What four years of course actuals and four years of application records tell us about raising price — what they support, what they only flag, and the three things we will do in 2027 to answer the rest. Prepared for the 2027 Early Pricing decision. Version 2, November 16, 2026.
Price elasticity is simply this: if we raise the price of a course by 10%, how many fewer families enroll?
If a 10% increase costs us 10% of our enrollments, demand is elastic — families are shopping on price, and raising it is self-defeating. If a 10% increase costs us almost nothing, demand is inelastic — families are choosing on fit, timing and what the course does for their child, and price is a secondary consideration. Most of what we sell appears to sit closer to the second case, but that conclusion needs to be held carefully, for a reason this report is built around.
In our inventory the measurement is awkward, and it is worth being plain about why:
So read every number in this report as a measurement of fill response to list price, within a catalogue of weakly substitutable products, over a period when we barely moved price. That is a real finding and a real limit at the same time, and the 2027 plan is designed to convert the limit into evidence.
This is the finding that should carry the Early Pricing decision. Across 337 same-course comparisons from 2023 to 2026, once we account for the year and how full a course already was, a 10% price increase moves fill by about zero points. Courses that raised price 5–10% saw fill essentially unchanged, and 2.8 points better than courses that held flat. The application record says the same thing from the demand side: across 9,731 cancellations, price was the stated reason in 2.4% of cases. Families leave over schedule conflicts, transfers, medical issues and motivation — not over cost.
Two honest qualifications travel with that thesis, and both are addressed in the 2027 plan rather than argued away. First, "no measurable penalty" is a weak statement on this much data, not a strong one — the plausible range runs from roughly −8 to +7 fill points, so the evidence cannot rule out a real cost. Second, price resistance is real where price is highest: price-related cancellations rise from 0.7% of cancellations on courses under $2,500 to 6.0% on courses above $9,000. The thesis supports the proposed step. It does not support an unlimited one.
Early Pricing holds current published prices through February 28. Standard Pricing begins March 1. Contingency incentives are held in reserve for courses at risk. The right-hand columns convert the proposed dollar increases into percentages against real 2026 inventory, which is how families experience them and the only unit in which elasticity can be judged.
| Course type | Early Pricing through Feb 28 | Standard Pricing from Mar 1 | Contingency incentive |
2026 price range | Increase at median price | Increase range across the tier |
|---|
2026 figures are demand-relevant sections only. Every tier lands in the same narrow band: a 9–10% step at the median. That consistency is good for communication and awkward for measurement — nothing in 2023–2026 approaches it as a coordinated, dated, network-wide move, which is precisely why the rollout is worth instrumenting.
All 337 same-course comparisons, grouped by size of price change. The bar at zero is courses that held price exactly flat from one year to the next.
127 of 337 comparisons (38%) held price flat and 120 more rose between 0% and 5%. Only 45 (13%) moved more than 5% in either direction. A price-sensitivity estimate built on this much variation will have wide error bars no matter how carefully it is done — which is the single most important thing to understand before reading the rest of this report.
Each dot is one course compared with itself the year before: percentage price change on the horizontal axis, change in fill rate on the vertical. The steep rust line is the raw relationship. The flat navy line is what remains after holding the year and the course's prior-year fill constant.
Hover or tap any point for the course, school, year and both prices.
The raw slope across all 337 comparisons is −13.4 fill points per +10% price. If that were causal, the 2027 step would be a serious mistake. But two things are tangled up in it. Courses that raised prices tended to start fuller (the +0–5% group averaged 85.8% prior fill against 80.2% for flat courses), and full courses have nowhere to go but down. And the largest increases cluster in 2024, when fill fell everywhere. Hold the year and prior fill constant and the slope flattens to −0.4 points per +10%.
For a Board audience the takeaway is compact: the scary number is an artifact; the reassuring number is real but imprecise.
Controlled estimate of fill change per +10% price increase, by course length, with 95% confidence intervals. A bar crossing zero means we cannot distinguish the effect from no effect. Only 8–10 day courses clear that bar.
8–10 day courses show a −15 point fill decline per +10% price across 104 comparisons, p=0.003. Even the optimistic end of that interval is a real decline, and the pairs are spread across schools and years rather than concentrated in 2024, which makes it harder to dismiss.
58 of those 104 pairs are 8-day courses, and the proposal's 1-week tier covers 5 to 8 days. So the largest block of short-course seats in the network — 40 sections and 402 seats in 2026 — sits in the only price-sensitive segment we can identify and is being asked to absorb +9.4%.
Why it matters beyond this tier
This is the tier closest to competitors on price and the entry point into the network. A family's first Outward Bound course is usually a short one; if we price that first step badly we lose the longer courses that follow it. The risk is not the revenue in this tier — it is the top of the funnel for every other tier.
The 1-week tier spans 5 to 8 days and 2026 prices from $1,825 to $3,380. The same +$300 is +9.4% on an 8-day course and +15.8% on a 5-day one — and 5-day courses already filled worst in 2026, at 43.5%.
