“A lot of research projects start with a deceptively simple question: Where should someone invest attention, time, money, or credibility?”

That question may sound strategic, but underneath it is a research problem. Before anyone can make a defensible recommendation, they need to know what already exists, what is credible, what is speculative, where the gaps are, and which claims are supported by evidence rather than vibes.

This is the kind of work I enjoy doing: targeted research that turns a messy issue into a decision-ready paper. The subject matter varies, but the method is fairly consistent. It involves formal literature searches, organizational research, standards and policy review, database work, source tracking, and—when possible—direct outreach to people operating the systems being studied.

It also involves a fair amount of humility. Research rarely gives you perfect certainty. The goal is not to know everything. The goal is to know enough, and to know how you know it.

You can Subscribe and Listen to the Podcast on Apple Podcasts, or wherever you listen to Podcasts.

And be sure to leave me a Rating and Review!

Start With the Question, Not the Search Box

The worst way to start a research project is to open a browser and type in a broad term.

That feels productive for about twenty minutes. Then you have fifty tabs open, three reports from organizations with suspiciously similar names, a white paper from 2018, a blog post that cites a dead link, and a growing sense that maybe the Internet was a mistake. It’s the librarian’s version of doomscrolling.

Before searching, I try to define the working question as precisely as possible. Not necessarily the final thesis, but the thing the research needs to answer.

For example:

• What infrastructure is missing?
• Which organizations are actually active in this space?
• What has already been funded?
• What standards, regulations, or deployment patterns shape what is possible?
• Where are the gaps between aspiration and implementation?
• What evidence would make a recommendation credible?

I want to highlight that last question. A defensible paper is not just a well-written opinion. It needs a visible evidence trail. The reader should be able to understand why certain sources were considered, why others were not central, and how the final recommendations connect to the research base.

Use Google Scholar, But Do Not Worship It

Google Scholar is usually one of my early stops, especially when I need to understand academic framing, established terminology, or prior work in adjacent fields.

It is useful for identifying:

• frequently cited papers;
• recent articles that summarize a field;
• debates over definitions;
• older foundational work that current projects still rely on;
• terminology that practitioners may not use consistently.

But Google Scholar has limits. It is not a complete map of what is happening in practice. Academic papers often lag behind deployment. They may describe theory beautifully while saying very little about how systems are procured, governed, maintained, funded, or used by actual people. They may also focus on concepts that are useful in scholarship but not quite aligned with how implementers talk about their work.

So I use Scholar as a grounding tool, not as the whole research universe.

A good pattern is to search for broad concepts first, then narrow them based on the terms that keep appearing. I also look at who is being cited, which disciplines are represented, and whether there are clusters of work that do not seem to talk to each other. That silence can be useful. If computer science, political theory, public administration, and standards communities are all using related terms differently, that tells you something.

Usually, it tells you that the paper will need to spend some time translating across worlds.

Search the Organizations That Shape the Field

After academic grounding, I turn to the institutions that shape policy, funding, standards, and deployment.

Depending on the topic, that may include:

• international organizations;
• development agencies;
• civil society networks;
• NGOs;
• standards bodies;
• public-sector innovation teams;
• philanthropic foundations;
• think tanks;
• industry consortia;
• regulators;
• procurement or implementation bodies.

This is where search becomes more deliberate. A generic web search may surface the loudest voices, not necessarily the most relevant ones. So I search within specific organizational sites and document libraries. I look for reports, issue briefs, meeting minutes, calls for proposals, grant announcements, regulatory guidance, technical specifications, implementation guides, and public consultation responses.

This work is not glamorous. It is a lot of clicking through PDF libraries and discovering that an organization’s “resources” page has not been maintained since 2021. Sometimes the best source is buried three layers deep under a title that sounded aggressively unhelpful.

But this layer of research is essential because it shows what institutions are actually saying and doing. It also helps distinguish between concepts that are fashionable and capabilities that are fundable, deployable, or constrained by governance realities.

These distinctions are not optional. A recommendation that ignores institutional context may be elegant, but it will also be brittle.

