Research insights
Multimodal Search: How Voice and Image Search Change Product Discovery

On this page
Multimodal search means people look for information and products by typing, speaking or using an image, often in the same journey: they may photograph a product, ask a voice assistant a question, then type a brand name to compare prices. For brands, the shift changes how products are discovered and which cues (pack design, product names, spoken phrases) help people find them. Consumer research shows how your customers actually search and where your brand is missing.
What is multimodal search?
Search used to mean typing keywords into a search box. Today, phones and apps accept several kinds of input:
- Text search: typed keywords or questions in search engines, marketplaces and social apps.
- Voice search: spoken questions to a phone, smart speaker or in-app microphone. Spoken queries tend to be longer and phrased as natural questions.
- Image or visual search: pointing the camera at an object, a pack or a screenshot to identify it or find similar items.
- Combined search: an image plus a typed or spoken refinement, such as a photo of a dress with “in blue”.
“Multimodal” refers to using these inputs together, and to search systems that understand text, sound and images at the same time.
Why are consumers using voice and image search?
Each mode solves a different problem. Voice is convenient when hands are busy or when typing in Vietnamese with full diacritics can feel slow. Image search helps when people do not know a product’s name, see something they like on social media or in a shop, or want to check a product in front of them. Text remains useful for precise comparisons, such as a specific model or price.
How much each mode is used differs by age, category, device and situation. These differences are what research should measure, rather than assuming that one mode will replace the others.
How does multimodal search affect purchase decisions?
- Discovery starts earlier and in more places. A shopper may first see a product in a video or a friend’s photo, then search for it by image.
- Packaging and visual identity matter for search. If a pack is hard to recognise from a photo, image search may return a competitor or a similar product.
- Spoken language matters. Voice queries use everyday words and local terms, which may differ from the category names brands use.
- Questions replace keywords. People ask “which milk is good for children” rather than typing a short phrase, so clear answers to common questions become important.
What should you research about search behaviour?
| Business question | What to measure | Method |
|---|---|---|
| How do our customers search in this category? | Use of text, voice and image search by channel and situation | Survey, in-depth interviews |
| What words do they use? | Spoken and typed phrases, local terms, questions asked | Interviews, search query data |
| Can our product be found by image? | Recognition of the pack and logo from photos | Packaging test, task-based testing |
| Where does search lead to purchase? | Path from search to channel and purchase | Journey research, purchase data |
| Which groups use new search modes most? | Differences by age, city and device | Survey with segmentation |
How can you research search behaviour in Vietnam?
- Observe real searches. In in-depth interviews, ask shoppers to find a product the way they normally would, and record which mode, words and apps they use.
- Collect the language of the category. Focus groups and interviews reveal the everyday and regional terms people say aloud, which can guide product names, website content and answers to common questions.
- Size usage with a survey. A quantitative survey estimates how often each search mode is used, in which situations and by whom. Keep questions concrete and recent, such as “the last time you looked for this product”, to reduce guessing. See how to design a bias-free questionnaire.
- Test pack recognition. Show photos of your pack and competitors’ packs in realistic conditions and check whether shoppers and image tools identify them correctly.
- Link search to purchase. Combine journey research with purchase data to see which searches end in a purchase and through which channel.
What are common mistakes?
- Assuming one mode will replace the others. Most people switch modes depending on the task.
- Using brand language instead of customer language. If people describe the product differently from your naming, they may not find you.
- Ignoring the pack as a search tool. Image search relies on clear, distinctive visual cues.
- Relying only on stated habits. People often misremember how they searched. Observation and behaviour data give a more accurate picture.
FAQ
What is multimodal search?
It is search that uses more than one type of input, such as text, voice and images, often combined in one journey. Search systems increasingly understand these inputs together.
How can a brand prepare for voice search?
Research the everyday words and questions customers use when they speak about the category, then use that language in product names, website content and clear answers to common questions.
How do you test whether a product can be found by image search?
Photograph the pack in realistic conditions, run it through common image search tools and ask shoppers to identify it. If it is often confused with other products, review its distinctive visual cues.
Understand how your customers search
INTAGE Vietnam is a market research company in Vietnam and part of the INTAGE Group, Japan. We help businesses understand search and purchase journeys with qualitative research, surveys and packaging tests. See our market research services or contact us to plan a study.
Related reading
Latest posts
- Effective M&A (Mergers & Acquisitions) Opportunity Research
- Market Research: Building a Market Research Report for Investment Funds
Survey Sample Size: How Many Respondents Are Enough?
Survey Data Collection Methods in Vietnam: How to Choose- Technology Adoption in Enterprises: How to Research Employee Behavior