Nine- and ten-day courses sit outside the published 5–8 and 12–15 day bands entirely — 34 sections and 390 seats with no assigned tier. Worth settling before Schools respond.
Network fill rate against average price per day, 2023–2026. Price per day rose about 8% across the four years while fill fell 10 points and then recovered 18.
Every segment moved together in 2024 and again in 2025. That is the signature of market conditions, not of course-level pricing decisions — and it is why a weak year is a far bigger threat to the 2027 forecast than the price step is.
| Segment | 2023 | 2024 | 2025 | 2026 | 2026 sections |
|---|
Cells are fill rate; shading runs from rust at 45% to teal at 90%. Section counts fall across the period because inventory was deliberately reduced, which is why fill and not enrollment volume is the demand measure here.
Sections grouped into price-per-day quartiles within their own segment and year, 2023–2026 pooled. If families were choosing on price, the low quartile would fill best. It does the opposite in nearly every segment.
This is correlation, not cause: higher-priced courses are often different products in different places and seasons. The claim is narrow and defensible: there is no evidence in our inventory that being the cheapest option in a segment wins enrollments. In 5 of 6 segments the cheapest quartile cancelled for low enrollment more often than the most expensive. Semester courses carry the lowest price per day in the network and the worst fill.
Share of each price band's cancellations where the stated reason was financial aid or choosing a non-OB program.
Across all 9,731 cancellations only 2.4% named price. But the gradient is clean and monotonic: 0.7% under $2,500 rising to 6.0% above $9,000, a ninefold increase. Price resistance is real and it concentrates at the top of the ladder — which is exactly where third-party financing, rather than discounting, is the right instrument.
All stated cancellation reasons, 2019–2022. Gold bars are supply-side events, not demand decisions.
Schedule conflicts, transfers between courses and schools, medical and motivational withdrawals dominate. This is the demand-side version of the core thesis: families are not leaving over cost. It also shows how much of our cancellation volume is movement within the network rather than loss from it — transfers alone are 23.6%.
A course that sells out shows 100% fill whether twelve families wanted it or thirty did. Every elasticity estimate in this report is compressed toward zero as a result, and every capacity decision is made half-blind. The application record lets us see the part we have been missing.
Why this matters for 2027 and beyond
Waitlist demand peaks in the $4,000–$9,000 bands, at 6.3% and 6.6% of applications — the two and three week courses. That is unmet demand on exactly the inventory we are raising prices on, and it is invisible in fill rate. The waitlist changes give us a demand signal above 100% for the first time. The operational prize is larger than the measurement one: a course that hits 100% early enough in the sales cycle, with a visible queue behind it, is a candidate for an additional crew rather than a closed door. Capturing that queue is also the only way to know whether a price increase suppressed demand or merely trimmed an overflow we never counted.
Cancellation rate by how far ahead the family applied. 2019 summer courses only — the clean pre-COVID year, school-side cancellations removed.
A February 28 application is 108 to 158 days ahead of a typical summer course start. In the clean 2019 data those are the lowest-cancellation buckets in the entire distribution — 27.1% and 21.7%, against 35.9% for families who applied inside two weeks of the start. Families who commit in that window are the ones who show up.
This is a real point in the Early Pricing model's favor, and it is worth stating carefully because the intuition runs the other way: pulling demand forward does not obviously mean pulling in less committed families. It only becomes a risk much further out, beyond roughly eight months, where cancellation climbs sharply again. A February deadline for a June–August course does not reach that zone.
Caveat worth carrying: this is 2019 behavior, and booking windows have moved since. It is a reason for confidence in the mechanic, not a forecast of 2027 cancellation rates. Test 1 measures the real thing.
The economics of our own campaigns are unusually clear. March 2026's early campaign cost $426 per enrollment. Late rescue discounting on low-enrolling courses cost about $1,000 per enrollment — roughly 2.3 times as much for a family who was already further from saying yes. Meanwhile the median enrollment window shortened from 152 days in 2022 to 110 in 2026, so the window in which an early incentive can work is narrowing.
Holding current prices as Early Pricing changes when a family gets the best price rather than what the best price is. Given that price position within a segment has never predicted fill, a broad price cut would almost certainly have given up margin without buying enrollments. Rewarding timing is the lower-risk lever, and it is the cheaper one.
Share of applications receiving any aid, and the effective discount against list, 2019–2022.
Between 6% and 12% of applications received aid, at an effective discount of 3.4% to 6.3% of list. Critically, the discount field was essentially unused — under 0.2% of applications carried a campaign discount; scholarships did all the work. That matters for 2027: contingency incentives are a new behavior for this network, not a continuation of an existing one. We have no baseline for what a campaign discount buys us, which is why Test 3 logs every activation from the first one.
Because year-level demand moves fill far more than course-level price, a broad discount is a bet on the wrong variable. Holding the −$150 / −$250 / −$350 / −$450 contingencies for courses that fall behind pace puts the money where the variance actually is, and keeps the published price intact for the 90% of courses that will not need it. The discipline this requires is agreeing the activation thresholds before the season, not during it — a reserve that gets spent reactively in April is just a late discount with extra steps. It also produces something our history lacks entirely: a mid-season price cut on identified courses, which is a cleaner elasticity reading than anything in the 2023–2026 file.
