Gains:
- Ability to prepare and verify opponent comparison (battlecard) tables with AI
- Ability to transform market, sector and trend research into structured output
- Ability to verify rival claims with sources against the risk of AI hallucination
In the middle of a sales call, the customer says: "Actually, we are also talking to company X, they do it cheaper." At that moment, you have two options: either you make a calm positioning because you really know the opponent, or you make a bad move and weaken the opportunity. Competitor and market research is one of the most neglected areas of sales but one that affects the win rate the most. Artificial intelligence can reduce this research from hours to minutes; prepares battlecards, structures market trends, proposes objection responses. However, there is a very serious trap here: artificial intelligence can produce outdated or completely fabricated information about competitors. In this unit, we will learn how to do quick research and how to avoid falling into this trap.
What is a Battlecard and why is it needed?
The Battlecard is a one-page sales tool that outlines how you will position yourself against a competitor. It includes the competitor's strengths and weaknesses, price position, ideal customer profile, and answers to your most frequently heard objections. A good battlecard prevents the agent from panicking during the call; because the answers have been thought out in advance. The AI is very good at producing the first draft of the battlecard, but remember: every claim it produces is a hypothesis, not fact. You should not disclose any competitor prices or features to the customer until they have been verified.
Battlecard table example
Size
Our Product
Competitor A
Competitor B
ideal customer
Institutions with 200-2000 people
SME-oriented
Enterprise
strength
Fast integration, native support
Low entry price
vast ecosystem
Weakness
Brand awareness is low
Limited reporting
long installation
price position
middle-upper
low
high
frequent objection
"We don't know you"
"Does not scale"
"Too expensive"
Our response
Reference + pilot offer
Show growth scenario
TCO comparison
Note: Competitor information in this table is an example. In your actual battlecard, each cell must be based on a resource.
Translating research into structured output: workflow
- Narrow the question. Don't say "Research my competitors" but rather "Research Competitor A's positioning towards mid-market customers."
- Ask the model for a hypothesis, not a fact. Have the output tagged as "claims that need to be verified."
- Add a source question to each claim. Let the model leave a note saying "confirm this with this source".
- Confirm with current and reliable sources. Competitor's official site for price, product documentation for feature, publicly available case studies for customer.
- Put it through an ethical filter. Translate it into a language that does not disparage the opponent and emphasizes your own value.
- Convert to Battlecard. Put the verified information into a one-page format.
Copiable prompt: battlecard draft
You are a B2B competitive analysis assistant. Prepare a battlecard DRAFT for the company "{{competitor_name}}" operating in the {{sector}} sector. Every information you produce is a HYPOTHESIS; state this clearly.For each item, fill in these three columns:| Claim | Confidence level (High/Medium/Low) | Type of source to be verified |Dimensions to be covered: ideal customer profile, strengths, weaknesses, price position (give exact figure, specify range/position), frequent objections.Rule: Do not produce exact price, exact number of customers or specific feature claim; mark them as "must be verified". Write "no data" where you don't know, don't make it up.
Copiable prompt: market/trend research
Configure the market dynamics for {{year}} in the {{sector}} sector. Output format:- Growth direction (increase/stable/decrease) + justification- 3 important trends- Changes in buyer behavior- Impact of these trends on our value proposition Put a label "[confirmation required]" next to each claim and write what type of source it will be confirmed by (sector report, official statistics, analyst publication). When giving numbers or rates, do not make up those without a source, say "source must be found".
Copiable prompt: response to objection
The customer expressed the following objection: "{{customer_objection}}"Competitor In the context of "{{competitor_name}}", produce 2 different answers to this objection, highlighting our own strengths, WITHOUT EVILING THE COMPETITOR. Each answer should be no more than 3 sentences and be based on a verifiable value proposition. Our values: {{value_proposals}}
Weak research / Strong research prompt
WEAK: "Give information about Competitor A and explain why it is worse than us." Problems: Model abandons objectivity, slides into derogatory language, may invent accurate but unverified prices/features, does not cite sources.
STRONG: "Draft a hypothesis-based battlecard for Competitor A. Add confidence level and type of source to be verified to each claim. Make up exact price/number of customers, write 'must be verified'. Base answers on our value proposition, without disparaging the competitor." Result: A draft that is auditable, ethical, and can be used safely in a sales call.
Mini case: the cost of the wrong price
On a team selling a cloud backup solution, the representative asked the AI, “What is Competitor B’s price?” he asks. “$12 per user per month,” the model confidently says. The agent tells this to a customer and claims "we're a better fit." However, Competitor B changed its pricing a year ago and the actual price is different. The customer knows this, the rep's credibility immediately collapses and the opportunity is lost. The same team rule is now: No competitor price given by artificial intelligence will be mentioned without confirmation from the competitor's official site or a current offer. This one rule has saved several deals. The lesson is clear: the model's confidence is not proof of its accuracy.
Tip: Add a small source tag next to each cell in the Battlecard (e.g. "source: competitor website, 07/2026"). This is how you know which cell to update when the information becomes stale. Competitor information gets stale quickly.
Caution: Artificial intelligence can hallucinate competing companies with outdated prices, non-existent features, and never-won customers. Before conveying a competitor's claim to the customer, be sure to confirm it from an up-to-date and primary source (competitor's official site, product documentation, current offer, publicly available case study). An unverified claim produces risk, not sales.
Common mistakes
- Mistaking the model's output for real. Artificial intelligence generates hypotheses; Using it without confirming with the source puts your credibility at risk.
- Matching exact price/features. If you make the model produce exact numbers, you will often get old or wrong numbers.
- To denigrate the opponent. Disparagement of the competitor makes the competitor grow in the eyes of the customer, not you; Focus on your own value.
- Not tracking resource obsolescence. Competitor knowledge quickly becomes stale; Battlecard without date and source tag becomes dangerous within months.
- Trusting a single source. Verify a critical claim with at least two independent sources.
In summary
- Battlecard is a one-page sales tool that allows you to position yourself calmly and prepared against your opponent.
- AI is fast at producing battlecards and market research outlines, but everything it produces is hypothesis.
- Confirm exact pricing, features and customer claims with up-to-date, primary sources.
- Have the model generate assertions with a "must be verified" tag and confidence level.
- Position yourself based on your own verifiable value proposition, without disparaging the competition.
- Keep battlecard updated with source and date tag.
Application task
Choose a real opponent and run the "battlecard draft" prompt above. The model will generate at least five assertions. For each claim, do this: go to a current, reliable source (competitor's official site, product page, publicly available case study) and either confirm or refute the claim. Make note of the claims you refute; You will see how often the model is wrong. Finally, from the verified information, create a one-page battlecard that does not disparage the opponent and add a source + date label to each cell.