The honest answer: the 2026 season does not show that the program lifted dollars overall. It does show one real bright spot — organizations in their second year of individual campaigns held onto donors measurably better than everyone else — and one big illusion we must not take credit for: the public-media surge.
Every claim on this page is backed by the numbers below, and the interactive charts let you check the two things that matter most: what happens when you set public media aside, and the difference between the big-org totals and the typical organization's experience.
Among organizations on the platform both seasons, dollars fell 4.2% and donors fell 8.6%. Alaskans applied for the 2026 dividend right after receiving a $1,000 PFD — 41% smaller than the year before, the smallest ever after inflation. That headwind hit everyone roughly equally, which is exactly why we compare participants against non-participants instead of against zero.
Tier 1 dollars rose 1.9% while non-participants fell 5.9%. But Alaska Public Media alone gained $39,650 — 3.5× the entire Tier 1 net gain. After Congress ended CPB funding in July 2025, every public station surged — stations we never advertised for actually grew more (+30.3%) than the ones we did (+20%). Flip the "set aside public media" switch below and watch the Tier 1 advantage disappear.
The 11 organizations in their second-plus year of individual campaigns beat comparable non-participants on donor counts by about +9.5% — and that holds (+6.9%) even with public media set aside. It's the only result in the whole analysis that survives every stress test. With only 11 organizations it's "promising evidence," not proof — but it's consistent with what the program is designed to do: re-engage known donors.
The 125 Tier 2 organizations tracked non-participants almost exactly — on dollars and on donors, in every way we sliced it. Generic category creative aimed at pooled donor lists is not producing detectable donations. That's a design conclusion, not a small-sample fluke: 125 organizations is a solid test.
Running the identical analysis on the 2025 season (when all 73 participants got individual campaigns), participants show no advantage over non-participants. Two seasons, one consistent story: the first year of a campaign does little we can measure; the donor-retention payoff shows up in year two.
You don't need a statistics background for this page. Every technical idea is explained in a box like this one right where you need it, every chart has a plain-English readout under it, and anything scientific you'd want to double-check lives in "Fine print" at the bottom.
On the surface, 2026 looks like a growth year: total pledges rose 5.1%. That number is misleading, and the reason matters for every comparison on this page.
Organizations present in both seasons gave less in 2026 (−$100K). All of the growth — $266,700 — came from 116 organizations that (re)joined the platform, minus $41,048 from 34 that left. Verified against the export to the penny.
Alaskans pledge PCG donations while applying for the next dividend, so the check they just cashed sets the mood. The October 2025 payment was $1,000 — the smallest inflation-adjusted dividend in state history.
PCG changed software platforms before the 2025 season and 100+ organizations missed the registration deadline — the platform shrank from 611 organizations to 548, then rebounded to 630 in 2026. Comparing raw season totals mixes real giving changes with organizations simply reappearing. So every year-over-year number on this page counts only organizations present in both seasons, and the returners are reported separately (see "Outside events"). One side effect we can't fully remove: donors whose organization vanished in 2025 may have redirected that gift to orgs that remained, slightly inflating 2025 baselines for everyone else.
The program's three populations, year over year: Tier 1 (22 orgs, individual campaigns), Tier 2 (143 orgs, grouped category campaigns), and non-participants (the comparison group). Use the controls — especially the red switch.
"Group total" adds up everyone's dollars, so a couple of large organizations can carry the whole number. "Typical org" is the median — line all the orgs up by their percent change and take the middle one. When the two disagree, the total is telling you about the big players and the median is telling you what the program felt like for most participants. Tier 1's total is up while its typical org is down — that gap is the Alaska Public Media story.
"Flat" means within ±2%. Even inside Tier 1, 11 of 21 organizations declined — the group's positive total rests on a few large gainers, which is why the median matters.
Comparing participants to non-participants over the same window is called a difference-in-differences estimate. It's the fairest question we can ask of this data: how did participants do relative to what we'd expect, given how everyone else moved?
