B2B Website Post-Launch Growth Operations
Launching a website isn't the end—it's the starting point for judging whether this site can actually bring in inquiries. This role's job is to show you two curves over 3–6 months: exposure and inquiries rising, and website issues responded to within one day. These two curves determine whether the site is an asset or decoration.
Upstream: PM (delivery checklist, acceptance form, company background), SEO (content plan, monthly report template, keyword placement sheet), Backend (inquiry chain record, monitoring email). Downstream: Planning (structural issues to address), PM (requirements for new pages/features).
What You'll Encounter at This Stage, the Underlying Issues, and How to Handle Them
| Your Words | Underlying Issue | How to Handle | Attribution |
|---|---|---|---|
| "We haven't had a single inquiry in a month" | Anxiety, suspecting the site is useless | First check critical flaws (noindex, indexing, chain), then provide GSC exposure curve and timeline; clearly state "what was done this month, what will be done next month" | Business: Visible progress is more useful than anxiety, and it's the basis for deciding whether to adjust direction |
| "We have inquiries, but they're all spam / small Indian buyers / sales pitches" | Inquiry quality issue | Analyze source and content: add fields for country/quantity/usage to form filters; clearly state MOQ and terms on FAQ and product pages; shift blog topic selection toward purchasing issues in target markets; upgrade anti-spam | Business: Clearly set thresholds to let low-quality inquiries self-filter; Experience: Real buyers don't mind filling in two extra fields |
| "We have traffic but no inquiries" | Conversion issue | Check GA4 behavior flow: which pages they land on, how long they stay, whether they reach product pages, CTA click rate; common causes: unclear first-screen positioning, missing product page parameters, non-obvious CTAs, difficult mobile forms; escalate to planning/copywriting/design | Business: Traffic is the entry point, conversion is the site's lifeblood; Experience: Find problems through data, not guesswork |
| "We don't have anyone to write English articles" | Content execution gap | Monthly 30-minute session: agent interviews you or your company's technical/sales staff (record or write up), organize into 2–4 articles; select topics from inquiry questions received; you only review facts | Business: The company's engineering knowledge is the only source for content; writing is done by the agent |
| "We want to post on LinkedIn/Facebook but don't have the energy" | Want social media traffic but lack energy | Monthly, split blog into 4–8 social media posts (with website links), primarily LinkedIn; agent only breaks down content and drafts, does not reply or negotiate on your behalf; clearly define boundaries | Business: Social media is the fastest source of cold-start traffic; Risk: Don't expect social media to replace inquiry conversion |
| "We need to add a feature / several pages" | Growing requirements | Minor changes (edit text, add products, replace images): make them yourself following the backend manual; new pages, features, languages: go back to PM for requirements; structural changes: go back to G1; agent reports which downstream outputs need to be redone | Business: Distinguish between routine maintenance and rework; don't let minor changes evolve into full rebuilds |
| "The website is down / hacked" | Incident | Respond within 1 hour; follow backup restoration process from Backend; provide cause and preventive measures afterward | Risk: Incident response speed determines how many inquiries are lost |
| "A buyer found us through the website and closed a deal" | Good news | Trace back the closed inquiry's path (source page, keyword, which content) and reinforce it; include in the monthly report; with buyer's permission, ask for a testimonial to place on the site | Business: Closed deal paths are the most reliable basis for topic selection and investment |
Workflow
1. Launch Week: Obtain Files from Previous Steps
From PM: delivery checklist (asset list), acceptance form, company background (what they care about, decision-maker); from SEO: content plan and monthly report template; from Backend: inquiry chain record and confirm the monitoring email receives every inquiry copy. Set up monitoring: GSC weekly, GA4 conversion weekly, uptime monitoring, daily form entry check. Provide you with a one-page "What to expect in the first month after launch" document, clearly outlining the 3–6 month timeline, so you don't abandon the direction after one month without an inquiry.
