Consumers Are Delegating the Work, Not the Final Say. AI and Purchase Decision in Indonesia

What four Indonesian studies show about AI, trust and the path to purchase

Most talk about AI starts with adoption. How many people use it? Which platforms are growing? How fast should brands move?

For marketers, there is a more useful question. Which part of the decision are consumers willing to give to AI?

We looked across four RB Consulting studies from 2026. They cover very different people: shoppers in Greater Jakarta, vehicle buyers, first-time Gen Z mothers, and vocational (SMK) students in 20 provinces.

The same pattern shows up in all four. Consumers use AI and other digital tools to find, learn, compare and simplify. But when the decision matters more, many still check with reviews, people, experts, stores or staff.

Consumers are handing AI the work faster than they are handing it the decision.

In Greater Jakarta:

•         62% used AI for shopping

•         3% said AI helped most at the final purchase decision

•         44% check reviews or a marketplace after an AI suggestion

One note on the numbers. The four studies use different samples, categories and questions. So please do not add them up into one national figure. Read together, they show the same shift in how people decide.

1. AI helps. It does not replace.

Two of the studies asked the same question in very different categories. Both point the same way.

	Greater Jakarta shoppers (SES A/B, age 25–45)	Vehicle buyers
Used AI	62% for shopping	About 40% for vehicle tasks
AI helped most with learning	52%	33%
AI helped most with finding or comparing	27% discovering something new	35% comparing brands
AI helped most at the final decision	3%	6%

So the shift is not simply from people deciding to AI deciding. Consumers give AI the thinking work: searching, sorting, comparing and shortlisting. The final decision is moving much more slowly.

2. Trust, checking and context

Low influence at the final decision does not mean consumers distrust AI. In Greater Jakarta, 64% rated AI recommendations as trusted. That is statistically the same as other customers' reviews (68%). Yet only 3% said AI helped most at the final decision.

The gap grows when the stakes rise. In automotive, only 4% completely trust an AI chatbot when choosing a vehicle. For family or friends it is 15%. For dealers or showroom staff it is 14%.

Among first-time Gen Z mothers, 85% rely on expert advice before deciding for their child. And 71% trust experts more than family or social media.

So the better question is not "Do consumers trust AI?" It is "What will they trust AI to do?"

There is also a clear checking step. This is what Greater Jakarta shoppers do right after an AI suggestion:

What they do next	Share
Search reviews or check a marketplace	44%
Visit a store	22%
Buy directly on the AI suggestion	18%
Ask someone they know	14%

AI can open the file. Another source often closes it.

We see the same pattern in other categories. Vehicle buyers still turn to dealers, family and reviews. Gen Z mothers look for other mothers' experience and a paediatrician's approval. SMK students find things on social media, but rely on friends, family and their own research to decide what to believe.

Context matters too. In our Greater Jakarta sample, AI use and trust were about the same for men and women, SES A and B, and age 25–34 and 35–45.

Outside this group, the picture changes. In automotive, younger consumers use AI more. In our SMK study across 20 provinces, AI is still a small part of how students check before buying. Social media, friends and family matter much more.

So AI can be mainstream in one group and minor in another. The same person may let AI pick a restaurant, use AI as one input for skincare, and still insist on a doctor for a health decision.

3. For brands: Be Found → Be Verified → Be Chosen

Finding, checking and deciding are now separate jobs. Brands need to manage all three.

•         Be Found: make sure AI and online search can find your brand and explain your products, facts and claims correctly.

•         Be Verified: make the next step believable, with reviews, marketplace pages, clear pricing, proof from other customers, evidence, and expert support where it matters.

•         Be Chosen: protect the last step, with stock availability, stores or showrooms, frontline staff, trial, service and delivery.

A brand can win the AI answer and still lose the sale. This happens when its reviews are weak, its marketplace information does not match, the product is out of stock, or the store experience breaks the promise.

The reverse is also true. A trusted brand can drop out of consideration if AI cannot find it or explain it correctly.

For marketers, this means treating AI as a role in the journey, not just another channel. Measure usage, trust and decision power separately. And keep the non-AI routes healthy: even in our connected Greater Jakarta sample, 21% used no AI at all.

The researcher's view: measure how far consumers let AI go

For marketing research, the next question is no longer just "Do consumers use AI?" It is "How far will they let AI go in a specific decision?"

We call this AI delegation depth. It has six levels:

1.       Ask for information

2.       Ask for a comparison

3.       Accept a shortlist

4.       Treat AI's answer as a recommendation

5.       Follow the recommendation

6.       Let AI buy or act for them

We should measure trust, checking and final outcomes at each level. This tells brands what AI really is in their category: just another information source, a shortlist maker, a trusted adviser, or an agent that acts for the consumer.

The journey is changing faster than the consumer

These studies do not show AI replacing anyone. Consumers are not leaving people, experts, stores, reviews or other digital channels behind. They are sharing the jobs out in a new way.

AI is very good at making choice easier. It can search, sort, compare and explain very fast. But who makes the final call still depends on four things:

•         how much is at stake

•         whether a trusted expert is available

•         whether people need to see or touch the product

•         their own habit of checking

The winning brand will not just be visible in AI. It will make the whole journey work, from the first recommendation to the final reassurance.

Study note

This article brings together four RB Consulting studies from 2026. It looks for patterns across them. It does not combine them into one number. Samples, categories, locations and questions differ, so read each number within its own study.

• Greater Jakarta AI & Trust: n=180, age 25–45, SES A1/A2/B, shopping in general.

• Automotive AI: 1,400+ consumers engaged with cars and motorcycles, surveyed at vehicle exhibitions, unweighted.

• Gen Z Mothers: first-time mothers in Jakarta and Semarang, parenting and baby products.

• IYG/SMK 2026: n=5,027, age 15–18, 20 provinces, urban and semi-urban, mostly mid-to-lower income.