A website migration rarely leaves your traffic charts perfectly calm. URLs may change, analytics tags can behave differently, and returning visitors may encounter an unfamiliar journey. Google also warns that a medium-sized website can take a few weeks to move through its index, while larger sites may take longer.
That does not mean every post-launch decline is normal. A genuine measurement failure can look remarkably similar to lost demand, and a broken redirect can resemble seasonal weakness. Luckily, you can separate these issues if you collect the right evidence before launch and compare like with like afterwards.
The aim is not to prove that the migration “worked” with one impressive percentage. It is to understand whether real people can still find the site, reach the right pages, and complete useful actions. That requires a stable baseline, consistent reporting windows, page and channel analysis, and enough historical context to recognise normal variation.
How to Create a Pre-Migration Traffic Baseline
A baseline is the “before” photograph of your site. Without one, a traffic change has no useful context. You may know that sessions fell by 12 per cent, but not whether that movement is unusual for a Tuesday, a school holiday, or the final week of the month.
Start with at least 28 complete days before launch. For a site with low traffic, irregular campaigns, or strong weekday patterns, eight to twelve weeks gives you a more dependable picture. Seasonal organisations should preserve the equivalent period from the previous year as well, because the month immediately before launch may be a poor benchmark.
Do not rely on a single site-wide session figure. Record sessions, users, engaged sessions, engagement rate, key events, session key event rate, and revenue or qualified leads where relevant. These measures capture both traffic volume and visitor quality, which matters because a migration can preserve total sessions while making important journeys harder to complete.
Add Search Console data to the baseline if Google Search contributes meaningful visits. Capture clicks, impressions, click-through rate, and the pages and queries responsible for the largest share of traffic. Search Console measures what happens in Google results, while GA4 measures activity after someone reaches the site, so the two sources answer different parts of the same question.
Save page-level data rather than only screenshots of headline charts. Your export should include each old landing page, its sessions, engaged sessions, key events, revenue or lead volume, and its main channels. The highest-traffic pages deserve particular attention, but do not discard lower-volume pages that support an important service, product, or regulated customer journey.
It is also worth preserving the raw detail while you still can. Standard GA4 properties allow user-level and event-level data retention of two or 14 months, and that setting affects Explorations and funnel reports rather than standard aggregated reports. Exporting the baseline now prevents a later investigation from being limited by retention settings or a changed report configuration.
Next, document the measurement setup itself. Record the GA4 property and web stream IDs, consent settings, internal traffic rules, referral exclusions, channel definitions, campaign naming conventions, and the time zone used for reporting. If any of these change during the move, note the exact time, because the apparent traffic difference may be a tracking difference rather than a change in human behaviour.
Create an old-to-new URL map alongside the baseline. For every meaningful old URL, record the intended new destination and its pre-migration traffic. Google’s site-move guidance recommends starting with important URLs found in sitemaps, analytics, server logs, and link data. That map later becomes your quickest route from “traffic is down” to “these specific pages did not transfer as expected”.
Finally, record the circumstances around the launch. Include the deployment time, campaign schedule, media spend, email sends, planned outages, public holidays, major promotions, and any other change likely to affect visits. Of course, it is tidier to change one major thing at a time, but real migrations rarely happen under laboratory conditions. A clear change log lets you account for the messiness instead of guessing.
How to Compare Visits Before and After Migration
Post-migration measurement is more like checking a patient’s recovery than reading a final exam result. The first reading matters, but the pattern over time matters far more. Launch-day data is often distorted by partial reporting, staff testing, tag fixes, cache clearing, and unusually heavy crawling, so keep it visible for diagnosis but exclude it from the main performance comparison.
Use several observation windows. A daily view helps your team catch a severe tracking or availability problem quickly. A rolling seven-day view smooths weekday effects, while a fixed 28-day window gives you a more credible formal comparison once enough time has passed.
Match the pre-migration and post-migration windows exactly. They should contain the same number of days and, for short comparisons, begin on the same weekday. GA4 supports comparisons with the previous period and the previous year, including a previous-period option that matches the day of the week. This small detail prevents a weekend-heavy period from being compared with a working-week-heavy one.
Calculate both the absolute and percentage change for every key metric. If sessions moved from 40,000 to 36,000, the absolute change is minus 4,000 and the percentage change is minus 10 per cent. The absolute figure shows the commercial scale, while the percentage makes it easier to compare pages and channels of different sizes.
That being said, percentage changes become dramatic when the starting number is small. A page moving from two sessions to four has grown by 100 per cent, but the difference is only two visits. Group low-volume pages by directory, template, service line, or content type before drawing conclusions from them.
Do not expect GA4 sessions and Search Console clicks to match. Google explains that clicks and sessions use different systems and calculations, so numerical equality is not the goal. Look for broadly similar direction and timing instead. If both fall around launch, the decline may be real; if Search Console is steady while GA4 collapses, tracking or consent deserves immediate attention.
Compare quality alongside volume. Sessions may recover while engagement rate, key event rate, or revenue remains weak because visitors are reaching the wrong destination or encountering a poorer experience. Equally, a modest fall in visits may be less concerning if qualified leads and revenue remain stable.
Use your own normal variability as the threshold for concern. A site whose daily traffic routinely moves by 15 per cent should not treat a one-day 8 per cent decline as proof of failure. Compare the movement with the range seen during the baseline, then ask whether it persists across several days and appears in more than one data source.
Keep the analysis human first. A dashboard can say that the migration retained 98 per cent of sessions, yet conceal that mobile visitors can no longer submit the main form. Test critical journeys yourself and combine analytics with customer-service messages, form records, ecommerce data, call tracking, and CRM outcomes. Traffic only has value when people can do what they came to do.