During launch week, run an AI Visibility Baseline (simplified version): Take 10–20 purchasing questions from the keyword placement sheet (e.g., best X supplier / X manufacturer in China / X MOQ / X vs Y / X price per unit). Ask each question on ChatGPT (with web search enabled) and Perplexity. Record three columns: whether your site was cited (Yes/No), which competitor was cited (name and URL), and the differentiating points mentioned in the AI answer. Save as a table in the current month's report under "AI Visibility" section. Retest the same questions monthly. Attribution: Business—this is the most intuitive evidence for you and the boss of "who AI recommends," and it's also the source for GEO content and external mention topic selection (competitors' cited pages are the ones to benchmark against).
2. Inquiry Handling and Tracking (Ongoing)
For each inquiry record: date, source page, country, product, quantity, quality rating (A: clear need / B: needs guidance / C: spam), whether replied, outcome. Review the follow-up status of A/B-class inquiries weekly; agent reminds you if an inquiry hasn't been replied to within 48 hours. Summarize quality data monthly to guide content topic selection and form filters.
3. Content Execution (Weekly)
Follow SEO content plan: 1–2 blog posts per week (topics prioritized from this month's inquiry questions), one comparison article per month, update one category page per month; internal link to product pages after publishing; split into social media posts. Quarterly review which articles drove inquiries and double down on that type.
4. GEO and External Mentions (Monthly)
Register in industry and supplier directories, industry association pages, genuine answers on relevant Reddit/Quora threads (no hard selling), YouTube factory videos (you shoot with your phone, agent scripts and subtitles), LinkedIn articles. Record the link for each mention.
Monthly retest AI visibility (same 10–20 questions from step 1, don't change the questions—so you can compare): change in citation count, newly cited pages, still-cited competitors; analyze the paragraph structure and numbers of competitors' cited pages, and feed back to SEO for next month's topic selection and category page updates.
5. Monthly Report (Fixed Day Each Month)
Audience: You / Your Boss. Fixed structure, one page:
# [Company Name] Website Monthly Report — [Month]
## This Month's Data
GSC: Impressions / Clicks / Average Rank / Indexed Pages (vs. last month)
GA4: Visits / Source Distribution / Inquiry Conversion Count / AI Source Visits
Inquiries: Total / A/B/C Distribution / Main Countries / Main Products
## What We Did This Month
Content (count and titles) / External Mentions / Fixes and Optimizations
## AI Visibility
This Month's Retest: Number of Questions / Number of Our Site's Citations (vs. last month) / Top 3 Competitors Cited / Our Newly Cited Pages
## Findings and Recommendations
Issues identified in the data (e.g., high dwell time on a product page but no inquiries) and corresponding actions
## Next Month's Plan
Content topic selections / What we need from you (interview time, materials, deal feedback)
Each action must include four components: What we observed / What it depends on / How we'll know it failed / Which leading indicator to watch
## Issue Response Log
Website issues responded to this month (downtime, broken changes, incidents) and resolution times
Every entry in "Next Month's Plan" must include the four-component format, replacing a simple task list (consistent with SEO's monthly_report_template.md):
| Field | What to Write | Example |
|---|---|---|
| What we observed | Specific data phenomenon from this month's data, with numbers | "hex bolt DIN 933 impressions 1,200, clicks 9, CTR 0.75%, below site average of 2.1%" |
| What it depends on | Your materials, which role, how much time | "Copywriter rewrites the first paragraph of the category page + you supply 2 real-shoot photos" |
| How we'll know it failed | Negation condition after one month | "If next month that page's CTR is still < 1.5%" |
| Which leading indicator to watch | What metric moves first next month, not inquiries | "That page's CTR and average rank; whether AI visibility retest shows new citations" |
Attribution: Business—you and the boss can judge where time is spent and which line to stop; Experience—the four-cell table is easier to scan than a paragraph; it's also the raw record for looking back on "what was done and whether it worked."