Casual Searching Has a Place

Not every useful lead comes from a formal database. Casual searching is part of the method, too. Search engines, conference agendas, GitHub repositories, project websites, blog posts, podcasts, mailing lists, and community discussion threads can all help identify active work. The trick is to treat casual search as lead generation, not proof.

A blog post may point to an interesting organization. A conference session may reveal who is actively working on a deployment. A GitHub issue may show where implementation pain is showing up. A project website may explain how a platform describes itself to users, which can differ sharply from how academics or funders describe the same category.

These sources are useful, but they need to be handled carefully. Some are current and credible. Some are marketing. Some are abandoned. Some are written as if the future has already happened, which is a charming habit of technology communication and a terrible basis for recommendations. In other words, these sources are great for trends and crap for details.

When casual searching turns up something important, I try to verify it elsewhere.

Build the Spreadsheet Early

A source spreadsheet is not administrative overhead. It is part of the research.

I usually start one very early, even before I know exactly what the final structure of the paper will be. At minimum, I want to track:

• source title;
• organization or author;
• publication date;
• URL or citation;
• source type;
• relevant topic area;
• key claim or finding;
• why it matters;
• limitations or concerns;
• whether it has been used, rejected, or held for later;
• follow-up questions.

For more complex projects, I add columns for geography, legal or regulatory implications, technical maturity, implementation status, contact information, outreach status, and recommendation mapping.

That last one is especially useful. If the eventual paper includes recommendations, I want to know which sources support each recommendation. Otherwise, it is very easy to end up with a beautifully organized bibliography that does not actually support the argument being made.

The spreadsheet also prevents a common research failure: rediscovering the same source five times and thinking each time that it is new. Not that I would ever do this. Obviously. Hypothetically.

A good spreadsheet gives you memory. It also gives you a way to audit your own thinking. If a source seemed important early but disappeared from the final paper, why? Was it superseded? Too weak? Duplicative? Outside scope? Useful background, but not central?

Those are not trivial questions. They are how a research project becomes defensible.

Verify Details in Databases

Once the research starts pointing toward specific organizations, grants, programs, or institutions, I move into verification. This is where databases matter.

Depending on the project, that may mean checking nonprofit records, foundation databases, public grant databases, company registries, tax filings, standards participation records, procurement portals, public consultation archives, regulatory databases, or project repositories.

This stage is slower, but it changes the quality of the work. It helps answer questions like:

• Is this organization still active?
• Is the project current or historical?
• Who funds it?
• Is it independent, acquired, merged, dormant, or rebranded?
• Has it received grants before?
• Is there evidence of implementation, or only aspiration?
• Are the public claims consistent with filings, reports, or observable activity?
• Is there a real contact path?

This is also where convenient narratives often start to fall apart.

A project that looks prominent may not be operationally active. A widely cited organization may have changed ownership or mission. A promising platform may have strong technical work but weak governance documentation. A potential grantee may be highly relevant substantively but not a good fit for the funding mechanism under consideration. I found one organization with quite a bit of material online, but a deeper dive showed that it was purchased only a month or so ago. Search results did not reflect that (yet).

These are not reasons to discard the source automatically; they are reasons to be precise. Defensible research does not require every candidate to be perfect. It requires the paper to describe candidates accurately and avoid pretending uncertainty is certainty.

Outreach Helps, But It Is Hard

At some point, desk research reaches its limit. Public materials can tell you what an organization says it does. They often cannot tell you how something actually works, what constraints operators face, what has changed since the last report, or whether a program is still actively maintained.

That is where outreach comes in.

I usually identify a small set of organizations or individuals and send concise questions. The goal is not to conduct a full interview study unless that is the scope of the project. Instead, the goal is to verify details, fill gaps, and understand whether the public record reflects current reality.

This sounds straightforward. I wish it were straightforward. Alas.

Outreach for this kind of research is basically cold-calling with better grammar. Many messages receive no response. Some contact addresses bounce. Some organizations have intake forms that appear to be where questions go to achieve enlightenment through silence. Sometimes the best contact path is through LinkedIn, Mastodon, a conference connection, or a person who knows a person.

