Build a Search Strategy That Tests Your Assumptions

In the last chapter, you turned your question into something sharp, bounded, and tied to a clear finish line. Now you finally get to go looking, but how you look decides what you find. Type your question into a search bar, grab the first confident-looking link, and you are right back to being confidently wrong. This chapter gives you the alternative: a deliberate plan for where to search and what to type, including a move most people never make on purpose, searching for evidence that you are wrong.

Choose Where to Search, and Why

Not all of the internet is the same place. Where you search shapes what you find, so the first decision in any search plan is matching the venue to what you actually need. A search venue is simply a type of place you can look, and each one is good at something different.

• General web search (Google, Bing, and the like): fast and broad, great for getting oriented and finding leads, but it mixes trustworthy sources and junk together on the same page.

• Scholarly and research sources: studies and research reviews, useful when you need evidence rather than opinion. You do not need an expensive database: Google Scholar, open-access journals, and your public library’s free online access already get you a long way.

• Official and institutional sources: government statistics portals, public-health bodies, regulators. Best when you need official numbers, rules, or definitions.

• Independent testing and consumer sources: consumer organizations and independent reviewers who test things themselves, useful for product and “does it actually work” questions, as long as they are not selling the thing they are reviewing.

Write Queries, Including Ones Meant to Prove You Wrong

Once you know where to look, you need to decide what to type. Typing your whole question in word for word rarely returns the best evidence. Good queries are built, not guessed.

Start from the key terms in your scoped question, the who, the what, and the outcome you defined last chapter. Then sharpen them: try synonyms a source might actually use, and add words that target your specific outcome, like “cost”, “injury rate”, or “independent test”. A few focused queries beat one long sentence.

Now the move that separates real research from collecting cheerleaders. Most of us, without realizing it, search in a way that confirms what we are already hoping is true. That tendency has a name: confirmation bias, the habit of noticing and trusting evidence that agrees with us while skipping past evidence that does not. Left unchecked, it quietly turns a search into a hunt for applause.

The fix is built right into your query list. A disconfirming search is one you run on purpose to find evidence against your hoped-for answer. If you are hoping standing desks are good for your back, you also search “standing desk back pain”, “standing desk risks”, and “standing desk criticism”.

Because it makes your answer honest either way. If the criticism you find turns out to be weak, your original answer is now stronger, you stress-tested it and it held. And if the criticism turns out to be strong, you just saved yourself from a confident mistake before it cost you anything. A disconfirming search never wastes your time; it only ever makes your conclusion more trustworthy.

Watch a Search Plan Come Together

Let’s build one end to end, on a civic question many residents actually ask. Suppose your scoped question is:

“In our city, did adding protected bike lanes on the main roads reduce cyclist injuries over the last five years?”

First, pick your venues to match what the question needs:

• Official: the city or national road-safety statistics portal, for the actual injury numbers.

• Scholarly: studies on protected bike lanes and injury rates, found through Google Scholar.

• General web and independent: news coverage and transport groups, both supporters and critics, to find leads and counterarguments.

Then write a query set that deliberately points in both directions:

• protected bike lanes cyclist injury rate (your city)

• city road safety statistics cyclists before and after bike lanes

• protected bike lane safety study

• protected bike lanes accidents increase

• “bike lanes” problems OR criticism

The first three go looking for the effect; the last two are disconfirming, actively hunting for evidence that the lanes made no difference or made things worse. Notice how this plan feeds straight into the success criteria you set last chapter: more than one independent source, matched to your scope, recent, and including at least one source that could have disagreed. The disconfirming queries are how you make sure that last condition is met.

Your Turn!

Time to build a small search plan of your own.

The situation. A friend with mild lower-back discomfort is thinking about buying a standing desk because they heard it is “better for your back”, and they ask how you would actually check that. You have already helped them scope the question:

“For an office worker with mild lower-back pain, does using a standing desk for part of the day reduce back pain compared with sitting all day, according to independent or research sources from the last few years?”

Your task. Plan the search.

• Pick at least two search venues and note why each fits.

• Write five search queries: three aimed at finding evidence, and two disconfirming queries designed to surface evidence that standing desks do not help, or cause other problems.

Deliver: two venues with a few words of justification, five queries (clearly mark the two disconfirming ones), and one line on what a good-enough answer would need. Keep this, it is the start of the Search Strategy log you will build on in your project.

Then check your answer against the model solution found at the end of this chapter.

Let’s Recap!

• Where you search matters as much as what you type: match the venue, general web, scholarly, official, or independent testing, to what you actually need.

• A high search ranking shows popularity or good optimization, not truth.

• Build queries from your scoped question’s key terms and outcome, rather than typing the whole question word for word.

• Confirmation bias makes us search for agreement, so a disconfirming search deliberately looks for evidence against your hoped-for answer.

• A good search plan pairs confirming and disconfirming queries, so the answer you reach has already survived a real challenge.

You can now frame a question and plan a search that actively tries to prove you wrong. In the next part, you’ll deal with what comes back: how to tell what kind of source you are looking at, trace a claim to where it really started, and decide which sources you can actually trust. Before you head there, take the short quiz that follows to check your understanding of Part 1.

Model Solution

There is no single correct plan, but a strong one looks in the right places and points in both directions. Here is one good version.

Venues:

• Scholarly (Google Scholar), for studies measuring standing-desk use against back pain.

• An independent or official health and ergonomics source, for practical, non-commercial guidance.

Queries:

• standing desk lower back pain study

• sit-stand desk back pain office workers evidence

• standing desk ergonomics health guidance

• standing desk back pain worse (disconfirming)

• “standing desk” no benefit OR risks (disconfirming)

Why this works:

• The venues match the need: research evidence plus practical, non-selling guidance, not a furniture shop.

• The first three queries chase the effect, using the scope’s key terms (office worker, back pain, recent evidence).

• The two disconfirming queries actively look for “it does not help” or “it causes other problems”, so a “no” could actually surface.

Now compare a weak attempt:

If your version named real venues and included two queries genuinely aimed at proving the idea wrong, you are searching like a researcher now. The topic matters far less than the habit of searching against yourself, not just for yourself.