Site Migration Without Losing Paid Traffic: A Google Ads Checklist for 2026

The redesign shipped on a Friday evening. By Monday: a wall of disapproved ads, zero weekend conversions, quality scores down on commercial keywords, Smart Bidding back in learning, CPA doubled. Nobody broke anything deliberately. Paid traffic was simply left out of the plan, because a site migration is almost never scheduled with advertising in mind.

There is no shortage of SEO migration checklists, and nearly all of them ignore advertising. Yet paid reacts to a URL change faster and harder than organic does: rankings erode over weeks, ads break within the hour. What follows is a site migration checklist written for Google Ads — what to do two weeks before launch, on launch day, and across the first two weeks after.

Why paid suffers more than organic

1. Advertising runs on hard-coded URLs. Ads, assets, sitelinks and tracking templates all contain specific addresses. Change the address and the ad points at a 404 or a redirect — and Google checks this automatically.
2. Moderation re-reviews landing pages. A mass final-URL change triggers a fresh review. If the new site does not meet Google Ads destination requirements, disapprovals arrive in bulk.
3. Bidding loses its history. A changed conversion structure or broken tracking severs the data feed. The strategy re-enters a learning period and behaves unpredictably for a week or two.

One rule saves weeks of pain: at launch, only pages should change. Not pages plus tags plus conversion structure plus domain, all at once. Split every migration you can.

Stage 1 — two weeks out

1.1 Export every URL advertising touches

Using Google Ads Editor or the final URL report, collect the complete list:

• final URLs on ads across every campaign type;
• keyword-level final URLs where they are set individually;
• URLs inside assets: sitelinks, promotions, prices, images;
• mobile final URLs, where defined;
• tracking templates at account, campaign, ad group and ad level;
• final URL suffixes;
• feed URLs for Shopping and Performance Max;
• destinations configured for app campaigns.

Bulk export and replacement is really only practical through Google Ads Editor bulk operations — manual edits do not scale past a small account.

1.2 Get an old-to-new URL map from engineering

Two columns: old address, new address. Demand it before launch, not after. Check specifically that:

• every advertising landing page appears in the map;
• none of them redirect to the generic homepage — collapsing landing pages into the homepage is the fastest way to destroy conversion rate;
• query parameters survive the redirect, or you will lose the GCLID;
• thank-you page addresses, where conversions fire, are accounted for.

1.3 Confirm tags and conversions will survive

The expensive part of a migration is never the URLs — it is measurement. Before launch, verify:

• the Google tag will be present on every page of the new site, including cart and thank-you;
• the same Measurement ID and conversion IDs carry over rather than being recreated;
• conversion events keep their existing names — renaming breaks history;
• if a new domain or subdomain enters the funnel, cross-domain tracking is configured in advance;
• the consent platform on the new site passes signals instead of blocking tags outright — why that matters is covered in our piece on third-party cookies after the Privacy Sandbox shutdown.

1.4 Record a baseline

Snapshot the 28 days before launch: CPC, CTR, conversions, CPA, conversion rate per key campaign, impression share, and landing page speed. Without a baseline you cannot later prove whether a dip is the migration or seasonality. Capture current Core Web Vitals too — new designs are almost always heavier than the ones they replace.

1.5 Pick the date deliberately

Never migrate before a weekend, in a holiday week, or at seasonal peak. Tuesday or Wednesday morning is ideal: the whole team is available and you have three working days of runway.

Stage 2 — two days out

• Prepare the URL replacement file in Editor but do not post it. Have it staged and ready.
• Review the staging site as a moderator would: contact details, privacy policy, refund terms, content that matches the ads.
• Pause automated rules and scripts that could react to an anomaly and, say, switch campaigns off on a CPA spike — see automated rules in Google Ads.
• Warn everyone who reads the reports about relearning, especially if decision-makers are not the people running the migration.
• Decide what happens to experiments. Running tests through a migration produces dirty data. Stop them and relaunch once things settle.

Stage 3 — launch day, in order

Sequence matters. Do it like this:

1. Ship the site with 301 redirects already live. Permanent, not 302. One hop, not a chain of three.
2. Manually check key pages: 200 status, tag loading, form submitting, thank-you page rendering.
3. Fire a test conversion. Walk the funnel end to end and confirm it reaches the interface, allowing for a few hours of reporting delay.
4. Only now replace final URLs in ads. One campaign first, verify, then bulk.
5. Update assets and the product feed. A feed URL change triggers reprocessing and some items may disapprove temporarily — keep the guide to product disapprovals and feed diagnostics to hand.
6. Audit tracking templates. Templates with the old domain hard-coded stop working after a domain change — see GCLID and tracking templates.
7. Timestamp the event. Leave a note; it makes later change-history diagnosis dramatically easier.

Why you cannot update ad URLs before the site ships

The new addresses do not exist yet. Google crawls, gets a 404, disapproves the ads, and the re-review after you fix it costs you time. The reverse mistake is just as costly: leaving old URLs live after launch works via redirect, but adds latency and drops parameters as soon as the redirect becomes a chain.