Even when responses come in, they may be partial. That is fine. A partial answer from an operator can still be extremely useful, especially when it clarifies identity, eligibility, governance, funding, or implementation details that public materials gloss over.

But outreach evidence should be documented carefully. Track who was contacted, when, how, what was asked, what came back, and how the response changed the analysis. If no one responds, that is also a finding of sorts, though it should be handled cautiously. Silence is not proof. It is a constraint on what you can say in your outcomes.

Use AI as an Assistant, Not a Research Substitute

I use AI in research work, but not as the authority.

AI can be useful for organizing notes, extracting recurring themes, identifying gaps across a source spreadsheet, suggesting alternative search terms, comparing draft recommendations against source categories, and helping turn messy findings into coherent prose.

It is especially good at asking, “You seem to have a lot of material in one category and very little in another. Is that intentional?” That kind of prompt can catch an imbalance before it becomes a structural problem in the paper.

But AI is not a substitute for source review. It cannot tell you, on its own, whether an organization is still active, whether a grant database entry maps to the same entity, whether a policy document has been superseded, whether a technical specification is implemented in practice, or whether a claim is being repeated because it is true or because everyone cites the same original source. The human researcher still has to make judgment calls.

AI can help keep the work organized. It can help notice patterns and stress-test an outline. It might help identify places where a claim needs a citation or where a recommendation outruns the evidence.

It should not be allowed to launder uncertainty into confidence, regardless of how confident it is in tone.

Write While Researching, But Do Not Fall in Love With the Draft

I rarely wait until all the research is complete before drafting. Early writing helps expose weak spots. If I cannot explain a point clearly, the issue may be the prose, but more often it is that the evidence is still fuzzy. Drafting turns vague confidence into visible gaps.

Think of a useful early draft as a diagnostic tool in your research, not a final polished argument.

I often create placeholders such as:

• needs source;
• verify current status;
• check whether this is still active;
• map to recommendation;
• possible appendix;
• too much background;
• unsupported leap.

Those notes are signs that the research process is working. Yay for critical thinking!

The danger is becoming attached to a draft structure too early. Sometimes the evidence refuses to support the original framing. Annoying, but useful. A defensible paper should be shaped by the evidence, not by the first outline that looked elegant.

Separate Background From Decision Material

One of the harder parts of research writing is deciding what not to include. By the time you have read deeply, everything feels relevant. It is not.

A decision-oriented paper should not become a museum of everything the researcher learned. Readers need enough context to trust the analysis, but they do not need every definitional rabbit hole, every adjacent debate, or every fascinating source that failed to affect the recommendation. There is a difference between doing a general review of the tech landscape and creating an actionable report.

This is where appendices earn their keep.

The main body should carry the argument. The appendix can hold methodology, source lists, detailed candidate screens, outreach logs, longer comparison tables, and background material that supports transparency without overwhelming the reader.

Appendices allow you to point to how you arrive at your conclusions without drowning the paper in justifications. Use them.

The Real Skill Is Judgment

Research is often described as finding information. That is only part of it.

The harder work is judgment:

• knowing which sources deserve weight;
• recognizing when terminology is inconsistent;
• distinguishing activity from aspiration;
• identifying when a source is outdated but still influential;
• noticing when a recommendation is attractive but unsupported;
• understanding which gaps are real and which are artifacts of poor documentation;
• deciding what belongs in the paper and what belongs in the appendix.

A good research process does not eliminate uncertainty. It disciplines it.

That is what makes the final paper defensible. Not that it answers every possible question. Not that it cites everything ever written. Not that it produces a grand unified theory of the topic, which is usually a sign that everyone should go outside for a while.

A defensible paper shows its work. It connects claims to evidence. It is honest about limits. It uses outreach where public materials are insufficient. It verifies details before relying on them. It keeps track of sources carefully enough that the argument can be reviewed, challenged, and improved.

That is the difference between research as content production and research as decision support.

The first creates a document.

The second creates something people can responsibly act on.

Have a project you’d like me to work on? Let’s chat about it!

📩 If you’d like to be notified of new posts rather than hoping you catch it on social media, I have an option for you! Subscribe to get a notification when new posts go live. No spam, just announcements of new posts. [Subscribe here]

Transcript

It’s probably been a while—if ever—since many of us had to write a formal research paper.

