What tools do you need for a website migration?

Seven categories, and they map to the four places a migration breaks. A redirect mapping tool for matching old URLs to new ones at scale. An agentic coding tool for the bulk scripting: data transforms, redirect file generation per server, schema additions across templates. A Search Console export tool for the pre-migration baseline, because the native interface only keeps 16 months. Cloud infrastructure for staging environments and parallel testing. A spreadsheet built for large datasets, since URL inventories break consumer spreadsheets. Analytics that reports immediately after cutover. A peer network for the problems that are not in any documentation. Nothing on that list replaces a person who knows which pages carry revenue.

How many URLs can you redirect by hand before it stops being viable?

The break point in practice is a few hundred. Below that, manual mapping is faster than configuring a tool and checking its output. Above it, three things go wrong at once: the hours stop fitting in the schedule, the error rate climbs as attention fades, and nobody can review the work because there is no record of how confident each match was. Automated mapping is worth buying at that point less for the speed than for the confidence score, which tells you which few hundred of your ten thousand matches deserve a human look.

What should you capture before a site migration goes live?

A page-level Search Console export covering the full 16 months, a crawl of the existing site, the current redirect rules already in place, and revenue by landing page from the commerce platform. Those four give you the only defensible answer to the question that arrives two weeks after launch, which is whether traffic fell because of the migration or because of something else. Capture them before cutover. Search Console history rolls off, staging environments get torn down, and the old site stops responding the day DNS moves.

How do you tell whether a post-migration traffic drop is the migration?

Compare against something that did not change. A holdout works best: a category or section left on the old structure, measured on the same days. Where a holdout is not possible, compare the migrated pages against their own pre-migration baseline and against sitewide seasonality at once, because a drop that shows up in both is not the migration. Server logs settle the remaining cases, since they record crawler behavior and referrer-stripped visits that browser analytics never sees.

Do you need paid tools for a small migration?

Under a few hundred URLs, no. A crawl, a spreadsheet, and a careful afternoon covers it. Paid tooling earns its cost on scale and on evidence: thousands of URLs to match, a staging environment to run in parallel, and a baseline you can produce months later when somebody asks what happened. The one exception at any size is the Search Console export, because that history disappears whether you paid for anything or not.

Website migration tools: the seven we reach for.

A migration goes wrong in four places: the redirect map, the data import, the scripting nobody budgeted for, and the week after launch when nobody can prove what changed. These are the seven tools we reach for, what each one covers, and where it sits in the timeline.

Available in:

Deutsch

Author: Tony Castillo

Published August 10, 2026

← All articles How we research

Seven small clay tools laid out in a row on cream paper, beside a clay arrow splitting one path into two

The short version

Migrations fail in four places, not one. The redirect map, the data import, the scripting nobody budgeted for, and the week after launch when nobody can prove what changed. A toolset that covers three of those still loses the argument in the fourth.
Redirect mapping is the job that scales worst by hand. Ten thousand URLs matched manually is weeks of work and a guaranteed error rate. It is the first place to spend money, and the one place where a confidence score per match is worth more than the match itself.
Baselines cannot be built after the fact. Search Console gives 16 months of history and then it is gone. Pull the page-level export before cutover, because the post-launch conversation is entirely about what the numbers were beforehand.
Buy the tools, keep the judgment. Every tool here compresses work a person would otherwise do by hand. None of them decides which redirects matter, which pages carry revenue, or whether a traffic drop is the migration or the season.

A Migration Breaks in Four Places

Nobody loses a migration on the design. It goes wrong in four places, and each one is somewhere the work outruns the hours in the schedule.

The redirect map goes first. Then the data import, usually discovered when a category page renders with half its products. Third comes the scripting nobody put in the estimate: URL normalization, redirect files in three server formats, schema markup across a few hundred templates. The fourth arrives two weeks after launch, when traffic is down and nobody can prove whether the migration caused it.

01The redirect mapThousands of old URLs to match against a structure that does not line up with the old one.

02The data importProducts, media and metadata arriving incomplete, in the wrong field, or not at all.

03The unbudgeted scriptingBulk transforms and validators that were assumed rather than scoped.

04The week after launchTraffic moves and there is no baseline to measure it against.

The seven tools below cover those four. They are the ones on our own machines, listed in the order a migration reaches for them.

The Seven

Redirects.net

Redirect map

Matching old URLs to new ones is the job that scales worst by hand. Redirects.net reads both URL sets and proposes matches using a natural language model, which holds up on the structures that defeat pattern rules: product slugs with SKUs buried in them, category paths that were reorganized, legacy URLs nobody can explain.

What it does: Proposes a target for every source URL and scores its confidence in each match. Tens of thousands of URLs process in minutes.
Why we use it: The confidence score, more than the matching. It sorts ten thousand rows into the few hundred that need a human and the rest that do not, which is the difference between reviewing the map and rubber-stamping it.
Where it fits: Before cutover, as soon as the new URL structure is stable enough to export.

