7  Who responded to the FY26 annual appeal?

The fall 2025 annual appeal was mailed to about 1,100 households — nearly all of them prior donors. A household responded if anyone in it gave a SUCCEEDED gift to the Annual Fund Drive - 2025-2026 campaign between October 1, 2025 and January 31, 2026. Giving and engagement are counted the same way as everywhere in this book: rolled up to the household, with engagement measured only for events that had happened before the appeal mailed (see Methodology).

import pandas as pd
import plotly.express as px
from hh.analytics.appeal import response_summary
from hh.analytics.donors import TIER_LABELS
from hh.io import read_parquet
from hh.viz import fmt_gift, money, pct, style, SEQUENTIAL_BLUE

t = read_parquet("processed", "appeal_households.parquet")
ap = t[t["appealed"]].copy()
n_universe = len(t)
n_appealed = len(ap)
n_responded = int(ap["responded"].sum())
resp_rate = n_responded / n_appealed
responder_gifts = ap.loc[ap["responded"], "afd_amount"]
resp_dollars = responder_gifts.sum()
median_gift = responder_gifts.median()
campaign_total = t["afd_amount"].sum()

not_appealed = t[~t["appealed"]]
n_not_appealed = len(not_appealed)
prior_donor_share_appealed = ap["prior_donor"].mean()
prior_donor_share_not = not_appealed["prior_donor"].mean()
arts_share_not = (not_appealed["eng_n_performance"] > 0).mean()
class_share_not = (not_appealed["eng_n_class"] > 0).mean()


def _fmt(raw: pd.DataFrame) -> pd.DataFrame:
    return raw.rename(
        columns={
            "households": "Households", "responded": "Gave", "rate": "Response rate",
            "dollars": "Total $", "avg_gift": "Avg gift", "median_gift": "Median gift",
        }
    ).assign(
        **{
            "Response rate": lambda d: d["Response rate"].map(pct),
            "Total $": lambda d: d["Total $"].map(money),
            "Avg gift": lambda d: fmt_gift(d["Avg gift"]),
            "Median gift": lambda d: fmt_gift(d["Median gift"]),
        }
    )


def display_summary(df, by, order=None, total=True):
    """response_summary for display, with a totals row (overall rate, responder-level
    avg/median — not averages of the group rows)."""
    raw = response_summary(df, by).set_index(by)
    if order is not None:
        raw = raw.reindex(order)
    parts = [_fmt(raw.reset_index())]
    if total:
        grand = response_summary(df.assign(**{"_total": "Total"}), "_total")
        parts.append(_fmt(grand.rename(columns={"_total": by})))
    return pd.concat(parts, ignore_index=True)

115 of 1,038 appealed households responded (11.1%), giving $34,495 of the $37,980 donated to the campaign during the window. The median responding household gave $100.

7.1 Response by prior giving

Prior giving is each household’s lifetime total before October 1, 2025, in the same four tiers as Are large donors going away?.

ap["prior tier"] = ap["prior_size_tier"].fillna("never gave")
tier_order = TIER_LABELS + ["never gave"]
by_tier = response_summary(ap, "prior tier").set_index("prior tier").reindex(tier_order)

fig = px.bar(by_tier.reset_index(), x="prior tier", y="rate")
fig.update_traces(
    texttemplate=[pct(r) for r in by_tier["rate"]], textposition="outside",
    marker_color=SEQUENTIAL_BLUE[7],
)
fig.update_layout(
    xaxis_title="Prior lifetime giving", yaxis_title="Share of households that responded",
    yaxis_tickformat=".0%",
)
style(fig).show()

Figure 7.1: Response rate by prior-lifetime-giving tier

display_summary(ap, "prior tier", order=tier_order)

	prior tier	Households	Gave	Total $	Avg gift	Median gift	Response rate
0	<$100	534	11	$1,449	$132	$50	2.1%
1	$100–999	330	39	$3,532	$91	$75	11.8%
2	$1,000–4,999	105	43	$8,127	$189	$104	41.0%
3	$5,000+	38	19	$21,252	$1,119	$500	50.0%
4	never gave	31	3	$135	$45	$25	9.7%
5	Total	1038	115	$34,495	$300	$100	11.1%

7.2 Response by giving recency

# a true partition of the appealed households (lapsed donors are a subset of
# "not in FY25", so the middle row excludes them to keep the groups disjoint)
groups = [
    ("gave in FY25 (year before appeal)", ap["prior_fy_donor"].fillna(False)),
    (
        "gave within 24 months, before FY25",
        ap["prior_donor"] & ~ap["prior_fy_donor"].fillna(False) & ~ap["lapsed_donor"].fillna(False),
    ),
    ("lapsed 24+ months", ap["lapsed_donor"].fillna(False)),
    ("never gave", ~ap["prior_donor"].fillna(False)),
]
status = pd.concat([ap[m].assign(**{"giving status": label}) for label, m in groups])
display_summary(status, "giving status", order=[g[0] for g in groups])

	giving status	Households	Gave	Total $	Avg gift	Median gift	Response rate
0	gave in FY25 (year before appeal)	315	91	$30,983	$340	$100	28.9%
1	gave within 24 months, before FY25	145	8	$1,200	$150	$100	5.5%
2	lapsed 24+ months	547	13	$2,176	$167	$100	2.4%
3	never gave	31	3	$135	$45	$25	9.7%
4	Total	1038	115	$34,495	$300	$100	11.1%

7.3 Response by engagement before the appeal

Engagement profiles count only events that started by September 30, 2025 — registering in September for a November show does not count.

profile_order = ["arts+class", "arts", "class", "community/other", "none"]
display_summary(ap, "eng_profile", order=profile_order)

	eng_profile	Households	Gave	Total $	Avg gift	Median gift	Response rate
0	arts+class	360	52	$19,839	$382	$100	14.4%
1	arts	367	48	$12,127	$253	$100	13.1%
2	class	190	3	$700	$233	$100	1.6%
3	community/other	12	1	$25	$25	$25	8.3%
4	none	109	11	$1,804	$164	$100	10.1%
5	Total	1038	115	$34,495	$300	$100	11.1%

7.4 Response by distance from the Hall

BAND_LABELS = {
    "[0, 1)": "0–1 mi", "[1, 5)": "1–5 mi", "[5, 10)": "5–10 mi",
    "[10, 20)": "10–20 mi", "[20, ∞)": "20+ mi",
}
ap["miles"] = ap["distance_band"].map(BAND_LABELS)
display_summary(ap, "miles", order=list(BAND_LABELS.values()))

	miles	Households	Gave	Total $	Avg gift	Median gift	Response rate
0	0–1 mi	140	18	$2,752	$153	$100	12.9%
1	1–5 mi	156	25	$11,643	$466	$200	16.0%
2	5–10 mi	237	29	$8,296	$286	$100	12.2%
3	10–20 mi	195	19	$3,257	$171	$100	9.7%
4	20+ mi	290	21	$8,173	$389	$104	7.2%
5	Total	1038	115	$34,495	$300	$100	11.1%

7.5 Households that were not appealed

Of the 4,654 households in the database, 3,616 were not appealed. The mailing list was overwhelmingly the donor file: 97.0% of appealed households had given before, versus 5.0% of those not appealed. Among the not-appealed, 47.9% had attended performances and 34.3% had taken classes before the appeal.