However, the underlying skill remains incredibly valuable, especially when making strategic recommendations.

Every recommendation ultimately answers a deceptively simple question:

Where should someone invest their attention, time, money, or credibility?

Although that sounds like a strategy question, it’s really a research question.

Before making any defensible recommendation, you need to understand:

• What already exists
• What is credible
• What remains speculative
• Where meaningful gaps exist
• Which claims are supported by evidence rather than opinion

This is the kind of research that transforms a messy problem into a decision-ready paper.

---

Start With the Question, Not the Search Engine

One of the easiest mistakes is opening a browser and immediately typing broad search terms.

For a few minutes, it feels productive.

Then reality arrives.

Suddenly you have:

• Dozens of browser tabs
• Multiple reports from similarly named organizations
• Outdated white papers
• Blog posts linking to dead resources
• Growing uncertainty about where to begin

Instead, define the research question before beginning your search.

The question doesn’t need to be your final thesis.

It simply needs to define what the research must answer.

For example:

• What infrastructure is missing?
• Which organizations are active in this area?
• What has already been funded?
• Which standards or regulations shape deployment?
• Where are the implementation gaps?
• What evidence would make a recommendation credible?

A well-defined question guides everything that follows.

---

Defensible Research Requires Visible Evidence

A defensible paper is more than a well-written opinion.

Readers should understand:

• Why certain sources were chosen
• Why others were less relevant
• How recommendations connect directly to the evidence

Transparency strengthens credibility.

The goal isn’t simply to reach a conclusion.

It’s to demonstrate how that conclusion was reached.

---

Begin With Academic Grounding

One of the first places to begin is Google Scholar.

Academic literature helps establish:

• Terminology
• Foundational concepts
• Frequently cited work
• Competing definitions
• Adjacent disciplines

Scholar is particularly useful for identifying:

• Highly cited papers
• Recent literature reviews
• Foundational publications
• Areas of scholarly disagreement

However, academic literature has limitations.

Research often lags behind real-world deployment.

Many papers describe theory exceptionally well while offering little insight into:

• Procurement
• Governance
• Operational maintenance
• Funding
• Day-to-day implementation

Scholar provides valuable context.

It should not become the only source of evidence.

---

Look for the Gaps Between Disciplines

As research progresses, pay attention to how different communities describe similar ideas.

For example:

• Computer science
• Political science
• Public administration
• Standards organizations
• Civil society groups

They often use different language to describe closely related concepts.

Those differences matter.

Sometimes the most valuable insight comes from recognizing where communities are not talking to one another.

Bridging those gaps often becomes one of the paper’s most important contributions.

---

Expand Beyond Academic Sources

Once the academic foundation is established, broaden the research.

Depending on the topic, useful sources may include:

• International organizations
• Government agencies
• Standards bodies
• NGOs
• Development organizations
• Think tanks
• Philanthropic foundations
• Public innovation teams

Rather than relying solely on search engines, search directly within organizational websites and document libraries.

Look for:

• Policy papers
• Technical specifications
• Regulatory guidance
• Grant announcements
• Meeting minutes
• Public consultations
• Implementation guidance

This work isn’t glamorous.

However, it reveals what institutions are actually doing—not simply what people are saying.

---

Separate Activity From Evidence

General web searches still have value.

Useful leads often come from:

• Conference agendas
• GitHub repositories
• Project websites
• Podcasts
• Technical blogs

These resources can identify:

• Active projects
• Emerging communities
• Current implementation work

However, treat them as leads rather than proof.

Whenever something appears important, verify it elsewhere.

That distinction is essential.

---

Build a Source Spreadsheet Early

One of the most valuable research tools isn’t complicated.

It’s a spreadsheet.

A source spreadsheet should begin early in the process.

Useful fields include:

• Source title
• Organization
• Author
• Publication date
• URL
• Citation
• Source type
• Topic category

For larger projects, additional columns may include:

• Geography
• Funding relevance
• Regulatory implications
• Technical maturity
• Recommendation mapping

This spreadsheet becomes much more than an administrative task.

