A Case Study

Closing the loop
before the send.

How Foundry DST turned a six-part email drip into a measurable, pre-send optimization loop for a 77-county Oklahoma audience.

The problem

Send first. Learn later.

The traditional email drip workflow is backwards. Draft. Send. Read open rates. Guess what went wrong. Adjust the next touch.

By the time a report lands, the email has already been delivered to a list that didn't want it. For a 77-county audience with real cultural fault lines, the cost of that guess is the campaign itself — deliverability, brand trust, and permission to re-engage.

DRAFT
SEND
READ REPORTS
GUESS

Four steps. Zero foresight. The list learns the lesson.

The intervention

A four-step loop before a single email goes out.

Steps 2 and 4 are the same Foundry run. The only thing that changes between them is the email.

01

Draft

Start with the original email as written.

02

Predict

Run it through Foundry DST for county-level sentiment — all 77 of them.

03

Rewrite

Rewrite against specific report guidance, county by county.

04

Revalidate

Re-run the new version. Ship only after sentiment improves.

The result

253 counties moved.

Across six emails, the rewrite loop moved Oklahoma sentiment by every measure we track.

+253
Receptive county-reads
29 before. 282 after.
−177
Resistant county-reads
187 before. 10 after.
6 / 6
Emails improved
Every email gained ground.
Ensemble confidence held or rose in every case.
Every email improved

Three emails had zero receptive counties. None do now.

Receptive Mixed Resistant
Side by side

The emails, before and after.

For each of the six emails: Foundry's pre-send sentiment shift, plus the condensed reasoning behind the rewrite. Email screenshots removed for this demo.

Beyond copy

What open rates would never have told you.

Foundry doesn't only rewrite emails. It surfaces targeting, list, and news-cycle risks a traditional post-send report cannot detect — because by the time the email has sent, the damage is done.

Suppress, don't send

Ten county-reads remain resistant after the rewrite. Foundry recommends suppression or no-sell alternatives for frontier ranching and high-poverty eastern counties — not a copy change.

Bilingual, not bolt-on

Five counties — Texas, Beaver, Harper, Kingfisher, Tillman — have Hispanic populations above 8%. Foundry flagged full bilingual versions; adding a Spanish-language contact line is the minimum.

News-cycle timing

The Oklahoma Watch “Invest in Oklahoma” story was active during the send window. Foundry flagged this as a radical-transparency send — named local contacts, explicit opt-outs, no opaque language.

Productize the loop

This is a repeatable service, not a one-off engagement.

The same loop applies to any Oklahoma-audience communication — policy, advocacy, fundraising, member outreach, crisis response. Plug in a draft. Measure sentiment. Rewrite with report-specific guidance. Revalidate. Ship.

6
Emails proven in
one campaign
12
Foundry runs,
before and after
77
Counties read
every time
Next moves

Where this goes next.

01

Close Email 6's residuals.

Apply Foundry's suppression, bilingual, and tribal-county recommendations to the final send list before launch.

02

Package Foundry Pre-Send.

Formalize the loop as a productized service. Price by email volume and audience complexity. Lead with this campaign as the anchor case.

03

Export to adjacent sectors.

The same loop applies in tribal relations, public affairs, healthcare member communications. Start where Saxum already has relationships.

04

Close the prediction-to-reality loop.

Measure real post-send engagement against the pre-send prediction. Every campaign becomes training data for the next.

Predict before you send.
Rewrite before you burn the list.

Make the unshakeable — unstoppable.