If you go looking for the ResearchGate Score (“RG Score”) on a profile today, it is not there. ResearchGate retired the RG Score years ago and replaced it with the Research Interest (RI) Score as the platform’s primary engagement metric. Because the RG Score is still referenced in older CVs, tenure files, and third-party “academic SEO” guides, it is worth being precise about what changed, why, and how the RI Score is actually calculated — since the two metrics measure genuinely different things and are not interchangeable.
What happened to the RG Score
The RG Score was ResearchGate’s original composite reputation metric, launched in 2011. It combined a researcher’s publications, questions, answers, and other on-platform activity into a single number, but ResearchGate never published the underlying formula. That opacity became the metric’s defining criticism: it could not be independently verified, it appeared possible to inflate through heavy Q&A participation rather than research output, and it carried no mechanism for negative signals such as retractions or research-integrity findings.
ResearchGate discontinued the RG Score around August 2022, stating that the metric no longer met the standards — intuitiveness, transparency, robustness, and relevance — it wanted for profile metrics going forward. (ResearchGate’s own product-update post announcing the change was not independently re-verifiable at the time of writing due to an access restriction; this timeline and rationale are corroborated across multiple secondary accounts, including a 2024 peer-reviewed analysis in Open Information Science, and should be treated as well-sourced but not primary-confirmed.) In its place, ResearchGate elevated a metric it had already been running quietly alongside the RG Score: the Research Interest (RI) Score.
What the RI Score measures
The RI Score is designed to track how the scientific community is actively engaging with a researcher’s published output, rather than rolling in general platform activity. Per ResearchGate’s own help documentation, it is built from four weighted interaction types on a researcher’s publications:
- Citations — weighted 0.5, the heaviest signal
- Recommendations — weighted 0.25
- Full-text reads (a publication viewed or downloaded in full) — weighted 0.15
- Other reads (e.g., a summary view or figure click) — weighted 0.05
Each research item on a profile earns its own RI Score from these weighted interactions; a researcher’s overall, profile-level RI Score is the sum of every item’s individual score. ResearchGate also displays a percentile ranking so a researcher can see how their RI Score compares to peers, and the score updates as new interactions accrue rather than being fixed at publication.
What counts, and what’s deliberately excluded
ResearchGate’s documentation specifies several exclusions built into the RI Score to reduce the kind of gaming the old RG Score was criticized for:
- Self-citations do not count.
- Reads by the item’s own author(s) do not count.
- Interactions from logged-out or unauthenticated users do not count.
- Repeated reads or recommendations from the same individual within a short window (documented as roughly one week) are not counted multiple times.
- Bot and crawler traffic is filtered out.
This is a narrower, more citation-weighted design than the old RG Score, and it is closer in spirit to platform-native engagement metrics like Altmetric’s Attention Score than to a bibliometric index. It is still fundamentally different from both: unlike the h-index, it is not citation-count-only and is not comparable across platforms; unlike Altmetric, it draws only on ResearchGate’s own logged interactions, not the broader web, news, and social media signals Altmetric aggregates.
Why this matters for research administrators and researchers
Two practical points follow directly from the mechanics above.
First, the RG Score and RI Score are not on the same scale and were never meant to be converted between one another. A profile that previously advertised a “high RG Score” cannot map that number onto an equivalent RI Score — the underlying formula changed entirely, not just the label. CVs, biosketches, or promotion files that still cite an “RG Score” are citing a metric ResearchGate itself no longer calculates or displays.
Second, like every platform-specific engagement metric, the RI Score should not stand in for peer-reviewed impact measures in formal evaluation contexts such as tenure and promotion, grant review, or hiring. This is the same caution research-assessment reform initiatives raise about attention scores and citation counts generally: engagement is not equivalent to quality or rigor. Institutions that have signed the San Francisco Declaration on Research Assessment (DORA) or that follow CoARA commitments generally discourage using any single platform-generated number, ResearchGate’s included, as a proxy for research quality in formal review.
How the RI Score compares to other profile and citation metrics
ResearchGate’s RI Score sits in a crowded field of researcher-level metrics, most of which measure different things:
- The h-index (native to Scopus, Web of Science, and Google Scholar) measures sustained citation impact across a body of work and is comparable, with caveats, across platforms that calculate it consistently.
- Altmetric’s Attention Score tracks online attention — news coverage, policy citations, social media mentions — rather than scholarly citation or platform reads.
- Citation impact metrics generally are field- and time-normalized in ways a raw RI Score is not.
- The RI Score is unique to ResearchGate, cannot be exported to or verified against an external database, and depends entirely on activity that happens to occur on ResearchGate’s own platform — a researcher who is highly cited but rarely uploads full text to ResearchGate will show a modest RI Score despite strong citation impact elsewhere.
For identifier and metrics governance purposes, none of this changes how a researcher’s identity should be tracked: an ORCID iD remains the interoperable, platform-independent identifier for linking a researcher to their outputs, and a Scopus Author ID or Google Scholar profile remain the more portable references for citation-based metrics. A ResearchGate RI Score is best treated as one platform’s supplementary engagement signal, not a system-of-record metric.
Frequently asked questions
When did ResearchGate remove the RG Score?
Multiple secondary and academic sources place the removal around August 2022, following an announcement earlier that year that the RG Score no longer met ResearchGate’s own bar for transparency and robustness. ResearchGate’s original announcement post could not be independently re-accessed at the time this page was written; treat the exact date as well-corroborated but not primary-source-confirmed.
Is there a way to see my old RG Score?
No. ResearchGate does not display or calculate the RG Score for any profile, current or historical. It has been fully replaced by the RI Score.
Can I convert my RG Score to an equivalent RI Score?
No. The two metrics use unrelated formulas and inputs, and ResearchGate has not published a conversion. A high RG Score under the old system does not imply a high RI Score under the current one.
Does a high RI Score mean a researcher is highly cited?
Not necessarily. Citations carry the heaviest weight (0.5) in the RI Score formula, but reads and recommendations also contribute, and the score only counts interactions that happen on ResearchGate itself. A researcher’s citation record on Scopus, Web of Science, or Google Scholar may look quite different from their ResearchGate RI Score.
Should institutions use the RI Score in tenure or grant review?
General good practice in responsible research assessment, consistent with DORA and CoARA commitments, is not to rely on any single platform-generated engagement metric as a proxy for research quality in formal evaluation. The RI Score can be a supplementary data point but should not substitute for peer review or field-appropriate citation analysis.
