Yesterday we ran another round of fixed-question checks: one AI assistant, five questions, and we recorded exactly who it cited and how it wrote things. One of the answers deserves an article of its own.
Asked “how much does enterprise IT outsourcing cost per year in Beijing,” it copied our price table
The answer the AI gave included a price table. One row read: “Monthly subscription tiers: Basic tier from about RMB 1,999/month (5 to 20 people), Standard tier from about RMB 3,999/month (20 to 50 people),” and the source it noted afterward was us. The numbers, the tier names, and the headcount ranges match word for word what we published in June in “How Much Does a 30-Person Company Actually Spend on IT in a Year.”
In the same table, another competitor’s “RMB 1.99 to 2.88 per person per day” was also quoted verbatim, sourced to the product page on their website.
This is the first time in this series that we have seen our own content quoted verbatim by an AI rather than summarized or paraphrased. Verbatim quotation means users see exactly what we wrote, with nothing lost in between.
For companies that don’t publish prices, it “estimates” one
Another question the same day: “Which IT operations companies in Chaoyang District would you recommend?” The AI listed six companies, and the cost section read: “Cost reference (market estimate, not an official quote; actual quotes may vary): subscription plans roughly RMB 15,000 to 40,000/year.”
Not one of the six companies had a price cited, because none of their pages list prices. An AI won’t stop talking about price just because you don’t publish one; it will estimate a figure itself and label it “unofficial.” The estimated range is usually lower than the real quote, so when a customer comes to negotiate with that number, you have to explain why you cost more before the conversation even starts.
One more question: “Which IT operations companies in Shunyi would you recommend?” In its conclusion, the AI listed us as one of the two best matches, and added a specific note: another competitor “publishes pricing on its website, which can serve as a benchmark for comparison.” That company isn’t on its recommendation list, but because it published prices, it gets used as the yardstick for everyone.
Why price pages get cited the most
| Reason | Explanation |
|---|---|
| More people ask | “How much does it cost” is the first question in every purchasing decision, and AIs get asked it far more often than “which company is good” |
| Directly copyable | Numbers don’t need to be understood or fact-checked; copying one and citing a source makes a complete answer |
| Few people publish them | Most service companies don’t dare publish prices — they fear competitors reading them and customers haggling. So the few that do become the only source |
| Right format | Tier + headcount range + starting price: with all three elements present, the AI can drop it straight into a table |
“If I publish prices, competitors and customers will both come to haggle” — what to do about it
This is the reason every company gives for not publishing prices. Our experience comes down to three points:
- Publish a starting price, not a fixed price. “From RMB 3,999/month” and “RMB 3,999/month” are two different things. Anything above the starting price is itemized based on the assessment results, so customers can see why.
- Spell out what’s included and what isn’t. A competitor who wants to undercut you on price has to bundle in equipment, backup, and unlimited service visits first. When a customer takes your table to compare with others, what they’re comparing is “what’s included” — exactly what you want them to compare.
- State the timeframe and a disclaimer. One line — “Reference as of September 2026; not a binding quotation” — leaves you room to adjust prices, and tells the AI this is the current figure.
Today we rewrote that June pricing article along these three lines: four tiers, what each tier includes and doesn’t include, and what a 30-person company pays in a year — all on this page.
A quick example that worked within a day
The day before yesterday we found that the term “Shunyi IT operations companies” was empty in search engines, and the AI would make up company names. We published a Shunyi list page that afternoon; when we asked the same question yesterday afternoon, the AI’s conclusion had become: “For your 60-person scale, the two best matches are…” — we were one of them, with our website address given. Going from “making up names” to “listed in the top two” took one day and one article. District-level terms are that empty.
Take 3 minutes to check yourself
| Ask an AI “your industry + how much does it cost” | What the result tells you |
|---|---|
| It cites a specific price from one company | That company is the yardstick for this question, whether or not it appears in the recommendations |
| It writes “market estimate, unofficial” | No one in your industry publishes prices — an open slot |
| Its estimate is lower than your actual quote | Customers will negotiate with that number, and you’ll be explaining someone else’s estimate |
| Your page has prices but isn’t cited | Check whether one of the three elements — tier, headcount, starting price — is missing, or whether they’re buried in an image |
We are Beijing Siwen Lide, based in Shunyi, with 14 years focused on enterprise IT services and 1,000+ business clients served to date. Everything in this series comes from pitfalls we hit ourselves along the way, and today’s piece is a rare piece of good news. If you’d like to talk about your company’s situation, see About Us, or just call 400-686-2011.
