A publication proof file is a dated folder that lets you show, weeks or months later, that a purchased guest article or link was live, where it lived and what it looked like. Six items go in it: the order record, the live URL with the check date, a screenshot, the link line copied from the page source, an archive capture and, when you own the linked site, a Search Console export. Each item answers a different question, and none of them answers all of them. Held together, they separate three things that are easy to blur: when the page said it was published, when you saw it, and when a third party captured it.
The people who need this are the ones who pay for placements and answer for them later: business owners, agencies, and the accounting and operations staff who close out orders. When a publisher edits a page, removes a link or disputes a delivery, the folder is what the conversation rests on.
Example scenario: A fencing club bought one package of ten guest articles to bring in new members for its beginner classes. The club secretary logs the delivery, and a volunteer treasurer approves the spending.
Why a proof file earns its keep
A proof file pays off whenever someone asks a question that memory cannot answer. Three situations produce those questions.
When a delivery is disputed
A message to a publisher is stronger when it names an address, a date and what was visible. Without a record, the exchange turns into one person's recollection against another's. With the folder, the message can say: this URL, on this date, showed this paragraph with this link line. For the wording and structure of such a message, see the guide to disputing a backlink purchase with evidence. If the article was not published at all, the route differs and is laid out in how the dispute process works when a guest post is not delivered. Either way, the order record and the check dates are the first things asked for.
When someone wants a report
A report is a summary of the folder. Whoever prepares it should be able to point at a row and then at the files behind that row. If the sheet and the files drift apart, the report becomes an opinion. A package with several placements makes this harder, so tracking by order number matters; the backlink package delivery tracking guide covers that side of the routine.
When the person who ran the order has left
Audits and handovers test whether the record survives its author. A folder named after the order number, with a fixed file pattern, can be read by a stranger. A folder named after a person's initials cannot. Practical suggestion: keep the folder in a shared team location and grant access by role rather than by individual.
The six items in the file
The order record
The order record is prepared before the publication arrives. It holds the order number, the order date, the publisher, the expected address or domain, the target page for the link, the expected anchor text and the position within the package. It works like an empty inspection form: once the publication arrives, expectation and reality sit side by side. Practical suggestion: a filled record might read F2208, ordered 2026-11-03, Publisher B, expected address on their blog, target the beginner classes page, anchor text "beginner fencing classes", position 2 of 10. If the published piece differs on any field, write the difference beside that field.
The live URL and the check date
Open the address the publisher sent and record the final URL shown in the browser bar. If the address redirects, record both the first address and the final one. Put the date and time of your check in its own column. That column is yours. The publication date printed on the page is the publisher's statement, and the two belong in separate cells. For reading the page's own date signals, see the guide on proof of publication date: page date, dateModified and sitemap lastmod.
The screenshot
A useful screenshot shows the address bar, the page headline, the paragraph holding the link and, where possible, the system clock. Practical suggestion: take one full-page capture and one close-up of the paragraph. Hovering over the link in the close-up displays the destination in the browser's status bar, which puts the target address in the picture.
The source line
A screen shows what the browser drew, and the source shows what the page contains. Copy the anchor element that holds the link and paste it as plain text. From that line you read the href, the rel value and the anchor text. Google's guidance on crawlable links (Opens in a new tab) says that Google can't reliably extract URLs from a elements that have no href attribute, and it lists a span carrying an href as not recommended. A link written that way is not dependable evidence. So a link that looks fine on screen can still fail the source check.
The archive capture
The Internet Archive help page (Opens in a new tab) describes "Save Page Now" as a way to save one specific page one time. The capture URL contains a timestamp in yyyymmddhhmmss order, which records when the capture was made. That is a statement about the capture, and the page's original publication date is a separate fact.
The Search Console export
If you own the linked site, the Links report in Search Console lists a sample of the external links Google has found. Export the table and store the file with its date in the name. Interface labels may differ if your Search Console is set to a language other than English.