Stage 4 — the first 48 hours

What to watch	Normal	Red flag
Ad status	A handful of disapprovals	Mass “Disapproved: policy violation”
Hourly conversions	Dip on launch day, recovery next day	Zero conversions for 12+ hours
Impression share	±10% movement	Down a third or more
Conversion rate	Within baseline ±20%	Halved at the same traffic level
404s in server logs	Occasional	Growing on ad landing pages
Landing page speed	Close to previous	LCP doubled

If ads stop en masse, start by reviewing the landing page through a moderator’s eyes, not by filing appeals. Common causes and a triage order are in why your ads are not showing. If it is conversions that vanished, work through conversion tracking diagnosis instead.

Stage 5 — the first two weeks

The dominant mistake here is fixing campaigns that are not broken. After a migration the algorithm is genuinely relearning, and every sharp edit extends that process.

• Leave bid strategies and targets alone for 7–14 days unless costs are truly runaway. Let learning finish.
• Do not restructure the account in the same window as the site move. One variable at a time.
• Watch lead quality, not just lead volume. A new form can increase submissions while degrading them — separate primary and secondary conversions.
• Check remarketing list sizes. If the tag landed on the new site late, lists have been draining.
• Reconcile three sources — Google Ads, GA4, CRM — a week after launch. A gap above 15% means something is still not arriving.

Who owns what

Half of failed migrations fail organisationally rather than technically: the advertising task floats between marketing and engineering and nobody picks it up. Write the ownership down before you start.

Area	Owner	Deliverable
Old-to-new URL map	Engineering	Sheet signed off by marketing
301 redirects and parameter preservation	Engineering	Staging verification report
Tag deployment on the new site	Engineering + analytics	Page list with confirmed tag coverage
Exporting and replacing ad URLs	Paid media	Editor file, before and after
Policy compliance review	Paid media	Screenshots of key pages
Post-launch test conversion	Analytics	Screenshot of the conversion in-platform
First 48 hours monitoring	Paid media	Report against the red-flag table
Rollback decision	Product owner	Rollback criteria defined in advance

That last row matters more than it looks. Define rollback criteria before launch — for example, “we roll back if conversions stay at zero for six hours while traffic is live.” Otherwise the call gets made in a panic at hour three, when everyone is tired and guessing.

Site migration: a day-by-day timeline

• Day −14. Export URLs, request the redirect map, capture baseline metrics.
• Day −7. Map received, reviewed, gaps closed. Tagging plan reviewed.
• Day −3. Staging reviewed through a policy lens; URL replacement file prepared.
• Day −2. Automated rules and scripts paused, experiments stopped, launch-day budgets trimmed.
• Day 0, Tuesday morning. Ship, verify pages, fire a test conversion, then replace URLs in ads, assets and feed.
• Day 0, evening. First read: ad statuses, hourly conversions, 404s in logs.
• Days +1 and +2. Monitor against the red-flag table. Only surgical fixes — disapprovals, broken links.
• Day +3. Restore budgets and re-enable automated rules.
• Day +7. Reconcile Ads / GA4 / CRM for the week and decide what genuinely needs fixing.
• Day +14. Return to normal optimisation cadence; relaunch experiments.

Reporting the outcome without fooling yourself

Comparing “week after” to “week before” is the easiest and most misleading method available. Seasonality, promotions and day-of-week effects will produce a difference on their own. Instead:

• Add a year-over-year comparison alongside the adjacent week to strip out seasonal drift.
• Watch conversion rate, not conversion count. If traffic fell because of disapprovals, absolutes drop even with a perfect landing page.
• Separate landing-page effects from learning effects. The first shows up immediately in CR; the second shows as unstable CPA across two weeks.
• Count money, not events. Conversions can rise while revenue falls — an easier form produces more, weaker leads.

Special cases

Full domain change

The hardest version. On top of the general checklist:

• add the new domain to the cross-domain list and keep both domains listed through the transition;
• re-verify ownership in Merchant Center and update the store URL;
• if you are a verified advertiser, check whether details need resubmitting — a domain change can trigger re-verification, so plan it rather than discovering it mid-season;
• make sure policy and legal pages exist on the new domain; their absence is a frequent disapproval trigger;
• keep the old domain alive with redirects for at least a year.

CMS replatform with unchanged URLs

It feels safe, which is exactly why it breaks tracking so often. The URLs match, but tags were re-implemented, events renamed and the consent platform swapped. Verify measurement as rigorously as in a full move.

Moving to an SPA or headless front end

The classic trap: navigation no longer reloads the document, so tags built around page load fire once or never. You need history-based event handling and an individual check on every conversion action.

Adding language versions

Introducing /en/, /de/ and friends rewrites URLs for a slice of traffic. Split campaigns by language version immediately and confirm that geo-based redirects do not bounce ad traffic away from the intended landing page — an ad that lands somewhere other than it promised is a disapproval waiting to happen.

If the migration already happened — and went badly

A common scenario: nobody consulted you, the site moved, performance collapsed. Recover in order of cost, cheapest first.

