A coloured donut appears beside your paper on the journal's page, with a number in the middle. Last week it said 4. Today it says 31. Is that good? What happened?
That number is an altmetric, and it is easy to misread. Here is what it counts, what it leaves out, and how to turn it into something more useful than a number.
What Altmetrics Are
Altmetrics are signals of the attention a research output gets outside formal citations: news stories, blog posts, policy documents, Wikipedia references, social media posts and similar mentions. Citations take years to build up. Mentions like these can start the week a paper comes out.
The one most researchers meet first is the Altmetric Attention Score, the number in the donut. Altmetric describes it as an indicator of the amount of attention an output has received. This post uses it as the worked case because its method is documented publicly.
How the Score Is Weighted
According to Altmetric's page on how the score is calculated, the score is a weighted count, because some sources reach more people than others. At the time of writing (October 2026), the default (minimum) weightings include:
| Source | Default weighting |
|---|---|
| News | 8 |
| Blog | 5 |
| Podcast | 5 |
| Policy document (per source) | 3 |
| Clinical guideline (per source) | 3 |
| Patent | 3 |
| Wikipedia | 3 |
| Peer review (Publons, PubPeer) | 1 |
| Bluesky, X, Reddit, YouTube | 0.25 |
| Facebook (curated list of public pages only) | 0.25 |
| LinkedIn (not trackable since 2014, historical data kept) | 0.5 |
| Mendeley readers | 0 |
| Dimensions citations | 0 |
Two details in that list are easy to miss. LinkedIn is listed as not trackable since 2014, so new posts there do not add to the score. And Mendeley readers and Dimensions citations carry no weight at all.
The same page says the score shown on a badge is always a whole number, so small contributions can be rounded up, and that the algorithm takes many other factors into account. You cannot reproduce the score by adding up mentions.
The Modifiers Behind the Number
Altmetric's page on score modifiers explains several rules that make two papers with similar mention counts end up with different scores:
- News outlets are assigned tiers, so a national outlet contributes more than a niche one.
- Only the first story from a given news source counts toward the score.
- Wikipedia is static: one or many Wikipedia mentions add 3 in total, partly to prevent gaming.
- Policy documents are scored per source and patents per jurisdiction, so ten documents from one source add the same as one.
- On Bluesky and X, a repost counts less than an original post, and modifiers reduce the weight of accounts that post about many papers or many papers from one journal.
- A post from a journal's own account counts for less than one from an unconnected researcher sharing the paper.
That last point matters if your paper's score comes mostly from the publisher's own promotion: it is designed not to count for much.
What the Score Cannot Tell You
Altmetric's page on how outputs are scored lists its own limits, and they are worth taking literally:
- It measures attention, not quality. People pay attention to papers for all sorts of reasons, not all of them positive.
- It only tracks public attention. Journal clubs, email threads and private forums are invisible to it.
- A news story or blog post is only counted if it links to the publication (journal article, DOI, PubMed ID or institutional repository) or meets Altmetric's text-mining criteria.
- The score only makes sense when comparing like with like, such as within a single discipline, because norms for attention differ between fields.
A low score does not mean nobody read your paper. A high one does not mean the attention was approving.
Reading Your Paper's Details Page
Clicking the donut opens a details page. It shows a "Mentioned by" list with the number of mentions of each type, tabs for each source, and, for Altmetric Explorer and Bookmarklet users, a timeline of attention and of the score over time. Here is how to read the main parts.
| What you see | What it means | What to do with it |
|---|---|---|
| News mentions | Counted by outlet tier; only the first story per outlet counts | Save the outlet, date and headline for your records |
| Policy mentions | Scored per source; Altmetric curates sources meant to influence policy or practice | Note which organisation cited you and in what document |
| Bluesky or X posts | Low weight, adjusted for who posted | Look at who shared it: practitioners, other fields, the public |
| Mendeley readers | Shown, but not counted in the score | Treat as a separate signal, not part of the number |
| Score history (Explorer and Bookmarklet users) | At the time of writing (October 2026): daily since July 25, 2025; monthly snapshots from August 1, 2022 for outputs also tracked in Dimensions; nothing earlier | Line up recent daily changes with when you shared the paper |
The policy details come from Altmetric's policy documents page, which says it tracks sources such as government guidelines and reports, policy institutes, advisory committees and international development organisations.
What to Do With the Numbers
Altmetric itself says the most important part of a report is the qualitative data, not the score. That suggests a simple routine:
- When you share your paper anywhere, link to the DOI or the journal page, since mentions without a link to the publication may not be counted.
- Once a month in the first few months, open the details page and copy out the specific mentions that matter: which outlet, which policy body, which clinicians or practitioners.
- Keep those in one file with dates, so they are ready for annual reviews, promotion files and grant reports.
- Quote the specifics, not just the score, and say what kind of evidence each item is.
This fits current thinking on research assessment. The San Francisco Declaration on Research Assessment (DORA) asks committees to assess research on its scientific content rather than publication metrics, and recommends that researchers use a range of article metrics and indicators in personal or supporting statements as evidence of the impact of individual outputs. Our guide to evidence of research impact for grant reports shows how to write those items up.
Do not try to game the number. Altmetric's modifiers are built to discount promotional patterns, and a score built from your own reposts tells a reviewer very little.
A Second Signal You Control
An attention score tells you what happened in public, on sites Altmetric tracks. It cannot tell you what happened when someone actually opened your work. That is where Amplify adds a different kind of evidence.
Amplify turns one paper into a video abstract, a vertical short, an audio brief and a one-page infographic, generated from the PDF you upload, each carrying the title, the authors and a link back to the paper of record. An embeddable widget puts all of them beside the paper on your own faculty or lab page, and its analytics show reads, listens, questions asked, languages and which format moved people. Those numbers sit alongside your altmetrics as their own record, useful for grant reports and broader-impact sections, and the widget and analytics are included in the per-paper price.
Read the Mentions, Not Just the Donut
An altmetric score is a rough, field-dependent count of public attention. The mentions behind it are where the value is: who noticed your work, where and why. Record those, link your paper properly whenever you share it, and keep a second record of how people engaged with it on your own page.
