How to Produce a Citation Report from Google Scholar, Scopus and Web of Science

Why a citation report matters and what it must show

A USCIS officer reviewing your petition has no login to Scopus, no institutional access to Web of Science, and no obligation to search Google Scholar on your behalf. Whatever citation evidence you submit has to stand on its own — the exhibit itself must contain enough detail that the officer can verify your claims without leaving the file. This is the central design principle behind a citation report: it is not a summary you assert, it is a record you prove.

What the exhibit needs to demonstrate

A complete citation report shows three things at once:

1. Total citation counts for each of your key publications, pulled from more than one database, so the numbers can be cross-checked against each other.
2. Citing-author independence — evidence that the people citing your work are not you, your coauthors, or your immediate collaborators, since a count inflated by self-citation is not the same claim as one built from unrelated researchers.
3. Traceability — every number in your summary table must be backed by a corresponding screenshot or exported file showing exactly where it came from and when.

The failure mode to avoid

A table of citation counts with no underlying screenshots, no database names, and no retrieval dates reads as an unsupported assertion of influence. That gap — a claim without the record behind it — is exactly the kind of thing that draws a Request for Evidence. The remedy is procedural, not persuasive: build the paper trail before you write the narrative that relies on it.

Pulling counts from Google Scholar

Step 1: Locate or build a profile

If the petitioner already has a Google Scholar profile (scholar.google.com/citations), sign in and open it — it lists every indexed paper with a running citation count. If no profile exists, do not create one solely to generate this exhibit; instead search each paper individually by exact title in Google Scholar's main search bar.

Step 2: Read the 'Cited by' number

Beneath each search result or profile entry sits a "Cited by N" link. That number is the total citation count Scholar currently reports for that specific paper. Record it alongside the retrieval date — Scholar counts shift week to week as it recrawls the web.

Step 3: Open the citing list and capture it

Click "Cited by N" to load the full list of citing works. Screenshot the list (multiple pages if needed) or use a browser print-to-PDF function to save the complete page, including the URL bar showing the search query and date. Save this as a source file referenced by the summary table, not just a number typed into a spreadsheet.

Step 4: Flag Scholar's inclusiveness

Google Scholar indexes preprints, theses, conference abstracts, and other non-peer-reviewed material alongside journal articles. Do not filter these out silently — note in the report that the Scholar count includes such sources, so a reviewer comparing it to Scopus or Web of Science understands why the figure is higher rather than assuming an error.

Pulling counts from Scopus

Scopus indexing differs from Google Scholar in scope and access model, and the report should reflect that difference rather than blend the two silently.

Locating the record

1. Go to scopus.com and search by document title, or by author name if a full publication list is needed.
2. If the petitioner has a Scopus Author Identifier (a unique numeric ID Scopus assigns to disambiguate authors with similar names), search by that ID instead — it avoids conflating the petitioner's work with someone else's under the same name. The ID appears on the petitioner's Scopus author profile page and should be noted in the report for reference.
3. Open each document record and locate the "Cited by" count near the top.

Exporting the citing list

1. Click the "Cited by" link to open the list of citing documents.
2. Select all entries, then use the export button (usually rendered as an arrow or "Export" label) to download the list as CSV or to a citation manager format.
3. Save the CSV alongside a screenshot of the citing-document list and the summary count.

If full access is unavailable

Scopus requires an institutional or personal subscription for full search and export. Without one, go to the petitioner's Scopus author profile (accessible via a free Scopus preview or through ORCID linkage) to view a citation overview — total citations and an h-index — even though the detailed citing-document list may be restricted. Note in the report which level of access was used, since it affects what can be verified from the exhibit alone.

Pulling counts from Web of Science

Web of Science (WoS), maintained by Clarivate, requires an institutional or personal subscription. If access is available through a university library or employer, use the Core Collection search rather than the broader "All Databases" option, since the Core Collection is the citation index most consistently referenced in bibliometric contexts.

Searching and pulling counts

1. Log in and select Web of Science Core Collection.
2. Search by author name (using a Researcher ID/Publons profile if the petitioner has one, to disambiguate common names and consolidate publications) or by exact article title.
3. Open each record and note the Times Cited count, displayed on the right side of the record page.
4. For a consolidated view across all of the petitioner's papers, select the relevant records and use Create Citation Report. This generates a summary table with per-article and aggregate citation counts, plus an h-index calculated from WoS data alone.
5. Export the citation report using the Export to Excel or Save to text file option, and also capture a screenshot of the report page showing the retrieval date.

Why the number will differ

WoS indexes fewer journals and conference proceedings than Scopus, and applies stricter inclusion criteria for source coverage. Expect Times Cited totals to be lower than both Google Scholar and Scopus for the same article — this is a known feature of database coverage differences, not an error requiring correction.