The finance conversation this report should trigger is not "is the increase safe?" but "how much fill can we afford to lose before the increase stops paying for itself?" That is arithmetic, and it depends on assumptions the Schools and finance officers should set rather than inherit. The calculator below is built so those assumptions can move — every price, increase, seat count and mix is editable. If the Early Pricing model changes after finance review, change the inputs and the answer updates.
| Tier | Seats | Price | Increase | Step |
|---|
The curve shows incremental revenue at every level of fill loss. Where it crosses zero is the break-even — the point at which lost tuition exactly cancels the increase. The marker is your current assumption.
Fill is only known in August. These are the leading indicators that distinguish a healthy rollout from a struggling one while there is still time to act.
| Window | What we watch | What it means |
|---|---|---|
| Feb 1 – Feb 28 Early Pricing | Weekly application pace against 2026 same-date pace, by segment | Pace running ahead of 2026 means the deadline is pulling demand forward. Running level means the mechanic is not landing and the March 1 step will meet an ordinary pipeline rather than a depleted one — which is the better of the two for revenue but tells us the incentive did not work. |
| Mar 1 – Mar 31 The step | Pace in the four weeks after the switch against the four weeks before, and against 2026's same four weeks | A drop against 2026 is the first real elasticity reading we will have. A drop against February alone is expected and means little — demand pulled forward has to come from somewhere. |
| By Apr 1 Decision point | Cumulative pace by segment, with the 1-week tier reported separately for 5–6 day and 7–8 day courses | More than 15% behind 2026 triggers the contingency incentive. This is the last point at which an intervention still has time to fill a summer course. |
| Throughout Demand ceiling | Waitlist additions on courses at or near 100% | Tells us whether a flat fill number is a demand ceiling or a capacity ceiling — and flags candidates for an additional crew while staffing can still be arranged. |
| Throughout Net price | Share of enrollments at Early, Standard and Contingency tiers | The realised mix against the 35/50/15 assumption. If the Early share runs well above 35%, the revenue gain is smaller than forecast even with fill fully intact. |
The principle running through all of these: the 2027 rollout is already a large, dated, network-wide price change. It is the controlled experiment our history never gave us, and it costs nothing extra to measure. Nothing below holds out courses, adds price variation, or changes the proposed structure.
Track weekly applications and enrollments by segment and by school for participating courses, against 2026 pace on the same calendar dates with the same course mix. Report the 1-week tier separately for 5–6 day and 7–8 day courses, since their percentage increases differ by nearly seven points.
Stamp the pricing tier — Early, Standard or Contingency — on the enrollment record at the point of application, alongside any scholarship or discount amount. Add a payment-type field if one does not already exist, so third-party financing uptake reports without a separate process.
A four-question instrument riding along with existing follow-up, asked of newly applied families, last year's families, and — most valuably — leads who never converted. The core question separates noticed the price from was deterred by the price, which no amount of enrollment data can distinguish.
Everything above measures a price change we are already making. None of it creates the one thing that would settle the question outright: comparable courses at deliberately different prices in the same season. That is the only design that isolates price from everything else, and it is the right thing to hold until 2028 — both because it needs to be built into inventory planning rather than bolted on, and because 2027 is already absorbing a large structural change.
The three actions above are the plan. These are the supporting records that make them readable, each with an agreed home so that no measurement depends on someone remembering to keep a spreadsheet.
| Measurement | Where it lives | What it produces |
|---|---|---|
| Weekly pace by segment and school Supports Action 1 | OE Command Center | The core elasticity read. 2026 baseline locked before Early Pricing opens. |
| Participating vs. non-participating courses | Tracked outside Salesforce this year, or a single flag field reportable through the Command Center | Separates the price step from 2027 market conditions. Uses differences created by School choice, so each School's reason for opting a course in or out should be recorded alongside — it is the only way to judge the comparison later. |
| Discount per enrollment, above trendline | Dedicated tab or chart in the OE Command Center | Separates campaign-based from reactive incentives once Standard Pricing is live. Benchmarks: $426 per enrollment on the March 2026 early campaign; about $1,000 on late rescue discounting. Sets activation rules for 2028. |
| Financing uptake | Payment-type field in School Salesforce instances, synced to OBUSA | Whether the barrier is cash flow or price. High uptake on longer courses supports premium positioning without deeper discounts. |
| 1-week tier timing flag | OE Command Center | Weekly pace check on 5–8 day courses after March 1, flagged against the 15%-behind trigger. The response to an activated contingency is itself a measured price cut — this test produces an answer either way. |
| Application-level reporting with cancellations | Standing report, 2023–2026 | Discounts, scholarships and cancellation reasons at application level, including cancellations landing near the 90-day tuition due mark. Requires a status-change date field, which the current export does not carry. |