LinkedIn Content Strategy for CX and AI Advisors | Netwarm
LinkedIn playbook · Case study
LinkedIn content strategy for CX and AI advisors: how a California customer experience author reached 137 CX and business leaders in four weeks
We read every reaction on 27 posts from a California advisor and author on customer experience and AI. 380 people, 137 CX and business leaders. Here is what brings buyers, what brings peers, and the 5 posts we’d publish next.
Max Sawyer · Netwarm··11 min read
The creator in this case studyA California advisor and author on customer experience and AI
Followers17,086
Posts~7 a week
Sellsadvisory for CEOs and customer leaders on AI decisions, an AI-powered experience assessment, roundtables and a CX newsletter
PeriodOct 7 – Nov 2
Ideal client: CEOs and CX, service and contact center leaders deciding what AI may do for, and to, their customers.The challenge: The advisor writes every day about AI agents meeting customers: the bank that blocks them, the retailer that lets them in, the chatbot that made an offer. Vendors and other CX consultants read along. Which posts reach the leaders who’d hire an advisor?
Key takeaways
These numbers are from 2 November. Your audience has already moved.
New CX leaders turn Hot every week and others go quiet. Run your free scan to see today’s version, with the names.
More than 1 in 3 readers is a CX or business leader.137 people in four weeks: 36.1% of the 380 people who reacted.
AI agent stories bring 1.5x more of them.Posts about AI agents acting for customers reached 44.8% leaders. Other CX business stories reached 29.8%.
15 leaders are regulars.They reacted to 3 posts or more. One of them reacted to 16 of the 27.
01
Who actually reads you, and what the LinkedIn algorithm changed
The advisor posts about customer experience in the age of AI agents: the companies letting personal agents in or keeping them out, tests run with a personal agent called Bob, a weekly newsletter and a roundtable for CX leaders. 27 posts between 7 October and 2 November.
The LinkedIn algorithm shows a post to the people who reacted to the last ones, then to people like them. In CX and AI, those people include vendors and other consultants. Reach can’t tell them apart from buyers. Who reacted can, so we sorted every person who reacted by role.
27posts in four weeks
380people reacted
137CX and business leaders
15came back 3+ times
Audience by roleThe 380 people who reacted, by roleFree
CX, service & business leaders (C-level, VPs, directors, heads of)137 · 36.1%
Leaders are the largest group: 137 people, more than 1 in 3. That is a high share for a niche this crowded with vendors.
Peers are 79 people: CX consultants, advisors, authors and sales leaders at AI vendors. They help reach, and some become partners. The managers and analysts in the third group often work for the leaders and bring the posts into their teams.
SeniorityAll people who reactedFree
Founder, owner, C-level17.6%
Director, VP11.3%
Manager12.9%
Independent5.8%
Other52.4%
IndustriesCX leaders onlyNetwarm
Customer experience, service and contact centers49.6%
Hidden in preview00%
Hidden in preview00%
Hidden in preview00%
Hidden in preview00%
See all industries and company sizes in your free scan.
These rules come from what worked on the advisor’s own profile, and from the best-performing posts by CX and AI advisors on LinkedIn over the last 3 months.
1
Open on a named company and a number. “BMW Group Toronto’s chatbot offered CAD $27,162.79.” A real company and a real figure in line one tell a CX leader this happened, and could happen to them.On this profile: the BMW post reached 55.8% leaders. The DoorDash price comparison reached 34, the most of any post.
Run the test yourself, and report back. Bob, the personal AI agent, is the advisor’s best format: a real errand, what worked, where a human had to step in. Leaders can’t run these tests themselves, so they read them.On this profile: the Bob refund post reached 23 leaders (50%), and the dentist appointment post 54.5%.
Name the decision nobody owns. The best posts end on a choice a leader has to make: who accepts the customer’s permission, who answers when the agent is wrong. That is also exactly what the advisor sells.On this profile: the JPMorganChase post, which separates customer permission from company acceptance, reached 48% leaders.
Bring the roundtable into the feed. Event invitations reach fewer leaders than what was said at the event. Share one case a CX leader raised (anonymised), and the question it left open.On this profile: the case from the roundtable reached 53.8% leaders. Event and roundtable invitations averaged 32.4%.
Show what the AI could not do. The part of the job that was never in the job description (the angry customer, the exception, the judgment call) is what CX leaders worry about. Name it.Across the niche: the Klarna post on what cutting people revealed drew 337 likes.
Ask the people who run the queue. A question for operators only (“vendors, please do not respond”) fills the comments with the leaders you want to meet, and keeps the sellers out.Across the niche: the dealer question drew 131 comments on 48 likes.
When CX leaders read: the best time to post on LinkedIn for CX and AI advisors
We compared the share of CX and business leaders by the day and the hour (Pacific time) each post was published. Hours are shown for slots with 3 posts or more.
