Skip to main content

​
Overview

Opportunity Sizing answers the question people actually ask about a metric, which is not “where does this number vary?” but “how do I improve it, and by how much?” It lives in the World Model Graph, reached through the Globe icon in the icon rail (a surface distinct from both the Context Graph and the Metric Tree). When you open a measure on the graph, Oxygen scans your data for segments that underperform a benchmark, puts a dollar figure on each gap, and then lets you drill that gap into its underlying drivers until you reach something specific enough to act on. The result is a ranked list of addressable upside, such as “mobile orders run $218.50/order below in-store, and closing that gap is worth +$264.6k,” in which every number is backed by a statistical test and every child in the breakdown provably sums back to its parent.
Opportunity Sizing operates on the semantic model, reading your existing measures, dimensions, and the relationships between views, so you never define opportunities anywhere yourself. The quality of the sizing depends directly on how well your model is defined (see What your model needs).

​
The core idea: size on rates, not totals

An opportunity has two ingredients, a gap and a volume:
The single most important design decision is that Oxygen compares per-unit rates and never raw totals. If you compared raw sums, a segment sitting below another would mostly reflect the fact that it is smaller, with fewer rows or a smaller market, rather than that it underperforms. Comparing revenue-per-check instead of revenue makes the comparison about performance, so that “smaller” never reads as “worse,” and multiplying the per-unit gap back by the segment’s real volume then turns it into an honest, addressable dollar figure.
Oxygen refuses to guess rather than emit a misleading number. If a measure is an additive total (like a sum) but its view declares no count measure from which to form a rate, the panel returns an empty result explaining why, and it will not silently fall back to comparing totals. This is deliberate, since a size-confounded number that looks precise is worse than no number at all.

​
Picking the benchmark

The benchmark is the bar every segment is measured against, and Oxygen picks it from the data automatically: Crucially, the benchmark is defined as a population filter (dim = best_peer or dim IN (segments at or above p75)) rather than as a bare number, which is what lets the drill query the benchmark population later instead of merely comparing against a fixed scalar.

​
Three sizing modes

How a segment is weighted into a dollar figure depends on what the measure supports: Only rows responses carry a directly addressable upside and get ranked in the Opportunities list, although the other bases still power the drill breakdown.

​
Trustworthy numbers: the evidence gate

Any tool can find a “top lever” in any dataset, including one in which the dimension is provably unrelated to the measure. Opportunity Sizing is deliberately rigged against itself and corrects for the two ways a naive scan fools itself.
The benchmark is the maximum (or P75) of the same segments being tested against it. If you draw a handful of values from pure noise, the largest one naturally sits well above the mean, so a fixed threshold would “discover” a leader in literally any dimension. Oxygen raises the bar by exactly that expected amount, so that noise does not clear it.
A full scan makes on the order of a hundred comparisons once related entities are followed, and at a naive 5% per test, five spurious levers per scan is the expected yield. Oxygen therefore applies a Šidák correction across the whole scan family, so that 20 dimensions × 5 segments requires a much higher bar of evidence than any single dimension would imply on its own.
The test itself is Welch’s t-test rather than a pooled-variance test, because the benchmark is routinely the thinner of the two populations, and a pooled estimate would understate the error exactly where it matters most. With a two-row benchmark the effective degrees of freedom collapse, so the gap has to be genuinely large to survive.

​
Three outcomes, kept distinct

Oxygen never blurs “proven” with “plausible”: The third row is the honest one. gated: false does not mean the gap is fake; it means nobody could prove it real, and the panel never presents it as proven upside. The skip reason also distinguishes “everything I found was noise” from “everything already matches the benchmark,” because those are different answers and you are entitled to know which one you got.

​
Drill: decomposing the gap into drivers

Finding a gap is only half the job. Drill decomposes that gap recursively, alternating between two kinds of split, until it reaches a driver specific enough to act on:
  • Component splits come from the measure’s own formula (revenue → entrée revenue + add-on revenue).
  • Dimension splits cut by an attribute (add-on revenue → by category).
The flow is recursive and domain-free, decomposing a per-unit gap through structure until it reaches something specific.

​
Two rules make the math honest

1

The denominator never changes

When the numerator is split by a dimension, the denominator is not filtered along with it, so sides/order means sides revenue divided by all of that segment’s orders rather than by the sides count. If the denominator changed, the children would stop summing to the parent.
2

The benchmark is inherited, never re-picked

Every level compares the same two populations chosen at the root. If each child picked its own best peer, the children would stop summing to the parent and every percentage would be fiction.

​
Why those rules matter

Given both rules, the decomposition becomes an arithmetic identity rather than a fresh comparison at each level:
This is what earns a child the right to say “91% of the parent gap.” The figure is not an estimated or heuristic ranking but plain arithmetic, and the children provably sum back to the parent at every level.