Each dot is our best estimate of the program's effect for that group — the amount by which participants beat (or trailed) comparable non-participants. The line through each dot is the range of doubt (a 95% confidence interval): the values the true effect could plausibly be, given how few organizations we have. If the line crosses zero, we cannot rule out "no effect." Solid dots = the result survives; hollow dots = inconclusive. Estimates use the equal-weighted method, so one giant org can't dominate; comparisons are against organizations never in the program.
Repeat Tier 1 participants held onto donors. The 11 organizations in year two-plus beat never-participating peers on donor counts by +9.5% (range +2.6 to +17.5) — and the result stands at +6.9% (range +0.8 to +13.6) with public media set aside. Eight of the 11 beat the comparison group's donor baseline: it's broad, not one lucky org.
Honest caveats: 11 organizations is a small group; they chose to come back (organizations that return may differ from ones that don't); and in their own first year they showed no advantage — so "the effect compounds" is our best interpretation, not a proven mechanism.
Tier 1 dollars overall: +7.4 points ahead of the comparison group — but the range of doubt spans zero, and setting aside public media drops it to +0.5 points. Not claimable.
Tier 1 first-timers: slightly behind the comparison group in their first season, mirroring what the 2025 cohort did in its first year.
Tier 2, every slice: statistically indistinguishable from no effect — repeat or first-time, dollars or donors.
Each of these checks has caught a real problem in a previous year's analysis, so all four were run. Every one is expandable — nothing is hidden.
34 current non-participants were in the program in 2024 or 2025. Former participants aren't a clean "untouched" comparison — and indeed they performed slightly worse than never-treated organizations (−7.2% vs −5.5% dollars), which would have flattered our results by about 0.4 points had we left them in. All headline numbers use the clean, never-participated comparison group. Watch-item: Fellowship in Serving Humanity fell 41% the year after its campaigns ended. If a post-program drop-off is real, it will matter more as our alumni pool grows.
A fair comparison requires that participants would have tracked non-participants had we done nothing. For first-time 2026 participants we can check the year before they joined: Tier 1 first-timers moved almost identically to the comparison group in 2024→2025 (within half a point on dollars) — the comparison is credible for them. Tier 2 first-timers are mixed: their group totals tracked the comparison group, but the median Tier 2 first-timer was outperforming by about 10 points before joining. If anything, that means Tier 2's "no effect" verdict is generous — organizations carrying positive momentum into the program still didn't beat the comparison group. For repeat participants no pre-program check is possible (they were already in the program the year before); their result rests on the assumption they'd otherwise have tracked the comparison group.
Applying the identical method to 2024→2025 (all 73 participants received individual campaigns): participants trailed the comparison group slightly on dollars (−1.1 points on totals; equal-weighted −2.0%, range −8.0 to +4.4) and on donors (−2.7 points; −2.5%, range −7.0 to +2.5). Nothing distinguishable from zero. Consistency across windows: no broad average effect in either season; 2026's apparent lift is concentrated in public media (external) and repeat participants (plausibly real).
Tier 1 2026: 11 repeat organizations (also in the 2024 pilot or 2025 program), 11 first-timers. Tier 2: 34 repeat, 109 first-time. The entire Tier 1 advantage sits in the repeat group (+18.7 points on dollar totals, +14.6 on donors vs the clean comparison; equal-weighted +12.6% and +9.5%, both ranges clear of zero). First-timers: −4.5 points dollars, −3.1 donors — indistinguishable from zero. Full honesty: the same 11 repeat organizations did not outperform during their own first year (2024→2025 equal-weighted −4.9% on dollars), so the year-two payoff could be compounding — or could partly reflect which organizations chose to return.
All 21 Tier 1 organizations with both seasons of data. Sort any column, search, and hover the bars. Net Tier 1 change: +$11,252 — and one bar below is 3.5× that entire net gain.
