GA4 active users on outwardbound.org program pages, Jan 1 – Jun 30, 2026, compared by state with OE enrollments on 2026 courses. Enrollments cover all 1,833 students on 2026 courses, including fall and winter courses still to run. City data (Jan 1 – Oct 12) is used as a cross-check, not in the state math.
Where catalog traffic comes from and where students come from line up closely (correlation 0.94 across states), so the catalog is drawing the markets we would expect. The exceptions sit mostly in the middle of the country. The Midwest and Plains generate about 20% of catalog traffic but only 12% of enrollments, and the South Central states (Texas, Kentucky, Oklahoma and neighbors) trail as well. Georgia and South Carolina under-convert despite bordering NCOBS. Washington has moved furthest from its own history: it sent 74–89 students a year in 2021–23, when NWOBS was operating, and 34 in 2026, while its catalog traffic per resident is still well above the national average. That looks like a market the network is not yet capturing. North Carolina, Colorado, California, New York and the rest of the Northeast corridor enroll well above their traffic share. That supports our working assumption that proximity to course areas and familiarity with OB help conversion, though we don't have geographic brand-awareness data to test it directly. Florida, the third-largest traffic source, converts right at the network rate.
100 = a state enrolls exactly in proportion to its share of catalog users. Teal states over-deliver; rust states send traffic that doesn't enroll at the same rate. Faded tiles have fewer than 4 expected enrollments, so treat them as directional.
Census divisions, which pool small states into samples large enough to trust. Bars show the 2026 index; the dark tick shows the same index using 2023–25 average enrollment, a check on whether the gap is new.
The line is what a state would enroll at the network rate of enrollments per 1,000 catalog users. States below it are generating traffic that doesn't convert proportionally.
Hover or tap a dot to see the state. Only the largest markets are labeled.
West North Central (MN, IA, MO, KS, NE, ND, SD) enrolled 54 students against about 122 expected from its traffic, an index of 44. It was 60 in 2023–25, so the gap predates this year and widened in 2026. East North Central indexes 68 (160 of 234 expected): Indiana (14 of 39), Michigan (27 of 54) and Wisconsin (17 of 28) pull it down, Ohio is under (44 of 56), and Illinois is at its share (58 of 57).
Minnesota is a 2026 change, not a pattern. It sends 2,500 catalog users and enrolled 26 of 46 expected (index 57), but it was at parity in 2023–25 (93). Minnesota students on VOBS courses fell from 31 in 2025 to 15 in 2026, and the Minneapolis–St Paul metro has the weakest enrollment-to-traffic ratio of any large metro in the city data.
Georgia has one of the largest single-state gaps in the network: 40 enrollments against 65 expected, from 3,600 catalog users. South Carolina enrolled 12 of 31. Both sit next to NCOBS's home territory and both under-indexed in 2023–25. North Carolina, by contrast, indexes 161. Atlanta also sits in the bottom third of large metros in the city data.
West South Central indexes 79 (95 of 120 expected). Texas is the biggest market there at 77 of 85, with Oklahoma (5 of 14) and Louisiana (4 of 11) further behind. Within Texas the gap is Dallas–Fort Worth; Houston and Austin enroll at or above their traffic. East South Central indexes 64, led by Kentucky (8 of 22). In the West, Utah (7 of 17) and Nevada (5 of 15) under-index even though Colorado next door converts at 1.47× its traffic share.
Washington enrolled 34 students against 56 expected (index 61, p=0.002). Its 2023–25 index (97) is inflated by NWOBS, which enrolled 61 Washington students in 2023–24 before closing. The cleaner comparison is 2025, the first year without NWOBS: 48 Washington students (index about 84), falling to 34 in 2026, mostly at OBCA (13 to 5), VOBS (7 to 2) and COBS (14 to 10). The Seattle metro shows the same drop. Looking further back, Washington sent 74–89 students a year in 2021–23, about 27–35 of them to NWOBS. Other Schools picked up some of that in 2025 but not in 2026. Washington still generates 39 catalog users per 100,000 residents against a national average of about 29, so interest is there. It is a good candidate for conversion work in the Pacific Northwest.
New York (154), Connecticut (174), Massachusetts (149), Pennsylvania (146), New Jersey (146), DC (154), North Carolina (161), Colorado (147) and California (124) all enroll well above their traffic share, and did in 2023–25 too. The Mid-Atlantic and New England divisions together enrolled 526 students against about 362 expected. New York alone is 63 above. At the metro level, Philadelphia, Washington DC, New York and the Bay Area lead. This fits our working assumptions about proximity and familiarity, but the data here does not prove why.
North Carolina likely needs an extra caveat. NCOBS works closely with local families on regional scholarships and held spots, often by phone or email, and some of those families may get a direct application link and never return to the catalog. If so, some NC enrollments have less catalog traffic behind them, which would push NC's index up. The enrollment data fits this but doesn't prove it: 88 of North Carolina's 151 students in 2026 went to NCOBS, and 21% of NCOBS students live in North Carolina. Among the other Schools, only OBCA (28% from California, a much larger state) draws a higher share from its home state.
Virginia's catalog traffic converts near average (index 96), but on the youth courses page it logs only 0.55 engaged sessions per user against about 0.85 elsewhere, with 75% new users. Its index drops to 67 when youth pages are the denominator. The city data now confirms the cause: Ashburn, Reston, Leesburg and Sterling (Northern Virginia's data-center corridor) account for about 900 catalog users, with lower engagement and more new users than real metros. Treat Virginia's traffic as overstated.
