My reports

Loading your device history…
The practical guide

Build a useful record from your local speed results

Published by InternetSpeedGeo · Reviewed

What belongs in this report history

×

The table contains completed speed results retained in this browser's local storage and accepted by the site's stored measurement checks. It does not automatically collect every application session or every test performed elsewhere. The retained record includes the measurement time, download, upload, ping and available jitter, along with other context used by the site where available. A failed or interrupted run should not be mistaken for a completed record merely because partial numbers appeared while it was running.

The history is associated with a browser profile and website origin. Another device, browser or private session can therefore show another collection. There is no required account that joins them into one synchronized archive. Begin by checking that you are looking in the same context where the measurement completed. If a result is visible immediately after a test but is absent after reopening the page, storage availability may be part of the question. A local record is useful evidence of the completed observation, but the word verified in the interface describes its measurement validation, not a laboratory certificate. Keep that distinction when presenting an exported result to a provider or another person.

Decide what you need the record to explain

×

A useful report has a purpose. You might want to document a recurring evening problem, compare a desk location, retain observations before an equipment change or show an administrator a reproducible failure. Each purpose calls for different context. An unlabelled collection of every test can be harder to interpret than a smaller, well described selection that directly addresses the issue.

Write a short question before exporting. For a hypothetical example, the question could be whether the same laptop receives more consistent throughput near the access point than in a distant room. The relevant record would identify those positions, preserve the device and connection method and state the time of each session. A provider billing question would need the applicable plan and provider process as well; the website record alone does not establish the contractual answer. This page helps organize technical evidence, not decide a dispute. A good purpose statement also tells you when to stop collecting. Once you can describe the pattern and its limits clearly, another repeated test may add less value than passing the evidence to the person who can act on it.

Read each row before relying on a summary

×

The date and time place a measurement in sequence. Download and upload are separate directions, normally displayed in Mbps. Ping and jitter use milliseconds and describe different aspects of timing. Jitter can be unavailable. A row should be read as a set of related observations from a completed run, not as interchangeable scores that can be added together into one number.

Check for context changes around a surprising row. A result taken on mobile data after several home Wi-Fi tests belongs to another path, even if it appears in the same local table. An older measurement method may also require qualification. A hypothetical series with a single high result may reflect a wired comparison rather than a sudden improvement lasting through the rest of the day. If the table does not contain the circumstance you remember, add it to your external notes rather than guessing from the value. The website cannot recover every environmental detail after the fact. Reading the rows first prevents a general average or best card from becoming a misleading headline in a report whose underlying observations came from mixed conditions.

Add the context the browser cannot know

×

The saved fields provide part of a measurement record, but they do not capture every factor affecting the connection. You may need to note the room, whether a backup was running, whether the device moved, whether the browser remained active and what symptom occurred in the application you care about. Keep those notes close to the corresponding timestamp in your own report.

A compact note can state the device, active local connection, upstream service, VPN state, broad position and relevant concurrent activity. It need not include a street address, password or complete hardware inventory. For a hypothetical work call investigation, noting that the upload of a large project started during the affected period may be more useful than recording an unrelated device serial number. Distinguish an observation from a suspicion: you can record that a backup was active without claiming it caused the delay until a comparison supports that conclusion. If a detail is unknown, mark it unknown. Clear uncertainty is better than a confident label reconstructed from memory. Context makes the raw values actionable while limiting the temptation to infer a cause from a single result.

Export while the needed observations are still available

×

The Export CSV action creates a downloadable file from the results currently retained in this browser. The export includes labelled columns for date, measured values, connection and server. It is a snapshot of the available table at export time, not a live link that continues updating automatically. Save it in a location you can find and open it to confirm that it contains the intended records.

Local history keeps up to fifty completed results. Older observations can leave the retained collection as new ones are added, and browser cleanup or manual deletion can remove the collection. If you need a before change baseline, export it before making the change or clearing data. A hypothetical visitor replacing a router should preserve the earlier report with its context rather than assume the website will reconstruct it afterward. The exported copy becomes yours to manage. Saving it in a synchronized folder, emailing it or placing it in a shared document creates other copies outside the Reports control. Choose the destination according to the purpose and sensitivity of the record, and avoid treating a download as either a permanent backup by default or an automatically private vault.

Open a CSV without changing its meaning

×

CSV is a text table format that many spreadsheet programs can open. The receiving program may interpret dates, decimal separators and text according to its own locale and import settings. If columns appear merged or dates look reversed, inspect the import settings before concluding that the original measurement is wrong. Preserve the original exported file so you can distinguish its contents from changes introduced during editing.

Use a working copy for calculations, filters and notes. A hypothetical spreadsheet might convert a date string into a different display format while preserving the same instant, or treat a blank jitter field as something requiring attention. Do not replace unavailable values with zero just to make a formula convenient. If you change units or round values for presentation, label that transformation and retain the original precision in your private working record when needed. A clean looking table can still be misleading if dates lose their timezone context or column headings lose their units. Before sharing, review a few rows against the on screen history and confirm that download, upload, ping and jitter remain in the intended columns with the intended meanings.