Badges: Repeat = also in the 2024 pilot or 2025 program · New = first year · Public media = affected by the CPB funding crisis (see "Outside events"). Planned Parenthood (22nd Tier 1 org) has no 2025 baseline — it sat out 2025 during the platform transition — so it can't appear in year-over-year math; it returned at $50,875 / 652 donors, well below its 2024 level. See "Outside events."
KUAC (Fairbanks public media) fell −26% — but only because it spiked +73% in 2025 ($20,150 → $34,800), a year before the statewide public-media surge. Its 2026 total is still +27% above 2024. Read it as a return to earth, not donor loss.
Largest Tier 1 organization after APM, first year in the program, and it shed 348 donors (−21%) while never-participating domestic-violence peers held steady (Kodiak Women's Resource +6.5%, AK Native Women's Resource Center +0.9%, Tundra Women's Coalition +13.5%). Despite the federal VOCA funding crisis (AWAIC is ~71% federally funded), no donor surge materialized for DV organizations in PCG. Something org-specific happened — list quality, creative fit, or their donors moving to other channels. Worth asking before next season.
Tier 2 exists because ad platforms raised minimum custom-audience sizes above what small nonprofits could meet individually — so donor lists were pooled into 24 thematic categories served category-generic creative. Every campaign ran as planned, with budgets scaled to list size. Here's every category's year-over-year dollars, against the platform's −4.2%.
The two categories that beat the platform meaningfully are the confounded one (Public Media — the CPB crisis, not our creative) and two single-organization categories where one org is the whole "category" (Civic Engagement = Alaska Black Caucus alone; Housing = one org). 18 of 24 categories declined. Nothing here suggests grouped generic creative moved donors — see "Cause & effect" for the formal test, and "Recommendations" for what we'd change.
Before crediting (or blaming) the program, we checked every external force that could move whole sectors — because if an outside event lifts a sector that happens to sit in one tier, the program looks better or worse than it is.
Congress voted on July 17, 2025 to claw back all federal public-media funding — roughly $1B nationally, about $15M/year to Alaska's 27 stations. A statewide emergency campaign (the Alaska Community Foundation's "Voices Across Alaska" fund) raised $3.5M+ by September, and the story saturated Alaska news through the PCG season. Donors responded everywhere:
The October–November 2025 federal shutdown delayed SNAP benefits; Alaska declared a disaster and food banks saw record demand and record giving — in November–December, before the PCG window. In PCG data, food-security orgs were only mildly resilient (Tier 2 food category −3.3%; our two Tier 1 food banks −2.4% and −4.7%), and never-participating pantries scattered in both directions. The crisis wave was spent before our season; no correction needed.
Federal VOCA victim-services funding fell ~40% and Alaska DV nonprofits publicly feared closures. No PCG donor surge followed: never-participating DV orgs were flat-to-slightly-up, while our two Tier 1 DV orgs (AWAIC, STAR) fell 16.5% and 12.3%. This sector pulled Tier 1's numbers down, not up.
Planned Parenthood (Tier 1) has no 2025 baseline and is excluded from every estimate; its return at −34% vs 2024 (against a platform-wide −5.2% over the same two years) reflects the post-2024 normalization of reproductive-rights giving, not the campaign. Others moved in both directions — ACLU of Alaska dollars +7.5% but donors −12%; Identity (LGBTQ+) −38%; pregnancy resource centers −17% to −35%; Alaska Black Caucus +29.6% — volatile, but no one-directional bias to correct.