Table: Items, the question each answers and when to capture
| Item | Question it answers | When to capture | Example file name |
|---|---|---|---|
| Order record | What was bought? | Before publication | F2208_record |
| Live URL entry | Where is it live? | First check day | F2208_01_url_20261120 |
| Screenshot | How did the page look? | First check day | F2208_01_screen-full_20261120 |
| Source line | How was the link written? | First check day | F2208_01_source_20261120 |
| Archive capture | Is there a third-party copy? | First check day | F2208_01_archive_20261120 |
| Search Console export | Has Google listed the link? | After indexing | F2208_01_gsc_20261201 |
Folder and naming rules
A file name should say what the file is before anyone opens it. A short, strict pattern survives staff changes.
A file name pattern
Practical suggestion: lead with the order number so that files stay together even when someone copies them out of their folder, then add the placement number, the item and the date. Writing the date last as year-month-day keeps repeated checks of one item in order. Use plain ASCII letters and hyphens, because spaces and accented characters cause trouble when files move between systems.
- Pattern: orderNo_placementNo_item_YYYYMMDD.extension
- Example: F2208_01_screen-close_20261120.png
- A repeat capture on the same day gets a suffix: -r2
- Do not overwrite an older file; keep both and let the date tell them apart
The folder tree
The tree runs order, placement, check day. The order-number folder sits at the top, one folder per placement sits under it, and each check gets its own dated subfolder.
- F2208 (order folder named after the order number)
- 00-record (the order record and an invoice summary)
- 01-publisher-a (one folder per placement)
- 01-publisher-a/2026-11-20 (one subfolder per check day: screen, source, archive)
- 02-publisher-b and so on
- log-sheet (one sheet, one row per placement)
Keep publisher labels short. If one publisher appears in several orders, the order number keeps the folders apart.
Where the files live: one shared folder or inside the order
Decision rule: keep all proof in one shared folder named after the order number, and put only that folder's link in the order record. Files attached inside the order record scatter once accounting and operations need different permissions. If the order system cannot hold attachments at all, the shared folder is the only option anyway. Write the folder link into the order note and grant access at team level.
What each proof can and cannot show
Each proof supports a limited kind of statement. Knowing which one keeps messages honest.
Table: What each proof shows and does not show
| Proof | What it shows | What it does not show |
|---|---|---|
| Screenshot | The page as the browser drew it at the check | Link structure or whether the link is crawlable |
| Source line | href, rel and anchor text | Whether the page is indexed |
| Archive capture | The timestamp of a third-party copy | The original publication date |
| Search Console export | A sample of links Google has found | That every link is listed, or whether a link is nofollow |
Limits of an archive capture
The Internet Archive help page states that "Save Page Now" does not add the URL to future crawls and does not save multiple pages, directories or entire sites. Pages may be absent from the archive if they were password protected, blocked by robots.txt or excluded at the site owner's request. JavaScript that needs to contact the originating server in order to work will fail when archived. The page also notes a lag of 3 to 10 hours between a crawl and the moment the capture appears in the Wayback Machine. Treat the archive as a delayed copy rather than instant evidence.
Limits of a screenshot
Practical suggestion: file the screenshot as a document that the source line and the archive capture back up. It shows what was drawn. Whether the thing drawn is a real link with an href cannot be read from the picture.
Limits of the Search Console Links report
According to Google's help page, the report is not a comprehensive list of every link to your site; it shows a sample. Tables are limited to 1,000 rows, and some URLs can be left out for reasons such as non-indexed pages or deduplicated URLs. Data is grouped by root domain. The report does not say whether a link is marked nofollow. Links in it may since have been removed, or the page may no longer exist. A link missing from the list therefore does not prove the placement is absent, and a link present in it does not guarantee that the link is still on the page.
Google's documentation can change, so check the current help page before relying on any limit quoted here.
Reading the link line
For an image link, Google uses the alt attribute of the img element as anchor text. A copied source line for an image link should therefore include the img element and its alt value. For a text link, the visible text between the opening and closing anchor tags is what gets read. The Search Console Links report help page (Opens in a new tab) describes a "Top linking text" view of the anchor text used in links from outside your property, which gives you something to compare against the anchor you ordered.
Logging ten publications from one package in one sheet
A package multiplies the files, so one sheet carries the status of all ten placements. One row per publication, one column per proof.
