Imagine someone sitting at the kitchen table in the evening with an AI assistant, describing that the dining room is only 2.80 meters wide, but should have room for four people plus guests. This sentence contains exactly what you need for advertising in ChatGPT: a real problem, a life situation and no solution yet. These are exactly the kinds of conversations you are looking for as Context Hints.
The video is only loaded from YouTube when you play it – data (including your IP address) is then transmitted to YouTube/Google. More in the privacy policy
Classic keyword tools are hardly helpful here, because they deliver search terms, not sentences. Instead, you need descriptions in which people explain their situation in full sentences. This guide shows you where to find such phrasings, how to translate product questions back into their preliminary problem stage, and how to use ChatGPT itself to reconstruct suitable conversation starters.
Key Takeaways
Key Takeaways
- Context Hints are complete sentences that describe a problem or a life situation, not individual search terms.
- Your own customer data, meaning support emails, return reasons and sales conversations, is the most valuable source.
- Question portals such as Gutefrage.net and Reddit show how people formulate their context in detail.
- The Keyword Planner is worthless for this task because it delivers terms without a situation.
- Stage 3 of the reverse analysis, meaning the life situation without any mention of a product, is the most valuable for ChatGPT Ads.
- A cross-check shows whether your product category appears in the conversations and whether competitors are already mentioned.
Step-by-Step Guide
Step 1: Analyze your own customer data
Start with yourself, because this is where the most authentic phrasings are. Open your support emails, return reasons, notes from sales conversations and inquiries via contact forms. Don't focus on the question about the product, but on the sentence before it: What did the customer experience before writing to you? A return because of "doesn't fit through the door" tells you, for example, that people are dealing with space problems in old-building apartments.
Step 2: Use briefings and frequent customer questions
Many companies already have documents in which questions and objections are collected. One example is a section in a comprehensive customer briefing that lists typical first questions. Read through such sections carefully and note which questions begin with a situation. Questions like "Does a 200 cm table fit in a room that is 320 cm wide?" are ideal templates because they are already concrete and phrased in complete sentences.
If you also have an FAQ collection, go through it in the same way. Even general questions like "Is oak care easier than walnut care?" show you which topics occupy people before they have a purchase intent.
Step 3: Search question portals like Gutefrage.net and Reddit
Question portals are a goldmine because users write at length there, state their budget and describe their living situation. On Gutefrage.net, search for a term from your field, such as "wood", and sort the results by relevance. You will quickly find questions like "Which wood should I use for an armchair that will be on the terrace in summer?". This is almost identical to what people would ask their AI today.
Reddit works similarly, but often offers even more depth. Enter a product term there and read the threads in which someone explains their situation and others give answers. Pay particular attention to posts with follow-up questions, because they contain the context you need for Context Hints. A sentence like "Solid wood board is sagging downward, what can I do?" already contains problem, material and expectation.
Step 4: Analyze search suggestions in Google
Google also provides conversation starters if you look at the right places. Enter a product term with a purchase intent, such as "buy solid wood table Vienna", and look at the "People also searched for" section. There you will find follow-up terms like "clearance sale", "extendable" or "rustic". These are keywords, but they show you which direction users are thinking in and which features matter to them.
Use these terms as a hint, not as a finished conversation idea. Then turn them into complete situations, because only those fit the dialogue format of an AI.
Step 5: Using ChatGPT as a Researcher
Now the AI itself comes into play. Create a prompt that clearly states what you want: not search terms, but descriptions of conversations people have with an AI assistant before they know your product. Upload your case study or your customer material as a PDF or Word file and ask for about 15 conversation candidates.
Make sure you give the model the context of your company. Without customer data, you will get generic suggestions; with it, you get wording that fits your offering. Whether you use the free version or a subscription with an extended thinking mode hardly matters for getting started.
The results are directly usable conversation candidates, for example "Our dining room is pretty narrow and we don't know how to use the space sensibly. What would you recommend?". Collect these sentences in a table and supplement them with your notes from steps 1 to 4.
Step 6: Translating Product Questions into Their Problem Precursor
Many teams write ads for questions like "Which dining table fits in a narrow room?". But that only reaches people who already know they are looking for a table. So turn the question backwards: What was this person thinking before they started looking for a table?
Distinguish three levels here. At level 1, the user knows the product category and is looking for a selection; this is essentially classic Googling with more words. At level 2, they know the problem but not yet the solution, for example "The dining area is narrow, but we still need space." At level 3, they only describe their life situation without mentioning a product, for example "We have a small apartment and are realizing that our dining area has less space than we thought."
Level 3 is particularly valuable for ChatGPT Ads. Here the AI can work in a suitable solution without the user having actively searched for a product. However, don't expect many people to buy directly through an ad. The ad is often a first point of contact that later leads to a purchase decision.
Step 7: Running a Cross-Check with Product Category and Providers
Before you add a conversation idea to your map, test it. Put together a realistic, rather casually worded request, such as "Our dining room is only 2.80 meters wide and there are four of us. We still want a table where guests can sit too. What would work?" Let ChatGPT answer and then check two things.
First: Does your product category appear in the answer, and at what point? If the table is mentioned, you are well positioned on the map. Second: Are other providers named? If not, the answer leaves room for your recommendation. If competitors show up, make a note of it and consider whether to phrase the situation a bit more narrowly or with a quality feature in which you are strong.
Summary – Finding Context Hints: Conversation Map for ChatGPT Ads
Context Hints arise where people describe a problem in full sentences. You find them in your own customer data, in question portals, in search suggestions, and with the help of ChatGPT itself. Keyword tools only help you marginally here. What matters is translating the product question back into the life situation that precedes it. Level 3 brings you the greatest reach, because that is where the AI can bring in a solution. With the cross-check, you verify whether your product category appears in the answers. This creates a conversation map with which you can specifically address situations in which your offering is a sensible answer.
FAQ
Do I need a paid ChatGPT subscription for Context Hints?No, even the free version delivers usable conversation candidates. However, an extended thinking mode can improve the quality of the results.
Why is the Keyword Planner worthless for Context Hints?The Keyword Planner delivers search terms without context. Context Hints, on the other hand, need complete sentences that describe a situation and a problem.
Which source delivers the best Context Hints?Your own customer data, meaning support emails, return reasons, and sales conversations, is the most valuable source because the wording is authentic.
What is the difference between the levels of the backward analysis?Level 1 knows the product category and looks for a selection. Level 2 knows the problem but not the solution. Level 3 only describes the life situation and is the most valuable for ChatGPT Ads.
Do I have to name the product category in my ad?The conversation idea should ideally include the product category so that the AI is more likely to pick it up. For level 3 conversations, the description of the life situation is often enough.
How many conversation candidates should I collect at minimum?A good guideline is about 15 candidates as a starting point. What matters, however, is that they cover different situations and levels.
What do I do if other providers are named in the AI answer?Make a note of the situation and consider how to phrase it more precisely or link it to a feature where your offering clearly convinces.
