
You updated your DNS records, refreshed your browser, and… nothing. Your old site is still there. Or worse, it works for you but a customer in another country still sees the old version. Cached DNS information can explain this symptom, but configuration errors can look similar. Investigate the answers before calling every delay DNS propagation.
This guide explains what DNS propagation really is, why different people see different results, how TTL controls the wait, and how to make your next DNS change almost painless.
DNS (the Domain Name System) translates a domain like hostreta.com into the IP address of the server that hosts it. When you change a DNS record — for example, pointing your domain to a new host — you update it at your DNS provider. Publication depends on the service; verify its authoritative answers after saving.
The delay comes from everywhere else. To keep the internet fast, DNS answers are cached by thousands of systems along the way:
Each of these keeps the old answer until it expires. “Propagation” is simply the time it takes for all those cached copies to expire and be replaced with your new record. Nothing is actively “spreading” across the internet — old answers are just timing out one by one.
Because every resolver cached your record at a different moment, they all expire at different moments. That’s why, during a change:
This is normal. The important thing during a migration is to keep the old server running until propagation is complete, so nobody lands on a dead site. And before any big change, take a fresh backup — see Website Backups: The 3-2-1 Rule Explained.

Read our TTL change-planning guide for the distinction between the previously cached lifetime and the value published after an edit.
Every DNS record has a TTL (Time To Live), measured in seconds. It tells resolvers how long they may cache the answer before asking again.
Here’s the key insight: the TTL that matters is the one that was set before your change. If your record had a 24-hour TTL, some resolvers will keep the old answer for up to 24 hours, no matter what you change now.
That’s where the famous “up to 48 hours” warning comes from. In practice, most changes are visible within minutes to a few hours — but a long old TTL, or a resolver that ignores TTLs, can stretch it out.
Checking only from your own computer tells you very little. Use these methods instead:
nslookup yourdomain.com 8.8.8.8 (Google) and nslookup yourdomain.com 1.1.1.1 (Cloudflare) to compare major resolvers. On Mac or Linux, dig yourdomain.com +short works too.dig @ns1.yourdns.com yourdomain.com. Compare the relevant authoritative servers and delegation. One correct response does not prove every other problem is only caching.To see the new site on your own device sooner, flush your local cache: restart the browser, run ipconfig /flushdns on Windows, or switch to a different network (for example, your phone’s mobile data).
You can’t force resolvers worldwide to forget your record, but you can plan so the wait is tiny:
Changing nameservers (not just records) usually takes longer, because the registry’s own TTL for nameserver records is often 24–48 hours. Choose the operation that matches the project: editing a website record and moving authoritative DNS are different tasks. Using Cloudflare? Read Cloudflare Free vs Pro: Is Upgrading Worth It?.
The Cloudflare TTL reference describes its provider-specific behavior. Use the documentation for your service rather than treating one provider’s timing as a universal rule.
Write down what was edited and when. A website address record, a mail routing change and a nameserver migration involve different dependencies. Record the name, type, previous value and replacement. Without that context, repeated checks can ask different questions and produce results that appear contradictory.
Keep the source of the new value, such as the destination host’s setup instructions. Confirm that it belongs to the intended application. A wrong destination will not become correct merely because more time passes. Assign one person to coordinate the investigation so simultaneous speculative edits do not hide the original problem.
Save observations separately from conclusions. For example, note that a named resolver returned the old address at a particular time. Then explain that caching is one possible cause. This is more useful than declaring that the entire domain is broken or that every unexpected answer is normal.
The authoritative service publishes zone data, while recursive resolvers can retain earlier answers. Check the active authority first. A hosting panel can display an inactive zone even when another company handles the public domain. Editing that inactive copy may never change the public result.
Compare the exact record and the relevant authoritative servers. If their answers differ unexpectedly, preserve the server names and responses for support. One correct answer does not establish a consistent configuration across a provider migration. Review delegation and the destination zone before deciding that waiting is the only remaining action.
Keep the original service available through the planned transition where practical. Record which system is responsible for live application data. DNS does not synchronize databases, so two destinations receiving customer actions can create a separate migration problem.

