Estimated Incremental Applications

Definition

How many extra applications did paid digital media help generate — above what we would have gotten without it?

In every year we can fully measure, students who saw our paid media applied at roughly two to three times the rate of students who did not. This page estimates how many of those applications paid media is responsible for.

Tip

The bottom line: Paid media adds hundreds of applications a year. If we turned it off, we would lose them. This is an estimate of impact, not a lab-proven cause — but the pattern is consistent and strong across every year.

Summary

Paid media’s contribution grew every year as investment scaled — from an estimated 159 extra applications in FY2023 to 559 in FY2025.

Fiscal Year	Students Who Saw Paid	Their Apply Rate	Everyone Else’s Rate	Extra Applications
2023	1,557	19.5%	9.3%	159
2024	1,734	36.5%	13.6%	397
2025	2,711	28.2%	7.6%	559
2026 (in progress)	4,203	17.9% (see note)	5.5%	522*

Note

Why FY2026’s rate looks lower — and why that’s temporary. In FY2026 we added a new performance vendor, EDDY, running paid search and paid social. Their campaign started late (first leads in mid-October, their landing site in January), so most of their students are only weeks or months into a decision that typically takes months to make. They simply haven’t had time to apply yet. Set the young EDDY students aside, and FY2026 paid media is converting at 33.4% — our best year ever. The blended 17.9% will climb as those recent students move through their decision window. EDDY’s own numbers already show this: their older students apply at 14.3%, their newest at 3.8%. Same students, more time.

Key Insight

Paid media works, and it’s working better than ever. Every year we can fully measure, paid-exposed students apply at 2–3x the rate of everyone else — adding hundreds of applications a year, up to 559 in FY2025. FY2026 looks softer at a glance only because a new vendor’s students joined too recently to have applied yet. On an apples-to-apples basis, FY2026 paid media is our strongest year. Don’t judge the newest investment on a half-finished scoreboard.

Why This Is Important

This metric translates paid media into the language that matters to a board: applications, not impressions or clicks. It answers one question directly — how many applications would we lose if we stopped spending?

Two things make the case stronger over time:

• The gap is widening. Students who see paid media apply at 2–3x the rate of those who don’t, and that multiple has grown as the general apply rate softened under the new application fee.
• The newest dollars are still maturing. The FY2026 EDDY investment hasn’t had time to show its full return. As those students decide, both the paid apply rate and the estimated extra applications should rise.

Natural next steps for this metric:

• Cost per extra application
• Cost per extra enrollment
• Extra net tuition revenue generated

Methodology

Each person is classified as paid_exposed or not_exposed based on whether they have a paid_exposed tag in person_tags. Persons are grouped by fiscal year based on their created_at date using the July 1 cutoff.

Application rates are calculated for each group per fiscal year. The incremental rate is the difference between the paid and non-paid rates. Estimated incremental applications is that incremental rate applied to the paid applicant count:

estimated_incremental_applications = paid_applicants × (incremental_rate / paid_rate)

This formula asks: of the paid applicants we observed, how many applied at a rate above the non-paid baseline? It does not assume all paid applicants were influenced — only the portion above what the background rate would predict.

Caution

This analysis includes all persons regardless of touch history. The non-paid group contains list imports and low-intent prospects with no recorded engagement, which suppresses the non-paid rate and inflates the lift ratio. See the paid media time-to-application and first-touch application rate pages for comparisons that control for this.

Query

CREATE OR REPLACE TABLE paid_exposure_application_lift AS
WITH paid AS (
  SELECT
    person_id,
    MIN(tag_date) AS first_paid_at
  FROM person_tags
  WHERE tag_type = 'paid_exposed'
  GROUP BY person_id
),

applied AS (
  SELECT
    person_id,
    MIN(tag_date) AS applied_at
  FROM person_tags
  WHERE tag_type = 'applied'
      AND tag_value = 'Yes'
  GROUP BY person_id
),

people AS (
  SELECT
    pd.person_id,
    pd.fiscal_year,

    CASE
      WHEN paid.person_id IS NOT NULL THEN 'paid_exposed'
      ELSE 'not_exposed'
    END AS exposure_group,

    applied.applied_at

  FROM persons_dashboard pd

  LEFT JOIN paid
    ON pd.person_id = paid.person_id

  LEFT JOIN applied
    ON pd.person_id = applied.person_id
),

aggregated AS (
  SELECT
    fiscal_year,
    exposure_group,

    COUNT(DISTINCT person_id) AS people,

    COUNT(DISTINCT CASE
      WHEN applied_at IS NOT NULL
      THEN person_id
    END) AS applicants,

    COUNT(DISTINCT CASE
      WHEN applied_at IS NOT NULL
      THEN person_id
    END) * 1.0

    /

    COUNT(DISTINCT person_id) AS application_rate

  FROM people

  GROUP BY 1,2
),

pivoted AS (
  SELECT
    fiscal_year,

    MAX(CASE
      WHEN exposure_group = 'paid_exposed'
      THEN people
    END) AS paid_people,

    MAX(CASE
      WHEN exposure_group = 'paid_exposed'
      THEN applicants
    END) AS paid_applicants,

    MAX(CASE
      WHEN exposure_group = 'paid_exposed'
      THEN application_rate
    END) AS paid_rate,

    MAX(CASE
      WHEN exposure_group = 'not_exposed'
      THEN people
    END) AS non_paid_people,

    MAX(CASE
      WHEN exposure_group = 'not_exposed'
      THEN applicants
    END) AS non_paid_applicants,

    MAX(CASE
      WHEN exposure_group = 'not_exposed'
      THEN application_rate
    END) AS non_paid_rate

  FROM aggregated

  GROUP BY fiscal_year
)

SELECT
  fiscal_year,

  paid_people,
  paid_applicants,
  ROUND(paid_rate, 3) AS paid_rate,

  non_paid_people,
  non_paid_applicants,
  ROUND(non_paid_rate, 3) AS non_paid_rate,

  ROUND(
    paid_rate / NULLIF(non_paid_rate, 0),
    2
  ) AS lift_ratio,

  ROUND(
    paid_rate - non_paid_rate,
    3
  ) AS incremental_rate,

  ROUND(
    paid_applicants
      * (
          (paid_rate - non_paid_rate)
          / NULLIF(paid_rate, 0)
        ),
    0
  ) AS estimated_incremental_applications

FROM pivoted
WHERE fiscal_year::INT IN(2023, 2024, 2025, 2026)
ORDER BY fiscal_year;