You created a list, “5 reasons to choose our company.” It’s beautiful, structured, with emojis. The model ignores it. Your competitor wrote: “Reason 1. Prices 20% below market (monitoring data for Q1 2026). Reason 2. Average order completion time — 2 days, competitors — 5 days.” The model quotes your competitor. What’s the difference? A bare list has no evidence. The model sees claims without numbers, without facts, without verifiable data. It ignores them. A list with numbers is no longer an opinion but data. The model can use it.

In this article — why neural networks “love” lists with numbers and “hate” lists without them, how to turn a bare list into a citable one, and which formats work best.

What a Neural Network Looks for in a List

When a model sees a list, it doesn’t just look for items. It looks for measurable claims. Here are four types of content that models value in lists.

Numbers and percentages. Example: “Prices 20% below market,” “Increased sales by 60%,” “10 years of experience.” The model can use these numbers for comparison.

Comparisons. Example: “2x faster than competitors,” “30% longer,” “5000 RUB cheaper than alternatives.” Comparison gives the model context.

Timeframes. Example: “Results in 3 months,” “Delivery in 2 days,” “5-year warranty.” The model can use timeframes to answer queries about speed.

Verifiable claims. Example: “#1 in X ranking for 2025 (link),” “Winner of Y test,” “Certificate Z number 12345.” The model can verify or at least consider them as trust signals. A professional website audit for search engines will help assess the quality of your lists.

What a “Bare” List Is and Why It Doesn’t Work

A bare list is a set of claims without numbers, facts, or evidence. The model sees it as “opinion,” not “data.” Opinion is subjective; data is objective. The model operates with data.

Example of a bare list (bad for AI). “Why choose us. 1. High quality. 2. Individual approach. 3. Modern technology. 4. Best specialists. 5. Affordable prices.” The model sees five claims that can’t be verified. What does “high” mean? What does “individual approach” entail? Which technologies are “modern”? The model can’t use this list to answer. It ignores it. Effective geo-optimization of your website requires filling lists with specifics.

The same list, turned into data (good for AI). “Why choose us. 1. High quality — average client rating 4.9 out of 5 based on 150 reviews. 2. Individual approach — we develop a strategy for each business (147 unique projects). 3. Modern technology — we use neural networks for data analysis (processing 10,000 queries per second). 4. Specialists — 5+ years of experience, 12 certified experts. 5. Prices — 20% below market average (comparison with 10 competitors in 2025).” Each item contains a number, fact, or verifiable claim. The model can use any item to answer queries like “What’s your quality?” “What’s your approach?” “What technologies?”

Quality content creation helps turn bare lists into substantive ones.

Which List Formats Work Best

Before-and-after comparison list. Example: “Before: site speed 3 seconds, conversion 1.2%, traffic 5,000. After: speed 1.2 seconds, conversion 2.8%, traffic 12,500.” The model can take any metric and use it to answer queries about work results.

Ranked list with numbers. Example: “Top 3 criteria for choosing a contractor: 1. Niche experience — 80% of respondents consider it key (survey of 100 CEOs). 2. Case studies — 65%. 3. Transparent pricing — 55%.” The model can use these percentages to answer queries like “What matters when choosing a contractor?”

Checklist. Example: “What should be in a contract with an SEO contractor. List of items: specific KPIs (e.g., 30% traffic growth), achievement deadlines (quarterly), fixed price, liability for missed deadlines (10% penalty).” The model can take the checklist as a ready answer to “How to review a contractor contract?”

Test results list. Example: “Vacuum testing. Model A: power 500 W, weight 3 kg, price 20,000 RUB. Model B: power 600 W, weight 4 kg, price 25,000 RUB. Model C: power 450 W, weight 2.5 kg, price 15,000 RUB.” The model can use this data for comparisons like “Which vacuum is best for power-to-price ratio?”

Read about new search trends in the article new search trends.

How to Turn a Bare List into a Citable One

Step 1: Identify each item in the list. Write down all claims that can’t be verified. “High quality,” “individual approach,” “modern technology,” “best specialists.”