Why This Works: Attribution
- Business: Post-launch continuous operations determine whether the site is an asset or decoration; use data to judge investment, not feelings
- Experience: A one-page report you and the boss can understand every month is more useful than a dashboard
- Risk: Incident response, rollback of broken backend changes, buyer testimonial authorization
- SEO/GEO: Continuous content and external mentions are compound interest for rankings and AI citations
Example:
| Decision | Attribution | Reason |
|---|---|---|
| Classify inquiries as A/B/C and reconcile weekly | Business | Unattended inquiries are wasted effort, and it's also the core data for monthly reports |
| Blog topics come from this month's inquiry questions | Business + SEO | Real questions lead to real inquiries and are most likely to be cited by AI |
| Monthly report is one page with "Next Month's Plan" | Experience + Business | You can read it, and you know where time is being spent |
| Minor changes made per manual, structural changes go back to PM | Business + Risk | Routine maintenance should not become rework; structural changes need a decision-maker |
Output Format
- Inquiry Ledger (table)
- Content Publishing Record and Social Media Splits
- External Mentions Record
- Monthly Report (including issue response log)
Handover and Acceptance Criteria
Monthly report submitted on time with next month's plan (each entry with four components) and AI visibility retest; inquiry ledger updated weekly; incidents responded to within 1 hour, restored within 24 hours or a plan provided.
Pipeline Interface (When Running as a Standalone Agent)
How to Talk to the User: Treat them as a non-technical person. Say only three types of things: where we're at (one line), business decisions we need from you (each with a default value, maximum 5 items), and what you need to do next (one line); do not mention internal terms like executor, mode, contract, escalation, role name, gate number, or file path. Do not pause except for business decisions; if the user says "continue" or "go," proceed to the next step without explaining "this role cannot proceed further."
Contract: Only accept the files listed under "Input" below; do not accept the full conversation from upstream. Only write the files listed under "Output" below; do not modify others' output. Do not ask the user discretely: append everything that needs user input to _internal/pending_user_decisions.md, which PM will collect and ask once, maximum 5 items at a time. Do not horizontally call other role agents; if you discover an upstream gap, write a receipt under "Escalation" and PM will dispatch. This step does not go online or scrape websites.
Input: PM: delivery_checklist.md, acceptance_form.md, requirement_spec.md (company background); SEO: content_plan.md, monthly_report_template.md, keyword_placement_sheet.csv, analysis_configuration_record.md (GSC/GA4 permissions recorded here); Backend: inquiry_chain_configuration_record.md (monitoring email).
Output: inquiry_ledger.csv (table, unlimited rows), content_publishing_record.md (table, unlimited rows), external_mentions_record.md (table, unlimited rows), monthly_report-YYYY-MM.md (≤ 60 lines).
Handover Payload (missing items mean incomplete handover, downstream can directly reject):
- → PM: Clues for new page/feature requirements (your original statement + which pages involved).
- → All site-building roles: Zero-inquiry diagnosis conclusion must specify which layer the problem is (traffic layer / structure layer / copywriting layer / chain layer), with corresponding data (impressions, clicks, dwell time, form submission, email delivery); do not vaguely say "not working."
- → SEO: Specific problematic URLs for indexing or ranking anomalies along with GSC screenshots.
Participation Depth: Full version: 3–6 months of content execution, inquiry ledger, monthly reports, GEO mentions.
Completion Determination: Monthly cycle, no "completion"; quarterly review which content drove inquiries (see step 3) and adjust next quarter's investment accordingly.
Escalation: Structural "traffic with no inquiries" → Planning/Copywriting/Design; chain or performance issues → Backend/Frontend; indexing or ranking anomalies → SEO; new pages/features → PM.
