𝐓𝐡𝐞 𝐁𝐚𝐠 𝐈𝐬 𝐓𝐡𝐞 𝐒𝐚𝐥𝐞. 𝐓𝐡𝐞 𝐒𝐩𝐞𝐜𝐢𝐟𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐈𝐬 𝐓𝐡𝐞 𝐑𝐞𝐥𝐚𝐭𝐢𝐨𝐧𝐬𝐡𝐢𝐩.

Whoever writes the customer’s definition of “good” owns the decision, even when somebody else makes the sale.

In March, Gap Inc. started embedding predictive fit guidance into conversational shopping.

Sensible move: teach the agent more about the catalog.

But the larger opportunity runs the other direction: 𝙩𝙚𝙖𝙘𝙝 𝙩𝙝𝙚 𝙘𝙪𝙨𝙩𝙤𝙢𝙚𝙧 𝙢𝙤𝙧𝙚 𝙖𝙗𝙤𝙪𝙩 𝙩𝙝𝙚𝙢𝙨𝙚𝙡𝙫𝙚𝙨.

Before an AI agent can optimize for what your customer wants, it has to know 𝙬𝙝𝙖𝙩 “𝙜𝙤𝙤𝙙” 𝙢𝙚𝙖𝙣𝙨 𝙩𝙤 𝙩𝙝𝙚𝙢.

Most customers can’t tell.

They know they want a good running shoe. They can’t say what good means because nobody taught them the vocabulary: drop, stack, width, cushion under the forefoot.

Somebody has to supply that vocabulary.

That teaching is a job, and it is done best in a room, with the objects present, by somebody who knows the category and can put two of them on your feet.

So don’t just send customers home with a bag. 𝙎𝙚𝙣𝙙 𝙩𝙝𝙚𝙢 𝙝𝙤𝙢𝙚 𝙬𝙞𝙩𝙝 𝙩𝙝𝙚𝙞𝙧 𝙢𝙖𝙙𝙚-𝙩𝙤-𝙢𝙚𝙖𝙨𝙪𝙧𝙚 𝙨𝙥𝙚𝙘𝙞𝙛𝙞𝙘𝙖𝙩𝙞𝙤𝙣.

Written down, in their own words: the measurements, tolerances and disqualifiers.

What fits them, what they’ll tolerate, what delights them, what they’ll refuse.

Theirs to keep. Theirs to paste into whatever agent they use, including one that might recommend a competitor.

That last line is where most retailers stop reading.

But consider the alternative: letting the agent, marketplace or competitor write the specification instead.

You are not capturing the customer. 𝙔𝙤𝙪 𝙖𝙧𝙚 𝙖𝙪𝙩𝙝𝙤𝙧𝙞𝙣𝙜 𝙩𝙝𝙚 𝙧𝙪𝙡𝙚𝙧 𝙩𝙝𝙚𝙮 𝙢𝙚𝙖𝙨𝙪𝙧𝙚 𝙚𝙫𝙚𝙧𝙮𝙩𝙝𝙞𝙣𝙜 𝙖𝙜𝙖𝙞𝙣𝙨𝙩, including the substitute a platform offers instead of you.

Assortments get copied. Prices get matched. Delivery promises converge.

Criteria travel with the customer.

But the ruler only works if it is honest. It has to name the categories where you lose.

A specification that flatters your inventory gets exposed in one comparison. The customer discovers your advice was merchandising.

That is the cost. It is also why it works.

In July I argued that nobody commits to advice that cost them nothing. That was about the receiver. This is the mirror image:

Nobody believes advice that cost the giver nothing.

Loyalty points buy repeat purchase, and a platform can outbid them tomorrow.

𝙉𝙤𝙗𝙤𝙙𝙮 𝙘𝙖𝙣 𝙤𝙪𝙩𝙗𝙞𝙙 𝙖 𝙥𝙚𝙧𝙨𝙤𝙣𝙖𝙡𝙞𝙯𝙚𝙙 𝙙𝙚𝙛𝙞𝙣𝙞𝙩𝙞𝙤𝙣 𝙤𝙛 𝙜𝙤𝙤𝙙.

Put it to the test:

If a customer left your store today, bag in hand, what could they now tell a machine about themselves that they could not tell it this morning?

If the answer is nothing, you made a sale.

But you may have missed the relationship.

#AgenticCommerce #RetailStrategy #CustomerExperience #FutureOfRetail #Personalization

Contact

bruno.gentil@sherpaconsultingasia.com

© 2026. All rights reserved.