The 100+ organizations that missed 2025 registration mean 2026's "growth" is mostly re-entry ($266,700 from 116 joiners: Planned Parenthood $50,875; 18 Tier 2 orgs totaling $47,975, led by The Door at $14,075 and ANDVSA at $6,325; 97 non-participants, $167,850). The 34 leavers took just $41,048 — the only formerly-treated one being Alaska State Firefighters Association ($1,300 → zero). Both-seasons-only math removes most of this; a small inflation of 2025 baselines (redirected gifts) remains and affects all groups roughly equally.
| Outside force | What it did | Cancelled out by our comparison? |
|---|---|---|
| PFD $1,702 → $1,000 (−41%) | Depressed all giving | Yes — hit everyone; explains the season's level, not group gaps |
| 2025 platform gap (100+ orgs missed deadline) | Distorts baselines and totals | Mostly — both-seasons rule handles it; small residual, roughly even |
| CPB defunding / public-media surge | Sector-wide giving spike, landed in both tiers | No — handled by reporting everything with & without public media |
| Shutdown / SNAP crisis (food security) | Crisis giving in Nov–Dec, before the season | Not in principle — but measured effect in-window was negligible |
| VOCA cuts (DV / victim services) | No PCG donor response detected | Not in principle — but no surge materialized; if anything hurt Tier 1 |
| New 2026 federal charitable deduction ($1,000/$2,000 for non-itemizers; 0.5% AGI floor for itemizers) | Mixed nudges on giving; PCG pledges are 2026-tax-year gifts | Yes — roughly uniform across organizations |
| Inflation / tariffs / consumer sentiment | Depressed discretionary giving | Yes — hit everyone roughly equally |
| Reproductive-politics giving cycles | Org-specific swings (Planned Parenthood) | No — handled by flagging & excluding from aggregates |
In priority order. The first one protects the relationship; the next three shape next season; the rest strengthen how we measure.
The +1.9% Tier 1 dollars figure is tempting to headline, but it does not survive removing Alaska Public Media, and PCG's own data shows never-participating stations grew more than ours. Saying it first is worth more than one season's talking point.
The one robust positive is donor retention in year two-plus (+7 to +9% vs comparison). Recruit 2027 around returning 2026 Tier 1 orgs, and set year-one expectations honestly with new participants: the first season shows little.
Whether the issue is generic creative or detection limits, the action is the same: don't expand as-is. Options: category creative that name-checks member organizations, pooled audiences serving org-specific creative in rotation, or reallocating budget toward fewer, deeper Tier 1 engagements.
Withhold ads from a random 10–20% slice of each organization's donor list. That yields clean donor-level effect estimates immune to sector shocks like this year's public-media surge — and settles the repeat-vs-new question definitively.
−348 donors against flat DV peers, in its first program year, as the second-largest Tier 1 org. If the list was stale or the creative missed, that's fixable; if donors migrated channels, we need to know.
Spend, impressions, and reach per organization and category. Budgets scale with list size, so dose-response analysis is currently impossible — delivery data separates "ads didn't work" from "ads didn't reach."
No 2025 baseline, polarizing-sector volatility, and a −34% two-year trajectory that would distort any group total it joins. Evaluate it separately in all reporting.
Former participants underperformed never-participants by ~1.7 points, and Fellowship in Serving Humanity fell 41% after treatment ended. One season isn't proof; two would be. If ads pull donations forward, sunsetting engagements needs a plan.
Everything a skeptical reader (or a statistician the client hires) would want to check.
The season arrives as a pledge-level export (25,386 rows). We consolidated it to organization level, counting both identified donors (rows with an ALN/PersonId, counted by pledge — 39 rows carry two pledges) and anonymous donors (summary rows with a pledge count but no ALN). Rows carrying only an "Anonymous:" grand-total label were excluded from counting but used as a built-in answer key. Anonymous donors are 36.8% of contributors and 32.5% of dollars — skipping them would have badly understated results.
Verification, three ways: (1) every one of the 630 organization blocks reconciles exactly against its own embedded grand-total row — anonymous count, non-anonymous count, and dollars; (2) the file-level total matches the export's own grand-total row to the penny ($2,571,175.00); (3) contributor totals reconcile: 13,963 anonymous + 23,963 non-anonymous = 37,926.
