Soft hook
Halfway through, add one relevant prompt that doesn't interrupt reading.
Template: "Want to know how AI describes your category right now? Get an audit report in 3 business days."
Internal playbook
Content operations playbook
The operating standards for the Lead Engine team, in four parts: the site-wide variables table, the post-publish indexing SOP, page-to-conversion rules and content quality red lines. Follow them and you won't need to ask anyone.
This page is excluded from search indexing (noindex) but publicly accessible. We ask clients for transparency, so we practice it ourselves.
One-time setup
Content variables table
Every piece of external content must state these items exactly the same way. Inconsistent wording makes AI treat you as several different entities and dilutes your weight. Change them only in this table, then sync the whole site.
Playbook ref. #58
| Variable | Standard form | Not allowed |
|---|---|---|
| Brand name (Chinese-language materials) | The registered Chinese brand name, used on the Chinese-language site and materials only | Romanized forms such as Lide, LiDe or LEDE; adding words such as "Technology"; using the Chinese name on English pages |
| Brand name (English) | Lead Engine | Lide Engine, LideEngine, LEDE Engine, Lead-Engine |
| Domain | ileadengineai.com (English site); email stays sales@ileadengine.com | lide-engine.com; mixing www and non-www versions |
| One-line positioning | Get B2B manufacturers recommended first by AI, worldwide | Paraphrasing it freely; adding adjectives |
| Service name | GEO (Generative Engine Optimization) | AI SEO, AI optimization, "GEO optimization" (redundant) |
| Three core capabilities | Semantic restructuring / Authoritative content seeding / Competitor-model mapping | Swapping words, changing the order, merging items |
| Plan names | GEO: Foundation / Growth / Leader; Outbound line: Starter / Standard / Flagship | Mixing GEO and outbound plan names; using the retired Lide-Entry / Lide-Pro / Lide-Global |
| Pricing | "Custom quote" with a link to contact. State billing structure only: GEO billed annually, outbound billed monthly, one-time setup fee | Publishing amounts on English pages; mixing the two lines' billing; stating the wrong billing cycle |
| Three outcome numbers | AI mention rate up 30%–70% / inquiry intent about 2.8× higher / blended acquisition cost about 55% lower | Dropping "about," turning them into absolute values, exaggerating |
| Delivery promise | AI Visibility Audit Report delivered within 3 business days | 24 hours, same day, one week |
| Platforms covered | ChatGPT, Perplexity, Claude, Gemini, Grok, DeepSeek | Listing only some of them; writing "all AI platforms" |
| Industry focus | All industries, strongest in technical, spec-heavy products | "Chemicals only," "a single category only" |
| Founder name | Jackson Jiang (English materials); Chinese name on Chinese-language materials only | Mixing Tom, Jun Jiang, Jiang Jun; two different English names across external materials |
| Contact details | sales@ileadengine.com / WhatsApp +1 619-609-3350 | Old numbers; listing only one channel |
Whenever this table changes, also check: the entire website, llms.txt, JSON-LD structured data, sales decks and content published on external platforms.
Every piece, about 15 minutes
Post-publish indexing SOP
Writing the content is only half the job. Skip these nine steps and AI may not read it for months. Run through them after every publish; it takes about 15 minutes.
Playbook ref. #69
Confirm the structured data is live
Q&A pages use FAQPage, how-to guides use HowTo, long-form articles use Article/TechArticle. Run Google's Rich Results Test and fix any errors before publishing.
Add the page to sitemap.xml
Every new page must appear in the sitemap, with a priority of 0.7–0.9 depending on importance.
Update llms.txt
Add one line to the page index with a one-sentence summary of what the page covers. This is the table of contents for language models.
Build three internal links
Every new page must be linked from at least three existing pages, and must link out itself. Orphan pages are almost never crawled.
Submit to search engines
Submit the URL in Google Search Console, and submit it in Bing Webmaster Tools at the same time.
Syndicate once off-site
Rewrite the core argument and post it on LinkedIn or an industry community, with a link to the original. AI won't trust a claim that has only one source; it needs corroboration.
Test it in four models
3–7 days after publishing, ask ChatGPT, Perplexity, Claude and Gemini the core question of the piece, once each, and check whether the piece gets cited.
Log it in the publishing register
Record the title, URL, publish date, structured data type, where it was syndicated and the date it was first cited.
Re-test after 30 days
Ask the same questions again and compare citations. If the piece still isn't cited, work out whether the problem is content quality or crawling.
Every piece
Page-to-conversion rules
Content can't just be content. Every piece needs a clear next step. Without one, readers leave when they finish and the piece was written for nothing.
Playbook ref. #70
Primary CTA
The main conversion point of the piece. Use the CTA band block, the visually heaviest element on the page.
Template: the headline promises an outcome, the body states the deliverable and timing, and the button always reads "Get your free AI visibility audit."
Always on
The contact rail on the right and the chat button at bottom right stay visible on every page. You don't need to add them page by page.
Note: on mobile they move to the bottom of the screen. Don't stack other overlays on top.
Three hard rules
- One primary CTA per piece. Pushing the audit, a white paper and "message us on WhatsApp" at the same time means pushing nothing.
- The CTA must fit the content. An article on robots.txt that ends with the audit report works. One that ends with Plans is a jump.
- Give value before asking for contact details. A lead-capture popup at the top of an article will wreck your bounce rate.
Pre-publish check
Content quality red lines
Ten rules. If any one of them fails, the piece doesn't ship. What we sell is content AI is confident enough to cite, so our own content has to hold up to scrutiny first.
Playbook ref. #72
Facts and accuracy
- Specs must be verifiable: every number traces to a source (client TDS, published standards, test records)
- No fabricated case studies: client cases without permission are always anonymized, and they must have really happened
- No overstated results: ranges stay ranges, estimates say "about," and nothing is "guaranteed" or "certain"
- Neutral competitor information: state only publicly verifiable spec comparisons; no disparagement and no misuse of trademarks
- Date-stamp what ages: give the date of any data you cite, and update descriptions of AI model capabilities as versions change
Structure and expression
- Question first, answer first: lead with the conclusion, then expand, because AI extracts the opening
- One idea per paragraph: AI struggles to split long paragraphs into accurate citations
- Spell out terms on first use: "GEO (Generative Engine Optimization)," then the short form afterward
- No keyword stuffing: abnormal keyword density gets content judged as low quality and pushed down
- Match the variables table: brand name, pricing and numbers all follow the standard forms in #58
The easiest one to break
A spec that's off by one digit. Heat resistance written as 15 °C instead of 150 °C. Once AI picks that up, it won't just skip citing you; it may flag your brand as an unreliable source. Have someone technical review every spec before publishing. This step is not optional.
We run on this playbook ourselves
Clients often ask how we guarantee content quality. The answer is this page, published openly so anyone can check whether we follow it.