Follow the dig and nslookup checks to compare the exact hostname, record type and selected server. Use full answers when short output leaves the cause unclear.
Use the same full hostname and type in each test. A result for www does not establish the root-domain answer, and an A lookup does not establish the AAAA result. Record whether a proxy or CDN is intended, because its public address may differ from the origin server.
Keep the answering resolver and time with each observation. Resolvers can fetch data at different moments and therefore show different remaining lifetimes. Preserve full output during diagnosis; a shortened address alone may omit the details that explain the difference.
For internal business names, use the organization’s permitted DNS services. Public resolvers may not have private-zone data, and internal hostnames should not be disclosed merely to follow a tutorial. Choose a comparison appropriate to the actual network.
A global checker provides observations from selected locations or resolvers. It is useful evidence, but it cannot establish that every visitor worldwide sees the same answer. Check which name and record type the tool queried before interpreting the display.
A map of successful lookups does not prove that the website certificate, database or mailbox works. Test the actual service separately. If one result remains unexpected, compare it with authoritative data and the change timeline instead of repeatedly refreshing without a specific question.
Keep representative observations in the project record. This helps the next administrator distinguish a stale cached answer from an unchanged authoritative record. It also prevents one local observation from being described as a global completion guarantee.
If the browser reaches the intended destination but shows an old page, inspect application and content caches. The browser, server, application or CDN can preserve content independently of DNS. Another DNS edit may not refresh an image or rebuild a page.
Test the affected page using an ordinary visitor context where practical. An administrator session may bypass a cache that customers use. Record whether the symptom is an old address, an old page, a certificate warning or a failed form; those observations point to different layers.
Use the cache controls appropriate to the identified layer. Do not treat cache clearing as a substitute for correcting an address, redirect or application setting that remains wrong. Preserve the evidence and repeat the same test after a targeted change.
A site can use IPv4, IPv6 and aliases across several services. Review the records involved in the actual design. If only one address family moved, visitors can have different experiences because the configuration is inconsistent rather than because one global waiting period has not ended.
An alias can introduce another relevant target lookup. Follow the relationship before interpreting the final result. A short TTL on one entry does not describe every dependency. Obtain destination values from the responsible platform instead of guessing from another site’s setup.
When a proxy is intended, compare the public result with that architecture. Do not force a record to expose the origin merely because it differs from a server address in the hosting panel. Check proxy routing and application behavior separately.

Review nameservers versus individual DNS records to identify which account controls the intended change. If validation fails after moving providers, consult the DNSSEC explanation before changing security records.
Moving authoritative DNS requires a prepared destination zone. Inventory mail, verification, security and application records, not just the website address. Review any automatic import against the service inventory, because a scan may miss required entries.
If DNSSEC is enabled, coordinate parent and signing information through the providers’ documented process. A website-record TTL is not a DNSSEC migration plan. Investigate validation failures instead of automatically dismissing them as a temporary cache delay.
After changing delegation, test representative records and dependent applications. Keep the previous service available for the agreed transition. Avoid repeatedly switching nameservers in response to a single early observation, which can make the intended configuration harder to assess.
A store or membership site can receive writes while different visitors reach different destinations. Plan final data synchronization and any controlled maintenance period with the application owner. DNS does not merge orders or user updates between two databases.
Decide when the new destination becomes the business system of record. Keep backups and a rollback decision process. A working home page does not establish that customer actions remain consistent during the cutover.
Include mail if the project also moves email services. Routing records do not transfer historical messages or create destination accounts. Test incoming and outgoing workflows independently so a website success does not hide a mailbox problem.
Include the original configuration, the change time, the previous TTL and the exact unexpected result. Identify the server queried and whether the symptom appears on one network or several. State what service is affected and what you expected to happen.
Ask which part of the configuration the provider can verify. A registrar, DNS host and application host have different responsibilities. A precise request helps support investigate the right system instead of repeating a generic instruction to wait.
Review diagnostics before sharing them. Public DNS observations may be suitable for support, while credentials and unrelated internal information need restricted handling. Never publish account passwords or API tokens in a public troubleshooting discussion.
Promotional recommendation: If the project requires a domain or hosting move, review the services at KAWEDMA’s domain and hosting services. Explain the current DNS provider, destination application and transition requirements. Ask which setup and migration work is included and how DNS propagation and application checks will be verified. Confirm the scope rather than assuming every plan includes a complete migration.
No. Verify authoritative configuration and the exact symptom. An incorrect record or missing application should not be ignored because a general waiting period was quoted.
No. A local cache action affects only a local part of the path. It does not clear every recursive resolver or other user’s device.
No. The previous lifetime, delegation and service configuration matter. A short supported TTL is a planning tool rather than an exact global switch.
Follow the verified migration and data-retention plan. Check remaining traffic, application data and service dependencies before ending a transition.
Record the final authoritative answers, representative resolver checks and important application tests. Include unresolved items and their owners. Another administrator should be able to understand what was verified without reconstructing the entire investigation.
Keep the successful configuration with domain and hosting documentation. Review it when services or staff change. The useful outcome is a working, maintainable setup with a clear support route, not merely a screenshot of a checker showing one favorable answer.
Before another person takes over, give them the original symptom and the smallest set of observations that explains the investigation so far. Include a working comparison where available. Keep commands, hostnames, record types and times together so the next person can reproduce the same question.
Do not describe a partial check as a global result. State which authoritative answers were inspected and which applications were tested. A precise DNS propagation report can say that selected resolvers now return the intended address while a particular user still reports an issue. That remaining symptom deserves its own follow-up.
After a targeted correction, repeat the original test and the affected customer action. If only one improves, record that distinction and continue investigating the other layer. Close the task when the agreed evidence supports the intended service state, and leave any remaining work assigned to a named operational owner.
Keep DNS propagation notes with the account map so future migrations start from the current configuration.
Review DNS propagation evidence again before closing the migration and retiring old services.
We may earn a commission on purchases made through these links, at no extra cost to you.
Don’t miss the chance to participate in the biggest giveaway of this year.