Names in program rosters routinely differ from names in PCG's contributions data — abbreviations, DBAs, renames, typos. Every roster entry was matched to a unique PCG vendor code and verified: match rate 100% on all four rosters (2024 pilot 10/10; 2025 participants 74/74; Tier 1 22/22; Tier 2 144 entries resolving to 143 unique organizations). The pre-existing overrides file was independently re-derived and confirmed — all 82 entries correct.
21 matches required documentary evidence rather than name similarity, including: Anchorage Ski Club = "Arctic Valley Ski Area" (DBA); BLBP = "Bright Lights Book Project"; Fairbanks Youth Advocates = "The Door"; KTOO = "KTOO KRNN KXLL 360TV Gavel Alaska"; Alaska Marine Conservation Council = "Alaska Marine Community Coalition" (renamed January 2026); Ester Village Library = "John Trigg Ester Library"; ACCA = "Alaska Center for Children and Adults"; North Star Council on Aging = "Fairbanks Senior Center Meals on Wheels" (same org across all three seasons' rosters); Palmer Food Bank = "Palmer Emergency Food and Community Services"; FISH = "Fellowship in Serving Humanity" (confirmed by program staff); "Literary Council of Alaska" = roster typo for Literacy Council of Alaska. One roster duplicate found and merged: "Lutheran School Association of Anchorage" and "Anchor Lutheran School" are the same organization listed twice in category 04.1 (one campaign ran). The full audit table lives in the project archive ("Org name-vendor code mapping (2026 analysis).csv").
| Group | n | $ 2025 | $ 2026 | Total %Δ$ | Median %Δ$ | Total %Δ donors | Median %Δ donors | Up/Flat/Down |
|---|---|---|---|---|---|---|---|---|
| Tier 1 | 21 | 585,373 | 596,625 | +1.9% | −2.4% | −4.3% | −6.2% | 6/4/11 |
| Tier 2 | 125 | 552,828 | 516,050 | −6.7% | −7.4% | −10.4% | −11.8% | 43/5/77 |
| Non-participants (all) | 368 | 1,266,732 | 1,191,800 | −5.9% | −9.6% | −9.6% | −11.4% | 123/18/227 |
| — never treated | 334 | 956,114 | 903,525 | −5.5% | −10.1% | −10.0% | −11.6% | 116/14/204 |
| — treated 2024/25 | 34 | 310,618 | 288,275 | −7.2% | −7.3% | −8.5% | −10.1% | 7/4/23 |
| All PCG (both years) | 514 | 2,404,932 | 2,304,475 | −4.2% | −8.1% | −8.6% | −11.3% | 172/27/315 |
Relative to the platform's −4.2%: Tier 1 +6.1 points above, Tier 2 −2.5 below, non-participants −1.7 below. Flat = within ±2%.
Three estimators: total (difference in group aggregate %Δ — dollar-weighted), median (difference of medians), and equal-weighted (mean of per-org log growth-ratio differences, back-converted to %; bootstrap 95% CI, 5,000 resamples). Equal-weighted is primary for inference. Comparison group: 334 never-treated organizations (324 excluding public media). Bold = CI excludes zero.
