SXO Is the Part After “We Ranked”: A Search Experience Scorecard
A ranking only opens the door. This scorecard follows the visit through intent fit, page clarity, performance, accessibility, task completion, lead quality, and commercial outcomes.
XenGrowth blog · Page 7
Field notes for the people responsible for what happens next.
Research, analysis, and practical arguments about AI-assisted discovery, demand, automation, CRM, sales, and the operating choices that connect them.
- articles published
- 215
- topic archives
- 6
Blog
Latest field notes
Read the question before reaching for another tactic.
Each article starts with a decision, failure mode, or buyer question. Sources and uncertainty stay visible so the work can be challenged and improved.
Your Marketing Agent Passed the Demo. Now Give It an Eval.
A smooth demo says little about stale sources, conflicting CRM records, tool failures, or missing approvers. Turn those real conditions into the tests an agent must pass before it earns more authority.
The Approval Gate Is a Product Decision, Not a Slack Message
Decide which AI-assisted actions may run, which must pause, what an approver needs to see, and how the workflow recovers when nobody answers the request.
CRM Deduplication: Choose the Record You Are Willing to Trust
Two similar CRM records are not automatically the same person or company. Match cautiously, preserve consent and history, review uncertain pairs, and fix the source that keeps creating duplicates.
Lifecycle Stages Are Contracts Between Teams
When marketing and sales disagree about an MQL, the dashboard cannot settle the argument. Define the evidence, owner, timestamp, exits, and exceptions that make each lifecycle stage operable.
Lead Routing That Survives Lunch, Leave, and Broken Syncs
An owner ID does not mean a person accepted the lead. Design eligibility, capacity, fallback, acceptance, reassignment, and monitoring around accountable follow-up.
Voice Agent QA Beyond Containment: Did the Call Actually Go Well?
A completed call can still leave the customer misinformed or exhausted. Evaluate the task outcome, disclosure, facts, tools, timing, handoff, effort, and safety with human-calibrated review.
One Customer Action, Two Pipes, One Event: Fix Meta CAPI Deduplication
The browser and server may describe the same customer action. Give that action one durable identity so Meta can deduplicate it—and so a green dashboard cannot hide a broken event contract.
Enhanced Conversions for Leads: The 2026 Migration Your CRM Cannot Fake
Connect Google Ads clicks with offline outcomes, validate the 2026 upload path, and stop teaching the bidder that every form submission is equally valuable.
Attributed Is Not Incremental: Design the Holdout Before Reading ROAS
Attributed ROAS assigns credit; it does not show what would have happened without the ads. Design a feasible holdout, respect conversion lag, and leave inconclusive results inconclusive.
A Creative Test Should Change the Next Brief
A creative test earns its cost when it changes what the team makes next. Preserve the variable, evidence, uncertainty, and downstream quality instead of awarding another short-lived winner badge.
A GTM Plan Is a Stack of Assumptions. Put Dates and Evidence Beside Them.
A working method for testing audience, problem, urgency, message, channel, sales capacity, and economics before a polished go-to-market deck hardens guesses into commitments.
Digital PR for AI Search: Earn References You Would Still Want Without the Algorithm
How to build source-worthy research, expert analysis, and accurate third-party coverage without manufacturing mentions or pretending anyone can guarantee an AI citation.
Your Form Said “Invalid.” Your Visitor Still Does Not Know What to Fix.
When a form fails, the visitor needs a next action—not a red border and a mystery. Use clear labels, useful errors, deliberate focus, preserved input, and accessible recovery.
A Performance Budget for the Page That Has to Earn the Lead
Budget LCP, INP, CLS, images, fonts, scripts, and third-party tags before they accumulate. Then judge speed beside the quality of the conversations the page earns.
The Thank-You Message Is an Operational Promise
“Message received” is a promise about the operating system behind the page. Store the inquiry once, route it safely, set an honest response window, and recover when CRM sync fails.
What Counts as Proof on a B2B Service Page?
Polish can earn a first look, but a serious buyer still needs inspectable evidence. Choose proof that fits the claim, has permission, and makes its limits visible.
Speed to Lead Without the Five-Minute Theater
How to measure inquiry age, define a useful first response, staff realistic service levels, recover missed routes, and test whether faster follow-up improves qualified outcomes in your business.
If You Cannot Publish One Price, Publish the Logic
When a fixed price would be misleading, publish the scope, assumptions, range, exclusions, and path to a firm quote. “Contact us” should not hide every useful detail.
Do Not Send the Whole CRM to an AI Tool Because the Prompt Was Easier
A bounded AI task rarely needs an entire customer history. Map the minimum fields, every copy, each retention period, and the deletion path before connecting the workflow.
The Web Page Is Data. Your Agent May Mistake It for an Order.
How indirect prompt injection reaches marketing agents through pages, email, documents, CRM notes, and tool output—and how to contain it without pretending one system prompt is a firewall.
Give the Agent the Smallest Key That Can Finish the Job
An instruction to “only update notes” is not access control. Separate read from write, scope records and fields, use short-lived credentials, approve material actions, and revoke stale grants.
If the Agent Changed the Campaign, the Log Should Explain More Than “Success”
What to record across prompts, sources, tool calls, approvals, state changes, cost, latency, and recovery so an AI-assisted marketing workflow can be debugged without creating a new data leak.
Evaluate the AI Vendor With Your Workflow, Not Their Best Demo
The vendor’s best demo avoids your worst data and failure paths. Test representative work, data controls, reliability, operating cost, change management, and the effort required to leave.
When the Marketing Agent Gets It Wrong, Stop the Side Effects First
An incident-response runbook for containing outbound messages, campaign changes, CRM writes, data exposure, and queued automations before investigating and safely restoring service.
Query Fan-Out Is Not a License to Publish 40 Thin Pages
AI search may retrieve across several related queries. That calls for stronger topic coverage and clearer evidence—not a doorway page for every wording a model might invent.
A Citation Count Does Not Tell You What the Answer Used
Measure AI-search visibility as a chain: retrieval, citation, prominence, factual use, referral, and business outcome. A displayed source is only one step.
The Non-Commodity Content Audit: What Can Only You Publish?
Replace polished sameness with evidence your team actually owns: decisions, field notes, failure modes, examples, tools, and limits a generic summary cannot reproduce.
ChatGPT Search Visibility: Separate Crawling, Citation, and Referral
A practical audit for OAI-SearchBot access, indexability, source clarity, observed citations, tagged referrals, and the customer journey after the click.
Is This Page Ready to Be an AI Overview Source?
A page-level checklist for blocked answers, weak provenance, misleadingly quotable passages, and experiences that disappoint once someone clicks through.