Keep original data and interpretation separate

×

A report often needs both raw observations and an explanation. Keep them distinguishable. The exported values record what the tool returned; your notes explain conditions; your conclusion interprets the comparison. Editing a number to match a theory destroys that separation. If a record is excluded from an analysis, document why rather than silently removing it from the only remaining copy.

For a hypothetical comparison, a test accidentally taken over mobile data might reasonably be excluded from a home broadband analysis. The reason is a different network context, not that its speed was inconvenient. A valid but unusually slow test taken under the intended conditions should not disappear merely because it weakens a preferred conclusion. You can investigate whether it corresponds to known activity and present it with that context. This practice makes the report reproducible and fair. Another person should be able to see which observations support the conclusion and which were outside the stated question. You do not need elaborate statistical software to maintain that discipline; a preserved original export, a labelled working copy and a short explanation of selection rules can be enough for a small troubleshooting task.

Keep a simple naming and handover convention

×

Give your saved report a name that identifies its purpose and observation period without embedding sensitive identifiers. An accompanying note can state which browser supplied it, who needs it and whether the file is an untouched export or an edited working copy. This small convention becomes valuable when several versions accumulate during troubleshooting. A file named only latest can stop being the latest as soon as another export is created, while an unexplained edited copy can be mistaken for original evidence.

When passing a report to someone else, explain the intended next step. A hypothetical handover might ask an administrator to review a recurring upload pattern, with the relevant rows and conditions identified. Another could ask a household member to compare an agreed device position using the same method. Specify what information would resolve the question, rather than asking the recipient to interpret every column without context. If they add observations, keep their source separate and note the new conditions. A short handover also identifies what the report cannot establish, such as the absence of a wired baseline or an unmeasured application destination. This reduces unnecessary back and forth and protects the evidence from being reused later as a broader claim than its author intended. Organization is part of accuracy when a record travels between people.

Build a before and after comparison

×

Before changing a setup, define the change and the conditions you will try to preserve. Use the same device, browser, measurement service and relevant connection mode where possible. Record a baseline, make the authorized change, and collect comparable observations afterward. Avoid combining a new router, a new laptop, a different room and another subscription into one unexplained after condition.

A hypothetical improvement after moving an access point can support that placement change if the other important conditions remain similar. It cannot automatically prove a universal benefit for every home. Keep the practical outcome tied to the task, such as fewer delays in a particular room, as well as the numerical difference. A faster download result may not address an upload problem or a distant application's delay. If the change has no clear effect, preserve that finding too; it can prevent unnecessary spending or repeated configuration work. The report should state what changed, what remained controlled, what was observed and what uncertainty remains. If conditions could not be matched, explain that limitation rather than overstating the strength of the comparison.

Document an intermittent problem without pretending to monitor continuously

×

Intermittent problems need a record of both the symptom and the sampled conditions. Note when the application problem occurred, when the speed test was run and whether the two overlapped. A measurement taken after a disruption ended may be useful but does not directly measure the disruption itself. Do not label a day as healthy solely because a short test during that day completed well.

A hypothetical report might show several completed observations during evening call difficulties and a smaller set during unaffected periods. State that these were selected snapshots, not continuous monitoring or a random daily survey. Record normal periods too when they help the question, but avoid running a test so frequently that it becomes new competing traffic. If an error prevented completion, preserve the error separately from successful rows; the local history is not a complete log of every failed attempt. A small incident note can include start time, symptom, affected application, connection state and the available measured evidence. This format gives support a clearer picture than an average that merges affected and unaffected periods without indicating which was which.

Separate devices and connections when assembling a combined report

×

Because history is local, combining observations from several devices requires a deliberate external workflow. Export from each relevant browser and label the source before merging anything. A matching timestamp or provider does not guarantee the same device conditions, and different measurement methods may appear in older records. Keep a source column in your own working document so the origin of each row remains visible.

In a hypothetical household comparison, one person may test on a wired desktop and another on a phone in another room. Both observations are useful, but their difference cannot be assigned solely to the provider because device and local link differ. A combined report should group or label those conditions. If the purpose is to compare devices, say so; if the purpose is to compare a provider over time, try to hold the device constant. Do not assume the website has synchronized the histories or removed duplicate records for your manual merge. Review identifiers and dates where available, and avoid counting the same exported observation twice. A carefully labelled small collection is easier to reason about than a large spreadsheet whose rows have lost their origin.

Prepare evidence for an internet provider

×

A provider conversation benefits from a concise symptom description, relevant dates and a controlled observation. Explain whether the issue affects one application or several, one device or several, and wired or wireless use. Include the plan information only as supplied by the provider, with the correct direction and units. A speed result can support a technical question, but it is not itself contractual certification or proof of a billing remedy.