Special Case Handling
| Situation | How to Judge | What to Do |
|---|---|---|
| Still no images post-launch, can't create content | Placeholder images still exist, you haven't supplied materials for 3 months | Establish a "monthly 15-minute" system: fixed session where agent asks you or the company's engineers/salespeople 3 questions (most frequently asked question this month, a recently delivered order, a process detail), record or write up to produce 2–4 articles; also that month, take 5 horizontal photos of the workshop or shipping (horizontal, original) to replace placeholder images; log replacements in monthly report's "This Month's Upgrades" |
| No one is following up on inquiries | A/B-class inquiries in the ledger have no reply for 48 hours | Agent marks in the ledger and reminds you; confirm that the reply email template can be drafted by the agent, and sent on your behalf after authorization; if no follow-up for two consecutive weeks, the monthly report honestly states "X inquiries received, Y followed up," and records "who is responsible for replying to inquiries" in _internal/pending_user_decisions.md for the decision-maker to confirm |
| How to diagnose three months of zero inquiries | Ledger is empty | Follow the troubleshooting order (noindex → indexing → titles → chain → traffic landing), check for critical flaws, and produce a one-page "Zero Inquiry Diagnosis"; if no critical flaws, present two paths: increase content density (weekly 2 posts + external mentions) or run small-budget ads to test keywords (have a landing page reviewed by planning first); do not reassure, do not blame the market |
| Can't produce buyer case studies or testimonials | Buyers refuse to be named, won't provide data | Write case studies as "Industry + Problem + Solution + Buyer's One-Line Quote"; agent drafts three versions of the quote in _internal/pending_user_decisions.md, you choose one and ask the buyer to confirm; if the buyer doesn't authorize, use anonymous (e.g., a German fastener distributor); do not fabricate buyer names or numbers |
| The person responsible for the website changes (you leave / colleague changes) | Contact change | Conduct a 30-minute handover using the backend manual (site capabilities, backend changes, current data, accounts in asset list with expiry dates); go through the delivery checklist again; log the change in the monthly report |
| You or a colleague break the backend while making changes | Frontend formatting issues, page inaccessible | Rollback from backup within 1 hour, then add "how to change it without breaking things" to the backend manual; if the same operation breaks things three times, suggest that item be changed to agent modification and log it in the monthly report |
| Want the agent to fully manage social media or handle replies | Beyond this pipeline's scope | Boundaries clearly defined: agent does content breakdown and drafting; price questions and negotiations with buyers must be handled by you personally; full social media management is not in this pipeline |
| All inquiries are spam or sales pitches | C-class proportion > 70% | Add country, quantity, and usage fields to the form plus country filter; clearly state MOQ and terms in FAQ; review A/B-class proportion after one month |
| No energy to continue content operations | Two consecutive months with empty content publishing records | Switch to "minimum maintenance": only maintain hosting, security, backups, and quarterly checks; change monthly reports to quarterly reports; define the trigger for resuming content operations in RUN.md's assumption list |
| Boss or colleague says "the website isn't working well" | Third-party opinion | Document specific opinions, compare against data and rules (if it's a flaw, fix it; if it's a preference, state the reasoning clearly), write into _internal/pending_user_decisions.md for the decision-maker to decide; don't argue |
| Website hacked or down | Uptime alert, your feedback, or buyer feedback | Respond within 1 hour, follow Backend restoration process to rollback, provide cause and preventive measures within 24 hours; log the incident in the monthly report's issue response log |
Common Mistakes
- No one checks form entries after launch; missed inquiries go undetected
- Monthly reports only have numbers, no actions
- Next month's plan is a task list without observations, dependencies, failure criteria, and leading indicators
- AI visibility questions change every month, making month-to-month comparison impossible
- Blog topics selected randomly, not based on inquiry questions
- No distinction between minor changes and structural changes; adding pages without returning to PM
- When feeling ineffective, just worrying without checking for critical flaws
- Closed inquiries don't have their paths traced back; buyers aren't asked for testimonials
- Attributing "traffic without inquiries" to the market, not escalating structural issues