​
What a drill looks like

The last split (sides/order = attach_rate × price_per_side) answers whether the gap comes from fewer sides or cheaper sides, which is the classic volume-versus-value question, and it does so inside the decomposition without any extra machinery.

​
At each level, the drill:

  1. Enumerates candidates, meaning the component edges from the measure’s formula plus every dimension not already consumed higher up.
  2. Computes each child’s gap and its share of the parent gap.
  3. Ranks by concentration, identifying which child holds the most of the gap.
  4. Shows all children with their shares rather than just the winner, so siblings stay visible and “follow the max” remains a reading aid rather than a hidden selection.
  5. Recurses into the largest child.
  6. Stops when the next split fails the evidence gate, cannot be evidenced, has no candidates left, or reaches the maximum depth.
Every level renders its stop reason, and a level that could not be evidence-checked is shown as ungated rather than as proven. The statistical error budget is also split across depth, so that an N-level drill does not quietly spend the full 5% error at every line.

​
How to use it

1

Open the World Model Graph

Click the Globe icon in the icon rail to open the graph of your semantic model.
2

Select a measure

Click a measure node. Oxygen scans its segments when the node opens and, if it finds addressable upside, reveals an Opportunities section labeled addressable upside · and where it comes from.
3

Read the ranked segments

Each row shows the segment, its per-unit rate against the benchmark (543.9 → 762.4 per order), the dollar upside, and a proportion bar. You can use the copy button on a row to drop a ready-made investigation question into chat.
4

Drill into a gap

Expand a segment to decompose its gap. The drill runs only on expand, never when the node first opens, because it issues recursive queries against your warehouse. Follow the highlighted child downward until the stop reason tells you there is nothing left to prove.

​
What your model needs

Opportunity Sizing is only as good as the semantic model underneath it, so to get the most out of it:
  • Declare a count measure on the view. Without a count denominator, a sum measure cannot be put on a per-unit basis, and Oxygen will refuse to size it rather than compare totals.
  • Model your metrics compositionally. A measure defined as a sum or difference of same-view measures (net_revenue = entrée_revenue + add_on_revenue) gives the drill component edges to follow. Mirror trees work too, so a gross_profit tree that parallels revenue lets you drill margin the same way you drill revenue.
  • Keep dimensions meaningful to benchmark across. A dimension in which being “higher” is a matter of arithmetic rather than performance (party size, for example, since a 6-top spends more than a 2-top by definition) will dominate the ranking without telling you anything, so prefer dimensions where a gap reflects something you can change.
  • Point it at measures where higher is better. Both benchmark bases are top-end rates, so a segment below the benchmark is always read as upside. That reading is right for revenue, margin, or conversion but backwards for a cost, since a cost sum with a count on its view would rank the cheapest segment as the biggest opportunity. Size the revenue or profit side instead, and let the drill walk into the cost components rather than rooting a scan on a cost measure.

​
Current scope

Opportunity Sizing is honest about what it can and cannot do, and about why it cannot. What it refuses, it refuses by design, because sizing those shapes would break the arithmetic identity the drill depends on. There is no “not yet” bucket: a shape either sizes today or marks a permanent boundary, and is never a quietly deferred promise. Works today
  • sum roots with a count on their view
  • Additive same-view composites (a + b − c), including nested ones, where a composite whose child is itself a same-view composite flattens transitively
  • Homogeneous +/− and ×/÷ expressions
  • Dimension splits and cross-grain numerators
Refused by design, because Oxygen would rather decline than emit a number that looks precise but isn’t. These shapes break the sum-to-parent identity, so no future work will “fix” them without a different model: Comparing a revenue drill against a margin drill side by side already works today, since it is simply two drills read together. Ranking add-on leaves by profit rather than revenue changes which leaf wins, so a true single-pass co-ranking (“big revenue gap, thin margin”) would be a presentation improvement rather than an engine limitation, given that the engine already sizes each measure on its own.

​
In one paragraph

Oxygen sizes an opportunity as (per-unit rate gap against a benchmark segment) × volume, working per unit rather than on raw totals so that “smaller” never reads as “worse.” The drill then decomposes that gap recursively, alternating splits from the measure’s own formula with splits by dimension, while holding the denominator and the benchmark fixed all the way down. Those two constraints turn the decomposition into an arithmetic identity, which is what lets a child legitimately claim “91% of the parent gap” and guarantees that children sum back to the parent. The drill stops when the evidence gate can no longer prove the next split, and it always tells you why it stopped.

Measures

Define the measures and composite metrics that Opportunity Sizing decomposes

Dimensions

The attributes the drill splits a gap by

The Metric Tree

How measures compose, and the driver edges the drill follows

World Model

How the semantic model, metric tree, and entity graph fit together