Reconciling the three counts into one table

Once counts are pulled from all three databases, consolidate them into a single table rather than presenting three separate exhibits. Reviewers should be able to see, for each publication, how the numbers compare side by side.

Worked example

List every publication the petition relies on as its own row, in the same order they appear elsewhere in the petition, so a reviewer can cross-check against the CV or the underlying evidence exhibits.

Why the numbers differ

Google Scholar, Scopus, and Web of Science index different sets of sources: Scholar sweeps in preprints, theses, and grey literature; Scopus and Web of Science restrict themselves to indexed peer-reviewed journals and conference proceedings, but their journal coverage lists are not identical. As a result, the three columns will almost never agree, and a gap between them is not a data error to reconcile or explain away — it is a predictable consequence of differing index scope. State this plainly near the table rather than letting a reviewer wonder why the figures conflict.

Filtering out self-citations and identifying independent citing authors

Raw citation counts include citations from the petitioner's own later papers and from coauthors, which do not demonstrate independent recognition. Each database's citing-list export must be reviewed manually to identify these.

Step 1: Pull the citing-author list

From Google Scholar's "Cited by" page, Scopus's exported CSV, and the Web of Science citation report, extract the author names for every citing work.

Step 2: Cross-check against the petitioner's coauthor history

Compile a list of every name that has appeared as a coauthor on the petitioner's own papers. Compare each citing work's author list against this coauthor list.

Step 3: Mark each citation in a spreadsheet

Add a column labeled "Citation type" next to each citing-work row, and mark it "Self" if the petitioner or a coauthor appears among the citing authors, or "Independent" if not. Keep a running count of each category per source database.

Step 4: Carry the split into the summary

Report both the total citation count and the independent-citation count for each publication, rather than presenting only the combined figure.

This manual review is unavoidable — none of the three databases performs this filtering automatically. Officers reviewing citation evidence may examine whether a large total count is substantially composed of self-citations, so the underlying breakdown should be available and consistent with the totals claimed elsewhere in the exhibit.

Formatting the exhibit and naming the files

File naming

Use a consistent scheme tied to the exhibit list in the petition index. For example, if the citation report is Exhibit J, name the main PDF Exhibit_J_Citation_Report.pdf. Name each underlying screenshot or export by database and retrieval date, kept as subordinate files or merged into the same PDF as an appendix: Exhibit_J_GoogleScholar_2025-03-14.pdf, Exhibit_J_Scopus_2025-03-14.csv (convert to PDF before filing), Exhibit_J_WoS_2025-03-14.pdf. Consistent dates across filenames let a reviewer confirm all three pulls were taken on the same day.

Document order

Structure the exhibit so a reader can verify without hunting:

1. Cover page stating what the exhibit contains and the retrieval date range.
2. The reconciled summary table (publication, year, three citation counts, date retrieved).
3. A short note on methodology — how self-citations were identified, which profile IDs were used.
4. Database screenshots or exports as an appendix, ordered by database, each page showing the search query, the citation count, and the date visible in the browser or export header.

Final checklist before filing

• Re-pull all three counts within a few days of filing; do not reuse a report pulled months earlier.
• Confirm every number in the summary table matches a corresponding screenshot in the appendix.
• Confirm each screenshot shows a visible date and search term.
• Confirm file names match the exhibit list and any table of contents.
• Save a dated copy of the raw exports separately from the formatted PDF, in case a discrepancy needs to be checked later.

Common mistakes that draw RFEs

The following patterns show up repeatedly in petitions that draw a request for evidence on citation impact.

1. Unsourced numbers. A sentence stating "cited over 200 times" with no database named, no retrieval date, and no exhibit reference cannot be verified by a reviewer who has no database access. Every count in the petition narrative must trace to a specific line in the citation report exhibit.

2. Screenshots that don't match the claimed totals. If the narrative says 214 citations but the attached Scopus screenshot shows 198 taken on a different date, the discrepancy itself becomes the issue, not the underlying number. Re-pull and re-check before filing.

3. Undisclosed self-citation. Presenting a raw total that includes citations by the petitioner or coauthors without flagging them invites scrutiny of the whole exhibit's reliability, not just that figure.

4. Relying on a single database. A report built only from Google Scholar, with no Scopus or Web of Science cross-check, reads as selective rather than comprehensive, especially given how differently the three platforms index.

5. Stale data. Counts pulled eight months before filing, with no re-verification, will not match what an officer could find if they searched the same title today. Databases update continuously; a report should carry a retrieval date close to the filing date, and any large gap should be closed with a fresh pull rather than left as-is.