How to Measure Changes by Landing Page and Channel
A site-wide total is like a building’s main electricity meter: it tells you something changed, but not which room has the fault. Landing-page and channel analysis turns a vague traffic decline into a short list of places to investigate. It also stops a healthy part of the site from hiding a serious loss elsewhere.
Begin with the old-to-new URL map. Add post-migration sessions, absolute change, percentage change, engagement rate, key events, redirect destination, and final HTTP status for every mapped page. Sort first by lost sessions, not by percentage decline, so pages with the greatest practical impact rise to the top.
Any page that had meaningful baseline traffic and now receives no landing sessions deserves attention. Check whether the old URL returns a permanent redirect, whether the destination loads successfully, and whether the analytics tag fires after consent. Also confirm that the new page serves the same visitor need. Sending several distinct pages to a generic category page may be technically convenient, but it can leave people in the wrong place.
Analyse groups as well as individual URLs. Compare product categories, service areas, editorial directories, language versions, and page templates. If nearly every page built with one template loses traffic or engagement, the common template is a stronger lead than dozens of apparently unrelated URLs.
Then segment the change by Session default channel group, Session source / medium, device, country, and where useful, new versus returning users. GA4’s Traffic acquisition report is designed to show where new and returning visitors’ sessions came from. Keep the same definitions and filters on both sides of the migration, or your comparison will mix real behaviour with classification changes.
Different patterns point to different problems. A decline across nearly every channel suggests an analytics, consent, uptime, or sitewide experience issue. A loss concentrated in Organic Search may reflect recrawling, indexing, redirect, or landing-page changes. Falling Referral traffic can indicate that external links now reach errors, while an increase in Direct alongside a fall in Email or Paid Social often points to missing campaign parameters.
Campaign traffic needs especially careful handling. If your acquisition plan includes services described as buy website traffic, paid placements, influencer links, or short-term media tests, tag those visits consistently and isolate them from the migration baseline. Otherwise, a burst of purchased or promoted sessions can make total traffic look healthy while core channels are declining.
Do not stop at sessions when reviewing a channel. Compare engaged sessions, engagement rate, key events, session key event rate, and revenue or qualified leads. If Paid Search volume is stable but its key event rate falls only on the new mobile landing pages, the likely problem is the post-click experience, not demand or campaign delivery.
Search Console adds another useful layer for Google traffic. Compare pages and queries across the same date windows, then review country, device, and search type separately. If impressions remain stable but clicks fall, visitors are still seeing the site in results but choosing it less often. If both impressions and clicks fall for a mapped group of pages, the investigation should focus on that group rather than the entire domain.
This page-by-channel view gives you a practical order of work. Prioritise large losses on commercially important pages, especially where redirects, tracking, engagement, and key events all point in the same direction. Leave tiny, isolated fluctuations until you have enough data to judge them properly.
How to Separate Migration Impact From Seasonal Fluctuations
Traffic has a calendar. Retail sites feel holidays, business-to-business sites feel working weeks, travel sites feel booking seasons, and local services react to weather and regional events. If a migration happens near one of those turning points, a simple before-and-after comparison can blame the deployment for a change that would have happened anyway.
Start with a year-over-year comparison using equivalent dates. Google recommends viewing up to 16 months of Search Console performance data to see whether a drop repeats each year. For GA4, compare the post-migration period with the same period in the previous year, then sense-check differences in campaign spend, promotions, opening hours, product availability, and tracking.
Year-over-year data is not automatically perfect. Easter moves, leap years add a day, and public holidays can fall on different weekdays. Align trading days where possible and annotate unusual events. For a business with strong weekly rhythms, comparing Monday to Sunday with another Monday to Sunday is often more meaningful than matching date numbers exactly.
Use external demand as a second line of evidence. Google Trends can show whether interest in important topics rose or fell across the wider market. Search Console query impressions provide a closer view of demand reaching your own site. If market interest and your impressions fall together, at least part of the decline is likely external.
A control group can make the distinction clearer. Choose a region, content section, product line, or campaign that was minimally affected by the migration but responds to the same seasonal forces. If total sessions fall by 18 per cent while the control group falls by 12 per cent, the rough incremental difference is six percentage points. It is not laboratory proof, but it is more informative than attributing the full 18 per cent to the move.
Also check what else changed near launch. Advertising budgets may have shifted, an email programme may have paused, a competitor may have run a major promotion, or a search platform may have updated its systems. Google lists seasonality, changing interests, technical problems, algorithmic changes, and migrations among the possible causes of search traffic drops. Good analysis tests these explanations rather than choosing the migration simply because it is the most visible event.
Watch the shape and location of the decline. Seasonality often appears gradually and affects related pages or channels in a familiar pattern. Migration problems are more likely to begin close to launch and concentrate around changed URLs, devices, templates, or tracking conditions. The distinction is not absolute, which is why several signals are better than one.
Of course, a temporary decline can still require action. Google says ranking fluctuations are expected while changed URLs are recrawled and reindexed, and that a medium-sized site may take a few weeks to move through the index. Continue monitoring the recovery curve, but investigate promptly if important landing pages remain at zero, redirects fail, tracking disappears, or key events deteriorate.
The most reliable judgement comes from combining four views: the stable baseline, matched comparison periods, landing-page and channel detail, and seasonal context. Together, they tell you not merely whether the traffic number moved, but whether the migration changed how real people find and use the site. That is the measure that should guide your next decision.