Table: Example scenario, a ten-row log sheet for a package
| No. | Publisher | Expected URL | Link target | Check date | Screen | Source line | Archive | Status |
|---|---|---|---|---|---|---|---|---|
| 1 | Publisher A | Expected address 1 | Club home page | Check day | Yes | Yes | Yes | Done |
| 2 | Publisher B | Expected address 2 | Beginner classes page | Check day | Yes | Yes | Pending | Awaiting archive |
| 3 | Publisher C | Expected address 3 | Club home page | Check day | Yes | Yes | Yes | Done |
| 4 | Publisher D | Expected address 4 | Beginner classes page | Check day | Yes | Missing | Yes | Source missing |
| 5 | Publisher E | Expected address 5 | Club home page | Check day | Yes | Yes | Yes | Done |
| 6 | Publisher F | Expected address 6 | Contact page | Check day | Yes | Yes | Yes | Redirected |
| 7 | Publisher G | Expected address 7 | Club home page | Check day | Yes | Yes | Yes | Done |
| 8 | Publisher H | Expected address 8 | Beginner classes page | Check day | Missing | Missing | Missing | Not live |
| 9 | Publisher I | Expected address 9 | Club home page | Check day | Yes | Yes | Yes | Done |
| 10 | Publisher J | Expected address 10 | Beginner classes page | Check day | Yes | Yes | Pending | Awaiting archive |
The order in which to fill it
Practical suggestion: fill the sheet on set check days, not as each article lands. That keeps every row's date comparable. Capture the URL and screen first, the source line second and the archive last. Because the archive copy appears with a delay, that column can be closed on a different day. Limit the status column to a short fixed list: Done, Awaiting archive, Source missing, Redirected, Not live. A short list makes filtering easy.
Closing a row
A row closes when the screen, source and archive cells are filled. If what was delivered differs from what was ordered, for instance a link pointing at the home page where the beginner classes page was ordered, write a difference note in the same row with the date the publisher was told. Rows are updated, not deleted.
What the example sheet tells the club
In the example scenario, row 8 is a dispute item because nothing is live. Row 4 lacks a source line, so the link's structure cannot be shown and the publisher is asked for a re-check. Row 6 is live but sits behind a redirect, so both addresses belong in the sheet. Rows 2 and 10 only wait for the archive, which is not a problem. The sheet decides which rows go to the treasurer, which go to the publisher and which need nothing.
Re-checks and change notes
Proof accumulates over time rather than on one day. Practical suggestion: take the first record on publication day, then repeat the same checks on the dates the team has fixed. A re-check takes a fresh screenshot and a fresh source line and does not replace the earlier ones. Differences go into a change note: did the text change, was the link removed, did the address start redirecting? Practical suggestion: write each change note in three parts: what the first record showed, what the re-check shows and what you did about it. In the example scenario, the club's re-check of row 3 finds that the paragraph holding the link has disappeared. The note reads: first record showed the link to the beginner classes page, the re-check shows the paragraph without a link, the publisher has been asked in writing. The note stays in the row, the reply goes into the folder beside the first record, and nothing is deleted. The routine for the weeks after publication is set out in the post-publication checklist for sponsored articles.
Footer and template placements follow the same folder logic. The guide to footer link follow-up after purchase shows one sequence for verification, indexing and reporting.
Who fills the sheet and who signs off
Two roles keep the file honest. The secretary, or whoever runs operations, collects proof and fills the sheet. A second person, here the treasurer, approves each row closure. Practical suggestion: record the approval in its own column with the approver's name and date, so a later question about who closed a row has an answer.
Keep personal data out of screenshots. A signed-in account name or an email address showing in a corner should not be part of the capture; checking while signed out avoids it. Payment details and service credentials do not belong in the folder at all. The retention period follows the organisation's own records policy and is written in the root note of the folder.
Practical suggestion: keep the root note to half a page. It says where the log sheet lives, what each status value means, who approves rows and when the next re-check falls. A new colleague can then open the folder without a briefing, and the treasurer can audit a closed order without asking the secretary.
Putting an existing backlog in order
A team that has not kept a proof file may want to build one for past orders. Some of that is possible and some is not. A screenshot of how a page looked last spring cannot be produced now. Only today's state can be recorded, and the row carries today's check date. If the Wayback Machine holds an earlier capture of that page, its timestamp can be copied into the sheet exactly as shown.