1. Confirm ads still reach live pages. Export final URLs and script a status-code check. Anything returning 404, a redirect chain or a homepage redirect gets fixed first.
2. Verify the tag on every conversion page. One forgotten thank-you page explains most of the drop more often than not.
3. Fire a test conversion. Until it completes by hand, every hypothesis stays a hypothesis.
4. Read the change history and line the launch date up against the inflection point on the chart. If the break predates the launch, the migration is not your culprit.
5. Review disapprovals by group, not by individual ad. Mass disapprovals nearly always share a single root cause.
6. Measure the new landing pages’ speed. If LCP doubled, part of the loss is genuine experience degradation rather than tracking.
7. Only then touch bidding. While data is incomplete, any strategy change bakes the error in.

If you discover conversions went unreported for weeks, do not backfill them haphazardly. Import the deals properly, with real timestamps and identifiers — a sloppy backfill damages learning more than the original gap did.

The five mistakes that repeat everywhere

1. Every landing page redirected to the homepage

It is easier for engineering: one rule instead of a hundred. For advertising it is a disaster — someone searched for a specific service and arrives at a generic shopfront. Conversion rate falls by multiples and Google separately downgrades landing page relevance. Insist on page-level mapping, and where a page genuinely no longer exists, redirect to the nearest equivalent section rather than the homepage.

2. Conversion actions recreated from scratch

New analytics stack, new action IDs, “clean slate.” The result: history wiped, bid strategies learning from zero, no valid period comparison. Carry the existing conversion IDs across even if it costs engineering extra work.

3. Redirects that eat parameters

The most invisible failure of all. The site works, pages load, but the GCLID and UTMs are dropped on the hop, so conversions stop tying back to clicks. One minute to check: open an old URL with ?gclid=TEST123 appended and see whether the parameter survives to the final address.

4. Friday releases

Google’s review systems work through the weekend; your team does not. Two days of unattended mass disapprovals cost far more than moving the launch to Tuesday.

5. Panic edits on day two

CPA is up, so someone lowers the target, switches strategy and rewrites the ads. Now three artificial relearning cycles sit on top of the natural one, and the account takes six weeks to recover instead of two. The rule is simple: in the first 72 hours, fix breakages, not metrics.

The consolidated checklist

• Complete export of every advertising URL.
• Old-to-new URL map received and reviewed.
• No landing pages collapsed into the homepage.
• 301 redirects, single hop, parameters preserved.
• Google tag and conversion IDs carried over unchanged.
• Conversion events keep their names.
• Cross-domain tracking updated for the new structure.
• Consent Mode working on the new site.
• 28-day baseline recorded.
• Automated rules and scripts paused.
• Experiments stopped.
• Launch scheduled early in the week.
• Manual test conversion completed after launch.
• URLs updated in ads, assets, feed and templates after launch.
• First 48 hours monitored against the table above.
• Two weeks without sharp strategy edits.
• Ads / GA4 / CRM reconciliation after week one.

If the account is still behaving oddly more than two weeks out, sweep it properly with the Google Ads account audit checklist. And when a migration coincides with entering a new category and running into platform-side limits, agency Google Ads accounts and advertiser verification are the next step — after you have confirmed the migration itself is not the cause.

FAQ

Should I pause campaigns during the migration?

Usually not. Pausing wipes accumulated signal and forces relearning on its own. Reducing budgets by 20–30% on launch day while keeping campaigns live is the better trade.

Can I just keep the old URLs and rely on redirects?

Temporarily, yes. Permanently, no. Redirects add latency, hurt mobile experience, and start dropping parameters once they chain. Update the URLs within a day or two.

How long does recovery take?

As a rough benchmark: three to five days to return to previous volume and up to two weeks for bid strategies to stabilise. If nothing has recovered after three weeks, that is no longer a migration effect — something is genuinely broken.

Why did quality score fall after the redesign?

Usually landing page speed and relevance. Heavier design and rewritten copy change both the page experience assessment and how well the page matches the query.

What do I do about mass disapprovals?

Do not fire off dozens of appeals. Open the new landing page and check it against policy: contact details, legal pages, content matching the ad, no cross-domain redirect. Fix the cause and the re-review will clear.

Will I lose my remarketing audiences?

Existing members stay — the lists live in the account. But they stop refilling if the tag arrives on the new site after launch, and members expire as membership duration elapses.

Should I create new conversion actions on the new site?

No, and doing so hurts: a new action resets history and restarts learning from zero. Reuse the existing conversion IDs.

How should Performance Max be handled during a URL change?

Update feed and asset links, then give the campaign 7–10 days untouched. PMax relearns more slowly than standard campaigns.

Does a migration affect advertiser verification?

A change of domain name can require resubmitting details. Plan for it rather than hitting it at peak season.

What is the single most important post-launch check?

The manual test conversion. Until one has completed by hand, nothing in the reporting interface proves anything.

We migrated a month ago and only now noticed a drop. What now?

Start with change history and a conversion-to-CRM reconciliation. It usually turns out some events stopped firing and the culprit is one overlooked tag, not the move itself.

Is a phased migration worth it?

Yes, where the architecture allows. Moving section by section surfaces errors on a small slice of traffic instead of putting the entire paid channel at risk in a single release.