A hypothetical support summary could state that the same authorized wired device repeatedly observed a particular pattern during named periods, while ordinary use showed a corresponding symptom. Attach a limited table rather than a folder of unlabelled screenshots. Note the Cloudflare measurement destination and method so the provider can understand what path was tested. If the provider requests its own diagnostic process, follow the authorized instructions and preserve the difference between those results and this site's records. Avoid treating disagreement between tools as proof that one organization is dishonest. Different endpoints and methods can produce different observations. A report should make the question answerable, such as asking the provider to investigate recurring outbound performance under the documented conditions.

Prepare evidence for an administrator or application service

×

For a workplace or school administrator, begin with the affected task and the permitted diagnostics you performed. Managed networks may intentionally route or filter traffic, so mention the relevant context without attempting to bypass controls. A test failure to one endpoint can be a policy issue rather than a general loss of connectivity. The administrator is better placed to interpret internal equipment, routing and access rules.

For an application service, describe its own symptom and destination information where available. A hypothetical game problem that continues while the website's speed result is strong still deserves investigation, but the report should not say that the website proved the game server was at fault. Instead, explain that one measured path performed well while the application showed a different behavior. This narrows the evidence without pretending to locate the cause. Share only the rows and context needed by the recipient. The same complete export is rarely appropriate for every audience. A local network administrator may need device context; an editorial correction may need only a screenshot of a label; a privacy request may need a result identifier and date. Match the record to the person who can act.

Redact carefully without obscuring the evidence

×

Review screenshots and exports before sending them. A browser image can expose tabs, notifications, account names, bookmarks and location details unrelated to the problem. Crop or redact those areas while keeping the relevant values, units, date and error readable. If the exact address or provider is not needed for the recipient's task, consider whether a broader description is sufficient.

Redaction should be explicit when it could affect interpretation. A hypothetical report can state that a precise location was removed while the country and connection type were retained. Do not alter measured values under the label of privacy cleanup, and do not remove the context that explains why two rows are comparable. Use a copy for sharing so your original evidence remains available privately. An exported file sent by email or stored in a shared drive can have a different audience from the browser history. Clearing Reports does not recall those copies. The practical aim is a narrow, accurate record: enough information for the question, without a surplus of identifiers or unrelated activity that introduces another concern into a simple troubleshooting exchange.

Clear history with its scope in mind

×

The clear action asks for confirmation and removes speed history from the current browser's local storage. It is useful when you want to remove those visible records from that context. Export anything you need first, because the website does not promise recovery after deletion. Clearing all site data through browser settings can also affect language and privacy preferences, depending on the scope you select.

Local deletion does not remove an exported CSV, a copy in an email, a backup or a contribution already accepted by the server. A successful home page test separately attempts to contribute measured values for provider averages. If your goal concerns that stored contribution, contact privacy@internetspeedgeo.com rather than assuming the local button performs a server deletion. On a shared device, also consider the download folder and browser profile used to access the site. A hypothetical person preparing to hand over a computer should review the relevant local copies, not only the table. These are practical distinctions about where records exist. They do not imply that Reports can inspect or erase every copy created by a user or processed by an external service.

When a record or export seems wrong

×

Start by identifying whether the issue is the measured value, the stored row, the export or the receiving program's display. A result can be correctly stored while a spreadsheet interprets its date unexpectedly. A missing jitter value can be accurately preserved as blank. Another browser can legitimately have no matching history. Naming the layer helps avoid a destructive troubleshooting step that removes useful evidence.

Check the same browser profile, compare the visible row with the exported text where practical and preserve the original file. If the discrepancy remains, send hello@internetspeedgeo.com the page, approximate time, browser and a minimal example. State whether the file was edited after download. Do not attach every unrelated result or sensitive network document. A hypothetical report might show that one export column differs from the same completed row, with both values clearly identified. That is more actionable than saying the statistics are wrong. If you are concerned about a server contribution rather than the local table, say so explicitly because its acceptance and storage are separate. The contact page offers additional templates for reporting tool, label and privacy issues.

Finish with a readable evidence summary

×

A useful final report can be short even when the underlying evidence is extensive. State the question, conditions, observed pattern and next action. Attach only the rows needed to support that pattern and preserve the larger original collection privately if appropriate. Explain the method and any known limitations, including mixed devices, missing metrics or sampling only during problem periods.

For a hypothetical example, a report might conclude that comparable observations improved after moving a device closer to its access point, while the provider and measurement method were held constant. That supports a local placement decision without claiming a provider wide diagnosis. Another report may conclude that the evidence is mixed and a wired baseline is still missing. Acknowledging that gap can be the most useful result because it identifies the next comparison. Reports is a workspace for retaining observations, not an automatic verdict. Its value comes from accurate records, careful context and a clear relationship between the evidence and the decision you need to make. Once those are present, move the question to the appropriate action or support channel instead of collecting numbers indefinitely.

Overview

Build a comparison you can trust

  1. Keep the same setup
  2. Change one condition
  3. Compare saved results
Keep the device, server and connection context consistent. A single reading is not a provider reliability score.

Reports keeps completed measurements available in the browser where you made them. Use the table, export and careful contextual notes to preserve evidence for a specific question.

ContactMethodologyWiki

Useful, at a glance

Relevant next steps to help you put this page into practice.