Practical suggestion: a first-week plan.
- Day one: agree the shared location, the folder tree and the file pattern, and name the person who approves rows.
- Day two: list the open orders and, for each, create the order record and an empty log sheet, marking what can no longer be reconstructed.
- Day three: sweep the live placements in a single sitting, taking the URL, the screen and the source line for each.
- Day four: request archive captures for every live address and note any earlier captures found in the calendar.
- Day five: send difference notes to publishers, collect row approvals and write the root note with the retention period.
New orders do not have this problem, because the record is prepared before publication.
Decision rules and a checklist
Decision rule: which proof leads in which case
Decision table: Leading proof by situation
| If you find | Pull first | Then add |
|---|---|---|
| The link points at the wrong page | Source line | Close-up screenshot |
| The page is gone at the re-check | First record and its archive timestamp | The re-check screenshot |
| The address redirects elsewhere | Both addresses in the URL entry | Archive capture of the final address |
| Nothing at the expected address | Order record and check dates | A note on the failed archive attempt |
| Accounting wants a status | Log sheet summary | Search Console export file |
Checklist by moment
Before the publication arrives:
- Order record filled in, including target page and expected anchor text
- Check calendar set, with the name of the person who checks
On each check day:
- Final URL and every redirect step written down
- Date and time of the check in their own column
- Full-page and close-up screenshots saved
- Anchor element pasted from the source as plain text, then href, rel and anchor text entered in the sheet
- Archive capture requested, or the waiting status written
When closing the row:
- File names match the pattern and no earlier version was overwritten
- Search Console export attached with its date, if the site is yours
- Approval recorded with the approver's name and date
Mistakes to avoid
- Skipping the archive request on the first check day: the page may change before anyone thinks of it.
- Typing free text in the status column: filtering stops working once every row words it differently.
- Recording a redirected placement under its first address only: the final address is the one readers reach.
- Saving a screenshot that shows a signed-in account name: personal data enters the folder.
- Tidying the copied source line before pasting it: the folder then holds your edit, not the page.
- Closing a row while the archive cell still says pending: the row looks complete when it is not.
- Keeping proof on one person's drive: the record leaves with the person.
Frequently Asked Questions
Can a PDF print of the page replace the source line?
Practical suggestion: no. A print records what was drawn, in the same way a screenshot does. Google's documentation says it can't reliably extract URLs from a elements that have no href, and a print does not show whether the link is built that way. Copy the anchor element from the page source and store it as text.
How soon after publication should the archive capture be taken?
The Internet Archive help page notes a lag of 3 to 10 hours between a crawl and the capture appearing in the Wayback Machine. Practical suggestion: request the capture on the first check day and confirm on a later check day that it displays properly, marking the sheet as awaiting archive in between.
What if the Wayback Machine cannot save the page?
The help page lists password protection, robots.txt blocks and owner exclusion requests as reasons pages may be missing. Note the failed attempt and its date in the row, and rely on the screenshot and the source line for that publication.
Does the Search Console Links report show nofollow?
No. The help page states that the report does not specify whether a link is marked nofollow. Read the rel value from the source line instead.
Does an image link carry anchor text?
Google's crawlable links documentation says that for images used as links, Google uses the alt attribute of the img element as anchor text. For an image link, the source line should therefore include the img element and its alt value.
Who should own the folder?
The documentation has no statement on this. Practical suggestion: assign ownership to a role, such as the operations lead, rather than a person, and keep the folder in shared storage so that it outlives any one employee.
Conclusion
A publication proof file is six items that each answer one question: the order record says what was expected, the screenshot shows how the page looked at the check, the source line shows how the link was written, the archive capture shows when a third party saved a copy, and the Search Console export shows a sample of what Google has found. None of them alone proves the original publication date, indexing or that the link will stay in place. A single shared folder named after the order number, a fixed file pattern and one log sheet let a ten-article package and a one-article order live in the same system.
If you want to follow a package order with this folder structure, you can browse the guest post packages.
Post a comment
Please log in to leave a comment. Login