Building a Programmatic Outreach Machine for 254 Texas Counties

We needed authoritative backlinks for a Texas property tax exemption site. The conventional approach would have been the same one most SEOs use: write a few guest posts, pitch some blogger roundups, maybe buy into a link network and hope Google doesn’t notice. Standard playbook, mediocre results.

Instead, we built a programmatic outreach pipeline that covered all 254 Texas counties, generated personalized email drafts in batches, and earned .edu backlinks from Texas A&M AgriLife Extension offices and county-level sources. Total human effort per email: zero. Total cost beyond our existing n8n server: about $20/month in API calls.

This pipeline is built on the same n8n marketing automation engine I use for all client work, connected to a self-hosted NocoDB instance that serves as our contact database. The architecture is the same as the lead routing workflow in that earlier post — webhook trigger, data enrichment, conditional branching, output — just applied to a different problem.

The Problem That a Spreadsheet Cant Solve

TexasLandTax.com provides property tax exemption information for every county in Texas. The site is useful. But useful content does not automatically rank. You need authority signals, and in the property tax space, the strongest signals come from .edu domains and local government sources.

Texas A&M AgriLife Extension operates a network of county-level offices across the state. Every county has an extension agent who publishes local agricultural and economic content. These offices link to external resources all the time. They are credible, they are local, and they are almost never approached for backlinks.

The manual approach would look like this: research 254 counties, find the right contact for each one, draft 254 unique emails, send them one at a time, track responses in a spreadsheet, follow up individually. That is roughly 40 hours of work before you have sent a single email. And most of those hours go into research and drafting, not the part that actually matters, which is personalization and follow-through.

We needed a different model.

The Architecture: NocoDB as a Contact Database, n8n as the Engine

The core insight was that all 254 counties follow the same pattern. Every one has an AgriLife Extension office. Every office has a website with contact information. The county names and agent names changed, but the outreach structure was identical. That is exactly the kind of problem n8n is built for.

We seeded a NocoDB table with all 254 county rows: county name, AgriLife Extension office URL, county seat. This data came from the Texas A&M AgriLife Extension directory, which is structured enough to scrape programmatically.

The n8n pipeline breaks down into four stages:

Stage 1 - Contact Discovery: An HTTP Request node fetches each county’s AgriLife Extension page. A Function node extracts the agent name and email from the page HTML. If the pattern fails (some counties use different page structures), the record gets flagged for manual review instead of skipped silently.

Stage 2 - Personalization: A Function node assembles a personalized email body. It inserts the county name, agent name, and a specific reference to the county’s property tax situation. The prompt is structured as a template with three variable slots. This is not AI-written content. It is a deterministic template with the right variables filled in. The tone is direct and helpful. We are offering a resource that answers questions their residents already ask.

Stage 3 - Gmail Draft Creation: Each personalized email becomes a Gmail draft via the Google Workspace API. We batch them in groups of 20 to stay under rate limits. The subject line includes the county name so the agent instantly knows the email is not mass spam.

Stage 4 - Review + Send Queue: Drafts land in a shared Gmail inbox labeled by county region. The human review step is a quick glance to catch any template rendering issues, then one click to send.

Here is the part that surprised me. We expected the hardest part to be personalization or deliverability. It was neither. The hardest part was the variability in county websites. Collin County structures its extension page one way. Tarrant County does it completely differently. Denton County buries the agent contact three links deep. The first week of development was entirely edge case handling for these differences.

What Happened When We Hit Send

The results were not instant. Backlink outreach never is. But the pipeline did what it was designed to do. It removed the bottleneck of manual drafting so we could send personalized outreach to all 254 counties within 48 hours of the pipeline finishing.

Within the first 30 days, we earned links from 11 .edu domains, including several AgriLife Extension office sites. The emails that performed best were not the ones with the most detailed personalization. They were the ones that led with a specific county-level data point. “I see your office published a guide on agricultural tax exemptions in 2023. Our page on Dallas County property tax exemptions covers the residential side. Links between the two would serve both audiences well.”

We tracked open rates, response rates, and link placements in a second NocoDB table. Open rates averaged 68 percent across all 254 counties. Response rate was lower, around 14 percent, but every response led to either a placed link or a referral to another office. This mirrors the reporting approach I covered in the 3 automations every marketing team needs post — track the right metric, let the data tell you what to adjust.

Why Programmatic Outreach Wins for Local SEO

Most SEO outreach treats local as a set of individual one-off projects. You email 50 bloggers, you email 10 journalists, you pitch 3 roundups. The effort scales linearly with the number of targets.

Programmatic outreach flips the model. You invest the upfront engineering time once. Then the cost per outreach target approaches zero. For a 254-county project, that makes the difference between a viable campaign and one that never ships because the manual workload looks insurmountable.

This approach works best when three conditions are met:

1. The target sites follow a predictable structure (same CMS, same page template, same contact pattern)
2. The value proposition is genuinely useful to the target (not a generic “please link to our content”)
3. You can fail noisily when the pattern breaks (the flagged-for-review system was worth its weight in debugging time)

Three Things I Would Do Differently

1. Add an Airtable or NocoDB column for “contact method preference.” Some county agents preferred email. Others ignored email entirely but responded to a contact form on their office site. We lost about two weeks in one region because we tunneled on email when the local agent only checks physical mail.

2. Automate the follow-up sequence. We sent one round of outreach, got good results, and stopped. A two-email follow-up sequence set to 10 days after the first send would likely have doubled our link count for negligible additional cost. The n8n pipeline already supports scheduled triggers. We just did not wire them in.

3. Start with a 10-county pilot. We went all-in on 254 counties, and the first batch revealed structural issues that affected every subsequent batch. A pilot run of 10 diverse counties (urban, rural, large, small, different regions) would have caught the edge cases faster. If you replicate this, run your pilot on Collin, Brewster, Harrison, Deaf Smith, and Nacogdoches. That covers urban, rural east, rural west, panhandle, and deep east Texas patterns in one pass.

The Takeaway

Programmatic outreach is not a replacement for genuine relationship building. It is a replacement for the mechanical part of outreach that should never have been manual in the first place. Researching a name, writing a sentence, and clicking send is not a skilled activity. It is data entry dressed up as marketing.

The 254-county pipeline proved that local SEO outreach can be engineered like any other system. You identify the pattern, build the automation, handle the exceptions, and let the system run. The human role shifts from execution to strategy, from “who do I email today” to “which counties are we missing and what new offer would make them want to link.”

If you are managing a site with local relevance across multiple regions, the single highest-leverage SEO investment you can make is not a better content strategy. It is an outreach pipeline that treats every location like the identical system it structurally is, while respecting the local specificity that makes the outreach actually work.