Get AI summaries of any video or article — Sign up free
You are choosing keywords for a paper WRONG! Do THIS instead... thumbnail

You are choosing keywords for a paper WRONG! Do THIS instead...

Andy Stapleton·
5 min read

Based on Andy Stapleton's video on YouTube. If you like this content, support the original creators by watching, liking and subscribing to their content.

TL;DR

Prioritize keyword alignment in the title and abstract because search engines use that text as core metadata for matching queries.

Briefing

Choosing the right keywords for a research paper isn’t a cosmetic step—it determines whether search engines can match the work to what scientists are actively typing. The biggest leverage comes from placing high-demand, field-relevant terms in the paper’s title and abstract, where search engines pull metadata used by Google Scholar and other indexes. Keywords typed into the “keyword” field during submission matter, but they’re often treated like an afterthought; the more reliable path to discoverability is aligning the paper’s most prominent text with the queries people already use.

A common mistake is selecting phrases that are technically accurate but too niche for anyone to search. Even researchers in the same specialty may not use the exact wording found in a paper’s methods or jargon. The practical goal is to aim for the “middle” of a search pyramid: broad enough to capture meaningful search volume, but specific enough to still point to the actual research contribution. Research lives at the top of the pyramid—too new or too obscure to be searched directly—so authors should instead target the terms that sit beneath it: the problems, concepts, and commonly searched subtopics that lead readers to the paper.

To find those terms, the transcript recommends a simple workflow built around how autocomplete and question prompts behave in search. Start with Google Scholar: type the field, then iterate through variations while watching what Google Scholar suggests in autocomplete. Because autocomplete is driven by frequent queries, it’s treated as a strong signal that real people search for those phrases. Keep a running list in a spreadsheet or document as the search terms evolve.

Next, repeat the exploration on regular Google, with the browser logged out to avoid personalized search history. Then use Answer the Public to map what people ask about a topic. If Answer the Public returns few or no results, that’s a sign the query is too far up the pyramid—go broader until you see enough traction. The transcript gives an example from solar-cell research, where “transparent semiconductor solar cells” was too narrow, while “organic photovoltaic solar cells” and “solar cells” provided better starting points.

Once a keyword list exists, effectiveness comes down to two criteria: search demand and how naturally the phrase fits into the title and abstract. Exact search-volume data is hard to obtain for academic queries because major tools are optimized for advertising keywords, not low-volume scientific terms. Still, autocomplete presence, Answer the Public results, and “People also ask” style prompts can serve as practical proxies. For additional estimates, Google Keyword Planner and tools like Ubersuggest can help, though they may be less reliable for science.

Finally, the transcript emphasizes a balancing act: high-volume terms must still read like normal academic English and pass a common-sense check for peer review. Keyword stuffing is discouraged; the phrase should integrate smoothly into sentences and reflect the actual research. Used consistently across papers, conference presentations, and any work with a title and abstract, this approach can make research easier to find—and easier to cite. It also suggests a secondary use: letting search demand guide future research directions by identifying which related questions attract more attention, then steering toward the most searchable branch of inquiry.

Cornell Notes

Discoverability hinges on matching what researchers search for with what appears in a paper’s title and abstract. Keywords chosen from the “keyword” field alone are often an afterthought; the transcript argues that search engines use title/abstract text as key metadata for results in Google Scholar and Google. The method is to build a keyword list by probing Google Scholar autocomplete and Google suggestions, then broaden or narrow using Answer the Public until the topic sits in the “middle” of the search pyramid (searched enough to matter, not so niche that nobody looks). Finally, pick phrases that both appear in search prompts and fit naturally into academic writing, avoiding keyword cramming. This can increase citations by improving how often the work is found.

Why do title and abstract keywords matter more than the submission “keyword” field?

Search engines rely heavily on metadata extracted from prominent text. The transcript highlights that Google Scholar and other indexes use the title and abstract to match papers to queries, while the journal “keywords” box is frequently treated as optional or filled with random terms. The practical takeaway is to prioritize search-aligned wording in the title and abstract first, then use the keyword field (if available) as an additional reinforcement.

How can an author tell whether a scientific phrase is actually searched for?

Autocomplete and question prompts act as practical signals. The transcript recommends typing into Google Scholar and watching autocomplete suggestions; because Google offers those suggestions, it implies meaningful query frequency. It also recommends checking Google (logged out) and using sections like “People also ask,” plus Answer the Public to see whether a topic generates real question patterns.

What does “search pyramid” mean, and where should keywords land?

The transcript frames search behavior as a pyramid: common terms at the base are searched often; niche terms sit in the middle; brand-new research concepts at the top aren’t searched directly because people don’t know them yet. Authors should target the middle—terms that are actively searched and still closely connected to the paper’s contribution—rather than ultra-niche jargon that only insiders recognize.

What should be done if Answer the Public returns few or no results?

Few or no results suggest the topic is too far up the pyramid—too narrow to generate search demand. The transcript advises going broader and trying related terms until enough results appear. It gives an example in solar-cell research: “transparent semiconductor solar cells” was too narrow, while “organic photovoltaic solar cells” and “solar cells” produced better starting points.

How should search demand be evaluated when exact volumes are hard to get for science keywords?

The transcript notes that tools like Google Keyword Planner and Ubersuggest are designed for advertising-style keywords and may be less accurate for scientific queries with low volumes. Instead, it recommends using proxies: whether a phrase appears in autocomplete, whether it shows up in Answer the Public or “People also ask,” and whether it can be incorporated into the title/abstract without sounding forced.

What is the rule for choosing between a high-volume keyword and a natural-sounding one?

The transcript treats natural fit as non-negotiable. A keyword should have enough search traction (autocomplete/questions) and also integrate smoothly into academic sentences. If a high-volume phrase doesn’t fit well or reads like keyword stuffing, it should be dropped in favor of a slightly less dominant term that still accurately reflects the research and would pass peer review.

Review Questions

  1. If a keyword appears in autocomplete but cannot be integrated naturally into the abstract, what decision should be made and why?
  2. How would you adjust a too-niche research topic using Answer the Public results?
  3. What are three different ways mentioned for generating a keyword list before selecting the final title/abstract wording?

Key Points

  1. 1

    Prioritize keyword alignment in the title and abstract because search engines use that text as core metadata for matching queries.

  2. 2

    Avoid selecting ultra-niche phrases that only appear in a paper’s jargon; target terms that people actually search for.

  3. 3

    Build a keyword list by iterating through Google Scholar autocomplete suggestions and recording promising phrases.

  4. 4

    Use Answer the Public to find question-based demand; broaden the topic if results are sparse or absent.

  5. 5

    Choose keywords using a two-part filter: evidence of search demand (autocomplete/questions) and natural integration into academic writing.

  6. 6

    Use keyword tools like Google Keyword Planner or Ubersuggest cautiously for science, since they may be optimized for advertising keywords rather than low-volume academic queries.

  7. 7

    Apply the same keyword strategy across papers, conference presentations, and any work with a title and abstract to improve discoverability over time.

Highlights

The most reliable discoverability gains come from putting searchable terms in the title and abstract, not treating the journal keyword box as an afterthought.
Answer the Public can act like a “search pyramid” compass: if it returns little, the topic is too narrow—go broader until demand appears.
Autocomplete and “People also ask” function as practical proxies for search volume when exact academic keyword volumes are hard to measure.
Keyword stuffing is a losing strategy; the best phrase is the one that both attracts searches and reads naturally enough to pass peer review.

Topics

  • Keyword Selection
  • Research Discoverability
  • Google Scholar
  • Answer The Public
  • Title and Abstract

Mentioned