Redirects.net Demo on YouTube

Claude Code

Scripting

Every migration carries a pile of repeatable engineering that nobody scoped. Claude Code is Anthropic's agentic command-line tool. It reads the repository, plans a change, runs the commands, and leaves a reviewable session behind.

What it does: Bulk URL normalization, redirect files generated for Apache, Nginx and Cloudflare from one source of truth, schema markup added across templates, staging against production diff checks.
Why we use it: It moves the constraint. The work that used to wait for an engineering week becomes an afternoon a practitioner can supervise, and the output is a diff rather than a black box.
Where it fits: Throughout. Heaviest in the two weeks before cutover and the first week after.

Claude Code Documentation

SEOGets

Baseline

Search Console holds 16 months and the native interface makes page-level history hard to pull at any useful size. SEOGets extracts it and keeps going past the window.

What it does: Exports the full history at page level, tracks metrics per URL over time, and surfaces the pages carrying the most to lose.
Why we use it: The post-launch conversation is entirely about what the numbers were beforehand. A baseline built after cutover is an opinion.
Where it fits: Before cutover, and it cannot be backfilled. This is the one item on the list with a deadline attached.

SEOGets Demo on YouTube

DigitalOcean

Infrastructure

A migration needs somewhere to run that is not production: a staging copy to test redirects against, a box to run transforms on, room to scale if traffic shifts after launch.

What it does: Droplets, managed databases and App Platform for staging environments, migration scripts and the production site itself.
Why we use it: Predictable pricing and a short path from decision to running environment. A staging box that takes a procurement cycle to approve does not get used.
Where it fits: From the first test import through the weeks after launch.

DigitalOcean

Row Zero

Data

URL inventories break consumer spreadsheets. A full crawl of a mid-size ecommerce catalog runs to hundreds of thousands of rows, and the tool that opens it has to stay responsive while you filter and pivot.

What it does: Opens and manipulates datasets far past the point where a desktop spreadsheet stalls, with the cleaning and transformation steps that redirect mapping needs.
Why we use it: Migration data arrives inconsistent, from several systems, in several shapes. Most of redirect mapping is standardizing it before any matching happens.
Where it fits: Alongside the redirect map, and again during import validation.

Row Zero Demo on YouTube

Plausible

Post-launch

The first 48 hours after cutover are the window where a broken redirect is cheap to fix. Reporting has to be immediate and simple enough to read under pressure.

What it does: Lightweight, privacy-friendly analytics with live traffic by page, referrer and goal, and no cookie banner to configure.
Why we use it: Speed of reading. A launch-night dashboard that needs three clicks and a custom report is a dashboard nobody checks at 2am.
Where it fits: Instrumented before cutover, watched hard for the first week after.

Plausible Analytics

The SEO Community Slack

People

Some migration problems are not in any documentation, because they are specific to a platform version, a plugin, or a combination nobody wrote up. The fastest route to those answers is somebody who hit it last quarter.

What it does: A Slack workspace with channels for technical SEO, Search Console and ecommerce. The technical SEO channel alone runs past 3,500 members.
Why we use it: Sanity checks before cutover, and a second opinion during one. Both are worth more the closer you get to the date.
Where it fits: Planning, and again at 11pm on launch night.

The SEO Community

Where Each One Sits

Ordering matters more than the list. Two of these have a deadline: the Search Console baseline and the staging environment both stop being available once DNS moves.

PhaseToolsWhat it is for

PlanningSEOGets, Row Zero, SlackPull the baseline while it still exists, inventory the URLs, pressure-test the plan.

BuildRedirects.net, Claude Code, DigitalOceanMap the redirects, script the transforms, run everything against a staging copy first.

CutoverPlausible, Claude CodeWatch live traffic, fix broken rules in the hours when it is still cheap.

AfterSEOGets, PlausibleCompare against the baseline and settle the question of what the migration caused.

What This List Does Not Do

Every tool here compresses work a person would otherwise do by hand. None of them decides which redirects matter. A tool scores a match at 60 percent confidence and stops. Whether that URL is a discontinued product nobody searches for or the page carrying a fifth of category revenue is a judgment call, and getting it wrong in either direction is expensive.

The same is true of the week after launch. Analytics reports the drop. Deciding whether it is the migration, the season, or an algorithm update that landed the same week takes someone who has watched the pattern before.

Under a few hundred URLs, most of this is unnecessary. A crawl, a spreadsheet and a careful afternoon will do it. The list earns its cost on scale and on evidence, and the Search Console export is the one item worth pulling at any size, because that history disappears whether you paid for anything or not.

It's Now Safer to Put Cloudflare in Front of Shopify: What O2O Fixed, and What the Warnings Still Mean

Website migration tools: the seven we reach for.

A Migration Breaks in Four Places

The Seven

Redirects.net

Claude Code

SEOGets

DigitalOcean

Row Zero

Plausible

The SEO Community Slack

Where Each One Sits

Have us audit the migration in depth.

What This List Does Not Do

Add our migration team to yours.