It becomes part of the research itself.

---

Track Why Sources Matter

A good spreadsheet does more than collect references.

It helps answer questions such as:

• Why was this source included?
• Why was another source removed?
• Which recommendation does it support?
• Has newer evidence replaced it?

Without that discipline, it’s remarkably easy to rediscover the same material repeatedly while believing it’s new.

The spreadsheet becomes an extension of your own memory.

---

Verify Organizations and Programs

Eventually, research moves beyond published reports.

Verification becomes critical.

Depending on the project, this may involve:

• Nonprofit records
• Foundation databases
• Grant databases
• Company registries
• Tax filings
• Standards participation records
• Public consultation archives

Verification answers practical questions such as:

• Is the organization still active?
• Who funds it?
• Has it received grants before?
• Is there evidence of deployment?
• Is there someone to contact?

Verification frequently challenges assumptions.

And that’s exactly why it matters.

---

Reach Out to People Doing the Work

Public information eventually reaches its limits.

Sometimes the only way to answer important questions is to ask the people directly.

Outreach doesn’t need to become a formal research interview.

Often the goal is simply to:

• Verify details
• Clarify uncertainties
• Confirm current practice

Responses may be incomplete.

Sometimes there may be no response at all.

Both outcomes provide useful context.

However, outreach should always be documented carefully.

Track:

• Who was contacted
• When
• How
• What was asked
• What was learned

Silence is not evidence.

It is simply another research constraint.

---

Use AI as an Assistant, Not an Authority

AI can make research more efficient.

For example, it can help:

• Organize notes
• Identify recurring themes
• Suggest additional search terms
• Compare draft recommendations
• Identify gaps in coverage
• Improve writing flow

However, AI cannot replace source evaluation.

It cannot determine whether:

• An organization remains active
• A policy has been superseded
• A specification has been implemented
• A claim is genuinely supported

Those remain human judgments.

AI should organize uncertainty—not hide it.

---

Draft Before Research Is Finished

Many researchers wait until every source has been collected before writing.

I rarely do.

Early drafting exposes weak points surprisingly quickly.

When something cannot be explained clearly, the problem is often not the writing.

It’s the research.

Early drafts become diagnostic tools.

My notes often include reminders such as:

• Verify source
• Confirm current status
• Add supporting evidence
• Move background to appendix
• Recommendation needs support

These aren’t failures.

They’re signs that the research process is working.

---

Let the Evidence Shape the Paper

One of the biggest challenges is avoiding attachment to the original outline.

Sometimes the evidence points somewhere unexpected.

That can be frustrating.

It’s also valuable.

A defensible paper should follow the evidence—not force the evidence to fit an existing narrative.

---

Decide What Doesn’t Belong

By the end of a project, almost everything feels important.

It isn’t.

A decision-oriented paper should not become a museum of everything the researcher learned.

Instead:

• Keep the main argument focused.
• Move supporting material to appendices.
• Preserve transparency without overwhelming the reader.

Appendices work well for:

• Methodology
• Source inventories
• Comparison tables
• Outreach logs
• Background research

This allows readers to verify the work without losing sight of the main argument.

---

Research Is Really About Judgment

Finding information is only part of research.

The harder work involves deciding:

• Which evidence deserves attention
• Which terminology is consistent
• Which organizations remain active
• Which claims are credible
• Which details belong in the final paper

Good research doesn’t eliminate uncertainty.

It disciplines uncertainty.

---

Final Thoughts

A defensible paper doesn’t answer every possible question.

Nor does it cite every source ever published.

Instead, it connects evidence directly to recommendations.

It verifies important details.

It acknowledges limitations honestly.

And it provides readers with enough transparency to understand how conclusions were reached.

That’s the difference between research that simply produces content and research that supports real decision-making.

---

Conclusion

Strong research is not about collecting the largest number of sources.

It’s about asking better questions, evaluating evidence carefully, and building recommendations that can withstand scrutiny.

By combining academic research, institutional evidence, verification, outreach, thoughtful use of AI, and disciplined writing, you create something far more valuable than a report.

You create a paper that people can trust—and act upon.