Differentiate informational, navigational, and transactional intent and classify keywords programmatically

WHAT THIS TESTS: This question tests whether you understand search intent as a business-critical signal rather than an abstract SEO concept. The interviewer wants to see that you can translate user psychology into an engineering spec: defining clear categories, choosing detectable linguistic markers, and building a system that is explainable and maintainable. At the senior level, they care about your ability to balance speed with accuracy and to validate query-level signals against SERP-level ground truth.

A GOOD ANSWER COVERS: A strong response starts by defining the three macro intents by the user's end goal, not the topic of the query. Informational means the user wants to learn or know something, navigational means they want to reach a specific site or page, and transactional means they want to complete an action like a purchase or a sign-up. Next, the candidate should explain a programmatic approach using rule-based classification on query modifiers. For example, regex or dictionary lookups can flag interrogatives like what, how, or why as informational; brand names, domain names, or login as navigational; and buy, cheap, near me, discount, or free trial as transactional. The best answers then describe a validation layer that checks SERP features and ranked domain types against the query label. If a query is labeled transactional but the SERP shows only blog posts and People Also Ask boxes, the classifier should downgrade its confidence or reclassify. Finally, a senior candidate mentions the need for niche-specific dictionaries and regular refresh cycles because modifier patterns evolve.

COMMON WRONG ANSWERS: A major red flag is proposing a black-box neural network or pure semantic similarity model without first establishing explainable modifier rules. Another common mistake is defining intent by keyword topic instead of user goal, such as claiming all software queries are informational. Candidates also err by ignoring navigational intent entirely or treating every local query as transactional rather than checking for store-finder versus buy-now signals. Finally, failing to mention validation against SERP features suggests a lack of production awareness.

LIKELY FOLLOW-UPS: The interviewer may ask how you would handle hybrid intents where a query blends learning and buying, such as best running shoes. They might also ask how to scale the classifier from a Google Sheet to BigQuery or Python, or how you would measure accuracy without a labeled gold-standard dataset. Another common follow-up is how you would detect micro intents, like brand-specific research phases, that generic templates miss.

ONE CONCRETE EXAMPLE: Suppose you are classifying keywords for an online electronics retailer. You would start with a regex rule set in Python or Google Sheets that scans for modifiers. The pattern buy or price or deal or discount or free shipping flags transactional intent, while what is or how to or why does or guide or tutorial flags informational intent. Navigational intent triggers on exact brand or domain matches like bestbuy or login. You then pull SERP data via an API and check whether Shopping ads, Local Pack, or product review domains appear. A query like best noise cancelling headphones 2024 might initially look informational due to the word best, but if the SERP is dominated by ecommerce product pages and Shopping results, your system should override or dual-label it as transactional with research micro intent.