| Estimate (2025→2026 unless noted) | n | $ total | $ equal-wt [95% CI] | Donors total | Donors equal-wt [95% CI] |
|---|---|---|---|---|---|
| Tier 1 vs all non-participants (contaminated ctrl) | 21 | +7.8pp | +5.3% [−2.9,+14.5] | +5.3pp | +5.7% [+0.0,+11.7] |
| Tier 1 vs never-treated (clean ctrl) | 21 | +7.4pp | +5.4% [−3.2,+14.8] | +5.7pp | +5.7% [−0.2,+12.0] |
| Tier 2 vs all non-participants | 125 | −0.7pp | −1.3% [−8.7,+6.5] | −0.8pp | −3.8% [−9.2,+1.9] |
| Tier 2 vs never-treated | 125 | −1.2pp | −1.2% [−8.9,+6.9] | −0.5pp | −3.8% [−9.3,+2.0] |
| Tier 1 repeat participants | 11 | +18.7pp | +12.6% [+1.5,+25.2] | +14.6pp | +9.5% [+2.6,+17.5] |
| Tier 1 first-time participants | 10 | −4.5pp | −2.0% [−11.3,+7.8] | −3.1pp | +1.6% [−5.7,+8.6] |
| Tier 2 repeat | 32 | −2.8pp | −2.1% [−10.3,+6.4] | −1.0pp | −4.2% [−10.5,+2.4] |
| Tier 2 first-time | 93 | −0.3pp | −0.9% [−10.0,+8.8] | −0.2pp | −3.7% [−10.1,+3.3] |
| Tier 1 excluding public media | 18 | +0.5pp | +4.0% [−3.4,+11.8] | +0.5pp | +4.5% [−1.3,+10.4] |
| Tier 1 repeat excluding public media | 9 | +3.4pp | +7.1% [−1.8,+16.3] | +6.0pp | +6.9% [+0.8,+13.6] |
| Tier 2 excluding public media | 121 | −1.6pp | −1.5% [−9.2,+6.4] | −1.0pp | −4.2% [−9.5,+1.6] |
| 2025 participants, 2024→2025 window, vs never-treated | 72 | −1.1pp | −2.0% [−8.0,+4.4] | −2.7pp | −2.5% [−7.0,+2.5] |
| Pre-trend: T1 first-timers, 2024→2025 (untreated) | 9 | −1.0pp | −0.5% [−12.2,+16.0] | −3.5pp | −1.7% [−7.4,+5.6] |
| Pre-trend: T2 first-timers, 2024→2025 (untreated) | 79 | +0.3pp | −1.1% [−12.9,+10.9] | +3.9pp | +4.1% [−6.1,+14.9] |
| Leave-one-out: Tier 1 minus Alaska Public Media | 20 | −0.7pp | +3.8% [−4.2,+12.5] | +0.6pp | +4.5% [−0.9,+10.1] |
| Leave-one-out: T1 repeat minus Alaska Public Media | 10 | +5.6pp | +9.9% [−0.4,+22.4] | +6.3pp | +7.4% [+1.3,+14.2] |
| T1 repeat cohort in its first treated year (2024→2025) | 10 | +2.6pp | −4.9% [−12.5,+3.3] | +0.2pp | −4.4% [−9.6,+1.9] |
| Group | Stations (both seasons) | $ 2025 | $ 2026 | Total %Δ$ | Total %Δ donors |
|---|---|---|---|---|---|
| Tier 1 stations (Alaska Public Media, KNBA, KUAC) | 3 | 175,800 | 211,150 | +20.1% | +15.3% |
| Tier 2 stations (KBBI, KCAW, KNOM, KTOO) | 4 | 26,875 | 32,200 | +19.8% | +18.6% |
| Never-participated stations (Glacier City, KBRW, KDLL, KFSK, KHNS, KRBD, KSTK, KUCB, KYUK, Seward) | 10 | 22,675 | 29,550 | +30.3% | +17.2% |
| Formerly treated (Big Cabbage Radio) | 1 | 4,450 | 4,400 | −1.1% | −14.5% |
Not in the both-seasons math: KCHU (left after 2025), KMXT and KTNA (returned in 2026 after missing 2025), KNOM listed under Tier 2's Mental & Behavioral Health category but treated as public media for the confounder analysis. KUAC's decline is mean-reversion from a +73% spike in 2025.
The 2025 season report told the client participants were up (+4.13% pledges, +8.82% dollars) "while the program as a whole was down," concluding meaningful program value. This analysis finds the 2025 program was a null. Two artifacts explain the difference:
Handling: this year's client report uses final-vintage baselines and like-for-like comparisons and notes that figures will not match prior preliminary reporting. Do not reuse last year's +4.13%/+8.82% figures.