convention. Monthly economic series should refer to the same month or the same monthly average. If one observation is seasonally adjusted and another is not, indexing will not repair the conceptual mismatch.

Decide how to handle gaps. Forward-filling can be reasonable for a balance that genuinely remains unchanged until the next observation, but it is misleading for many prices, rates, and flows. Interpolation creates estimated values and should be visible in methodology. Often the cleanest editorial choice is to begin at the first date shared by every series.

Also distinguish nominal and real values. Indexing nominal wages and a price index can show their relative paths, but it does not automatically calculate real wages. Deflating a nominal series is a separate transformation. Likewise, currency conversion, per-capita adjustment, seasonal adjustment, and smoothing should happen deliberately and be described in the method note.

For a simple series indexed to your selected base period, subtract 100 to describe the percentage difference from that base. An index of 142 is 42% above the base level; an index of 76 is 24% below. Say “the series rose 42% from the base period,” not “it rose 42 index points,” unless you specifically mean the movement on the index scale.

Comparing two later index values requires care. If Series A is at 140 and Series B is at 120, A grew 40% from its base and B grew 20%. It is tempting to say A grew “20% more,” but the direct difference is 20 index points. Depending on the question, you may instead say A’s relative gain was twice B’s. Be explicit about the denominator.

Do not apply the subtract-100 shortcut to every published composite index. Measures such as consumer price indices or stock-market indices may have distant or constructed base periods and are commonly used to calculate change between two current dates. For any two index observations, calculate percentage change as (later ÷ earlier − 1) × 100.

A trustworthy indexed chart should be reproducible from the source data. Preserve the raw series, the exact base period, the transformation formula, and any alignment or filtering decisions. Downloading only the rebased output makes later updates and corrections harder.

The title should describe relative movement, not absolute leadership. “Gold outperformed equities after the rate decision” is appropriate for an indexed comparison. “Gold became more valuable than equities” is not supported because the raw units and total values were removed.

Put the base period next to the chart, ideally in the subtitle or vertical-axis label: “Jan. 1, 2025 = 100.” The source line should name the original providers, while the methodology note explains the transformation. For a newsletter, the full method can sit below the image; for a social graphic, include at least the base and concise source on the image itself.

Most errors come from an ambiguous baseline or incompatible inputs rather than the formula itself. A clean-looking chart can still be wrong if one series begins on a different date, one uses month-end values while another uses averages, or a zero has been treated as a missing value. Validate the base row and one manual calculation before publication.