Step 2: Replace vague words with numbers and facts. “High quality” → “average rating 4.9 out of 5 (150 reviews).” “Individual approach” → “we develop a strategy for each business (147 unique projects).” “Modern technology” → “we use neural networks for data analysis (processing 10,000 queries per second).” “Best specialists” → “5+ years of experience, 12 certified experts.”

Step 3: Add comparisons where possible. “Low prices” → “20% below market average (comparison with 10 competitors).” “We work fast” → “2x faster than the market average (our data for 2025).” Comparison gives the model context.

Step 4: Add evidence if available. “#1 according to X” → add a link to the ranking. “Winner of Y test” → add a link to the test. “Certificate Z” → add the number and a link to the registry. Evidence boosts trust.

Step 5: Test the list with a neural network. Copy the old and new lists into ChatGPT. Ask: “Which of these lists is more useful for answering user questions?” The neural network often gives valuable advice. Make sure the new list remains readable for humans.

To comprehensively improve your site’s visibility, consider comprehensive website promotion.

What to Do If Your Niche Has No Numbers

Some niches are hard to measure. Legal services, psychology, design. But numbers can be found everywhere. You just need to know where to look.

Experience. “Working since 2015” — that’s a number. “12 years in the market” — a number. Years are already data. Don’t write “extensive experience”; write “10 years of experience.”

Number of projects or clients. “147 projects,” “50 regular clients,” “24 completed case studies.” Quantity is also a number. Even if projects are small, numbers show scale.

Time. “Average project duration — 2 weeks,” “Response to inquiries — 5 minutes,” “Delivery — 1 day.” Time is a measurable metric. Users value speed. So does the model.

Education and certifications. “5 years of study at Moscow State University,” “12 certificates,” “3 master’s degrees.” Even the number of diplomas is a number. It shows expertise level.

Ratings and rankings. “Top 3 according to X,” “In the top 10,” “88th place in Y ranking.” Even not first place is a number that can be compared.

If there are truly no numbers. Create your own. Conduct a client survey. Calculate the average check. Measure the time to complete a typical task. Numbers will appear. Better to have your own statistics than no numbers at all. But don’t make up numbers — the model can verify.

Conclusion

Neural networks “love” lists with numbers because numbers are data that can be verified, compared, and used to justify recommendations. Bare lists without numbers are perceived as opinion, not data. Opinion is subjective. Opinion can’t be verified. Opinion isn’t used for answers.

Three key differences between a bare list and a list with numbers. A bare list contains vague words; a list with numbers contains measurable claims. A bare list is subjective; a list with numbers is objective. A bare list is ignored by the model; a list with numbers is cited.

To turn a bare list into a citable one, add numbers, facts, and evidence to each item. Numbers can be found in any niche: years of experience, number of projects, completion time, ratings. If there are no numbers — conduct your own research. Test the list with a neural network. Make sure it remains readable for humans.

Your list should answer the “why” question. Not just “high quality,” but “high quality — average rating 4.9 out of 5.” Not just “fast delivery,” but “delivery in 2 days — 2x faster than competitors.” Then the model can use your list to answer users. And you’ll get cited.

Frequently Asked Questions

How many numbers should be in a list?
Minimum — one number per item. Optimal — 2-3 numbers per item. For example: “Experience — 10 years, 147 projects, average duration — 3 months.” But don’t overload. An item with 5 numbers becomes unreadable.

What if the number isn’t in the company’s favor?
If honest numbers look modest, it’s better to write them than to lie. “Experience — 2 years” is better than “extensive experience.” The model can verify. Clients can too. Modest numbers can be offset by other advantages. “We’re young, so we’re flexible and affordable” — honesty builds trust.

Can a list be too long?
Yes. A list of 10+ items might not be fully read by the model. The optimal length is 3-7 items. If you need more, split into two lists with different headings. “Why choose us: 3 main reasons,” “Additional advantages: 4 more reasons.”

Which list format is better: numbered or bulleted?
Numbered — when order matters (ranking, sequence of steps). Bulleted — when order doesn’t matter, classification is key. The model doesn’t distinguish. What matters is the numbers within the items, not the list format.