Florida is on neither list because it enrolls almost exactly what its traffic predicts: 94 students against 95 expected (index 99). It has held at 78–105 students a year since 2021. South Florida (Miami, Fort Lauderdale, Palm Beach) supplies 26–43 a year, with Tampa Bay, Orlando, Southwest Florida and Jacksonville at roughly 6–20 each. The gap in Florida is reach: 22 catalog users per 100,000 residents against a national average of about 29. When Florida families find the catalog, they enroll at the normal rate. Not enough of them are finding it.
The 22 states that under-index in both 2026 and 2023–25 enrolled 194 students against about 433 their traffic share predicts. Because expected enrollments are a share of a fixed total, this gap is mirrored by the over-converting states. It shows where demand isn't converting, not enrollments that went missing. The gap is statistically clear in 16 of the 22 states; the rest could be noise, and some of it reflects real distance from course areas.
Splitting markets by whether they have a reach gap (too few visitors) or a conversion gap (visitors who don't enroll) points to different kinds of tests. Reach is catalog users per 100,000 residents (2024 Census estimates), national average about 29. Reach and index are statewide; the named metros are where to focus within each state.
These markets convert at or above the network rate but send few visitors for their size. The test is whether more targeted traffic brings them up to baseline enrollment.
| Market | Reach | Index |
|---|---|---|
| Florida: South Florida, Tampa Bay, Orlando | 22 | 99 |
| Texas: Houston and Austin | 15 | 91* |
| New Jersey | 19 | 146 |
*Statewide index. Houston and Austin enroll at or above the metro average; Dallas–Fort Worth does not, so it is left out of the test. Some New Jersey visitors may register in GA4 as New York, which would understate New Jersey's reach.
These markets already send above-average traffic but enroll well below their share. The test is better conversion from the traffic that already exists, through School-specific follow-up, retargeting, info sessions or offers, rather than buying more traffic.
| Market | Reach | Index |
|---|---|---|
| Washington: Seattle | 39 | 61 |
| Minnesota: Twin Cities | 44 | 57 |
| Georgia: Atlanta | 32 | 62 |
All three gaps are statistically clear (p<0.01). Minnesota and Washington are 2026 declines from stronger history, so they are the most recoverable.
Active users summed by page and region (a person who views two pages counts twice), all geographies, H1 2026. Read the share column; totals are page-level, not unique people.
| Page group | Page-level users | Share | Engaged sessions / user |
|---|
GA4 catalog users by city (Jan 1 – Oct 12, 2026) grouped into 20 metros, against enrollments from the same metros by home ZIP code. GA4 assigns many suburban visitors to small towns, so these 20 metros hold 25% of catalog users but 56% of enrollments. The index is therefore relative to the other metros (100 = average for this group), not to the state scale above. Read it for ranking and direction.
| Metro | Catalog users | Enrolled 2026 | Avg 2023–25 | Relative index 2026 | Relative index 2023–25 |
|---|---|---|---|---|---|
| Minneapolis-St Paul | 1,495 | 13 | 20 | 30 | 41 |
| Dallas-Fort Worth | 1,234 | 14 | 24 | 39 | 58 |
| Seattle | 1,667 | 22 | 51 | 46 | 92 |
| Phoenix | 1,012 | 14 | 16 | 48 | 49 |
| Portland OR | 1,137 | 20 | 30 | 61 | 80 |
| Baltimore | 894 | 16 | 22 | 62 | 75 |
| Atlanta | 1,327 | 26 | 39 | 68 | 89 |
| Los Angeles | 3,243 | 64 | 71 | 68 | 66 |
| Raleigh-Durham | 1,469 | 32 | 39 | 75 | 79 |
| Chicago | 2,284 | 53 | 54 | 80 | 71 |
| Charlotte | 1,297 | 32 | 35 | 85 | 81 |
| Boston | 2,342 | 63 | 78 | 93 | 100 |
| Austin | 727 | 21 | 24 | 100 | 101 |
| Denver | 2,377 | 71 | 77 | 103 | 98 |
| Houston | 958 | 29 | 25 | 105 | 79 |
| New York | 4,990 | 208 | 212 | 144 | 128 |
| SF Bay Area | 2,904 | 123 | 147 | 147 | 153 |
| Asheville | 669 | 32 | 24 | 166 | 110 |
| Washington DC | 1,497 | 80 | 98 | 185 | 198 |
| Philadelphia | 1,093 | 67 | 62 | 212 | 170 |
Minneapolis–St Paul, Dallas–Fort Worth, Seattle, Phoenix and Portland trail, with Atlanta close behind; Seattle and Minneapolis fell sharply from their 2023–25 levels. Philadelphia, Washington DC, New York and the Bay Area lead. Portland includes Portland, Maine, which GA4 does not separate.
Internal NCOBS traffic. Asheville, Black Mountain, Candler and Woodfin log 15 to 40 views per user against about 5 everywhere else, consistent with staff use. It is only about 500 users, so the North Carolina index (which counts users, not views) barely moves, but any view-based report for NC is inflated. Consider a GA4 internal-traffic filter.
Data centers. Northern Virginia (see Virginia above) and Prineville, Oregon (180 users, 92% new, 24% engagement) look like non-human traffic.
Unlocated. 4.6% of catalog users have no city ("not set").
Click a column to sort. Index = enrollment share ÷ traffic share × 100. p is a two-sided Poisson test of actual vs. expected; below 0.05 the gap is unlikely to be chance.
| State | Catalog users | All program users | Enrolled 2026 | Expected | Gap | Index (catalog) | Index (all pages) | Index 2023–25 | Enr. / 1K users | p | Top school |
|---|