1031 Property Selection: a WordPress to Astro rebuild for search and answer engines

A nine year old WordPress site on managed hosting, rebuilt as a static Astro site with its URL structure intact, structured data a search engine can read, and analytics that send nothing to a third party.

The problem: an aging WordPress site in a competitive category#

The site sold a specific thing to a specific reader: Delaware Statutory Trust and 1031 exchange property selection for accredited investors. The category is unusually well suited to search, because the people who need it arrive with a question rather than a brand in mind. They type what is a Delaware Statutory Trust, or DST vs TIC, or 1031 exchange deadlines, and then they read for an hour before they ever speak to anyone.

What they found was a GoDaddy Managed WordPress install dating to around 2021, carrying a few genuinely good pages and roughly thirty thin syndicated tax articles that existed because a plugin had put them there. The good pages were competing for attention with the filler, the whole thing was slow, and the parts a search engine most needs in order to understand a page, the structured data, were not present at all.

The domain itself was the asset worth protecting. It was aged, on topic, already indexed, and fully under the client's control, which is a combination that cannot be bought and is very easy to destroy with a careless rebuild.

Migration without losing the URLs#

The failure mode in a migration like this is not a broken page. It is a page that still works but now lives at a different address, so every link pointing at it and every ranking it had accumulated quietly stops counting.

• The URL shape was preserved exactlyAstro was configured to keep trailing slashes and directory style paths, so every address the old site answered, the new one answers identically. No redirect, no hop, no lost signal.
• Content was ported verbatim, typos includedRewriting during a migration means that when rankings move you cannot tell whether the platform or the copy caused it. The words were moved first and improved afterwards, as a separate change.
• The thin articles were consolidated, not deletedAround thirty syndicated tax pages were redirected permanently into a single glossary, which keeps whatever authority they carried and removes thirty weak pages competing with the fifteen strong ones.
• The result is a static siteNo database, no PHP, no plugin surface. Pages are files on a CDN, which is both faster for the reader and a materially smaller thing to attack.

Structured data, because the questions are the traffic#

In this category the traffic is questions, so the page has to be legible as an answer to a machine as well as to a person. That is what structured data is for, and the old site had none. The rebuild publishes four blocks, all of which validate on the live site today:

The Person block matters more than its size suggests. Credentials that appear only as prose are a claim; credentials tied to a public regulatory identifier are a claim that can be checked, and in a category this heavily impersonated, being checkable is the differentiator.

Answer engine optimization, and the compliance constraint#

A growing share of this audience never reaches a results page. They ask an assistant and read what it tells them. Being cited there needs two things: the crawlers have to be allowed in, and the answers have to be short enough and self contained enough to lift.

The site's robots file explicitly welcomes GPTBot, ClaudeBot, PerplexityBot, Google-Extended and CCBot, which is a deliberate choice rather than an oversight. Blocking them protects nothing here. The content is already public, and the whole commercial point is to be the source an assistant quotes when someone asks what a Delaware Statutory Trust is.

We then built the clusters those questions fall into: a DST versus TIC comparison, a pros and cons page, a passive income page, and an FAQ hub of twenty seven questions whose answers are written at forty to sixty words each, which is the length an answer engine will quote whole rather than paraphrase.

First party analytics with zero third party scripts#

The site runs no Google Analytics, no Meta pixel and no third party tag of any kind. It has its own analytics instead, written as serverless functions against the host's object store.

It records what the client actually needs: which channel a visitor came from, including paid click identifiers from Google, Meta, Bing, Reddit and TikTok, and referrals from AI search tools, which are attributed as their own channel rather than being lost in direct traffic. First touch is kept, so a visitor who reads for two weeks and then converts is credited to whatever first brought them. Visitor IP addresses are never stored.

For a site whose readers are investigating a six or seven figure tax decision, a page that reports their visit to five advertising networks is a bad look and an unnecessary liability. Running our own was less work than the consent banner the alternative would have required.

Where it stands#

Sixteen pages, live since July 2026 and extended since. Over the last ninety days the site earned impressions on more than four thousand eight hundred distinct search queries, concentrated exactly where the work was aimed: the Delaware Statutory Trust question family, DST versus TIC, and 1031 exchange mechanics.

Those positions are still climbing out of the range where impressions turn into clicks, which is the honest state of a six month old rebuild in a category contested by firms with far larger budgets. The structural work is done, the measurement is in place, and the queries it is being shown for are the right ones.