Share of CX leaders by posting dayAmong the people who reacted to each postFree
30.8%
Mon
41.0%
Tue
31.6%
Wed
47.5%
Thu
47.8%
Fri
26.2%
Sat
41.0%
Sun
Best time to post, by hourShare of CX leaders among people who reactedFree
39.0%
6:00
35.3%
8:00
41.0%
10:00
Thursday is the best day: 47.5% leaders across 6 posts, and the most leaders in total, 48. Friday is as high but rests on 2 posts. Saturday had the lowest share (26.2%), even though the DoorDash post went out that day.
Most posts go out at 6 a.m. Pacific (9 a.m. on the East Coast), and that slot works: 39% leaders. Put the agent stories and the roundtable cases on Tuesday to Thursday mornings, and keep newsletter promotions for Sunday.
05
5 posts ready to publish to target CX leaders
Each post follows the rules above, in the advisor’s voice. Swap in your own details before posting: the real test or the real case is what makes it land.
1Ready to postNamed incident + decision
Your chatbot just made an offer. Who honours it?Rules 1 and 3: a named company and a number, then the decision nobody owns. Modelled on the BMW chatbot post, the highest share of leaders.
Hook
Your chatbot just made a customer an offer. Who in your company decides whether you honour it?BMW Group Toronto’s chatbot recently offered a customer CAD $27,162.79 for his car. A salesperson then said the maximum was $20,000. The dealership reinstated the offer after the CBC called.Around the same time, a United Airlines chatbot told a customer her $200 travel credit was good for five years. An email later said it expired in weeks.Two different industries. The same gap.Nobody had decided, before launch:
→ What the bot is allowed to promise
→ Which promises a human has to confirm
→ Who answers the customer when the two disagreeCustomers don’t read your chatbot’s disclaimer. They read the offer.And once they have it in writing, “the bot made a mistake” sounds a lot like “we didn’t mean it.”If your chatbot made an offer this afternoon, who would decide whether you stand behind it?
Visual · Text onlyNo image. The two figures ($27,162.79 and $200) carry the post.Check before postingBoth stories come from your own earlier posts: re-check the figures and the sources before posting.
I asked Bob to change a flightRule 2: run the test yourself and report back. Modelled on the Bob refund post.
Hook
I asked Bob, my AI agent, to move a flight by one day. It took him four minutes. And then it took me forty.Bob found the new flight, compared the fare difference and filled in the change form.Then the airline asked for a one-time code sent to my phone. Bob couldn’t read it. I was out walking Izzy.By the time I got back, the code had expired and the fare had gone up.I don’t blame the airline for checking it was me. That check is there for good reasons.But it was designed for a world where the customer is always the one typing.That world is ending. More customers will send an agent first and step in only when asked.So the questions for CX leaders are getting very practical:
→ Which steps really need the customer, and which only need their permission?
→ How long does that permission last?
→ What happens to the price while you wait for the human?How does your journey work when the customer isn’t the one typing?
Visual · Single imageA simple timeline: Bob’s 4 minutes, the code request, the 40 minutes after.Check before postingUse a real errand you ran with Bob, with the real steps and times. Name the company only if you named it in the test.
Four decisions before you let an agent inRule 3: name the decision nobody owns. Modelled on the JPMorganChase post.
Hook
Before you let a customer’s AI agent in, someone has to make four decisions.JPMorganChase decided outside agents can’t operate retail accounts. Delta won’t take bookings from them. Walmart built a gateway to let them shop.Three answers. Each one a decision somebody owned.Most companies haven’t made it yet. Their website’s anti-bot rules are making it for them.Here are the four I’d put on a leadership agenda:1/ What the agent may do
Look up, compare, book, pay, cancel. Pick per journey, not for the whole company.2/ What the customer has to agree to
And how you’ll know the permission is real and still valid.3/ Who owns the outcome
When the action crosses teams (payments, service, fraud), one person answers.4/ When a human takes over
The exact moment, and how the context travels with the customer.Customer permission and your willingness to accept it are two separate decisions.Which of the four has an owner in your company today?
Visual · Single imageA one-page grid: the four decisions, with a column for “owner” left blank.Check before postingThe three company examples come from your October posts: check each one is still current before posting.
What one CX leader measured after the agent went liveRule 4: bring the roundtable into the feed. Modelled on the containment vs sales conversion post.
Hook
A CX leader told our roundtable her AI agent’s best metric was hiding her worst one.After launch, the dashboard looked great.
The agent handled more contacts. Handle time went down. The cost per contact fell.Then her team looked at the same customers a week later.Many had come back. Same problem, different channel.The agent had closed the conversation. It hadn’t closed the issue.What she changed:
→ She stopped reporting containment on its own
→ She added a 7-day repeat-contact rate next to it
→ She asked the agent’s owner, not the vendor, to explain every riseThe dashboard looks less impressive now. The customers call less.A conversation can end long before the customer’s problem does.What does your dashboard say happened after the agent said goodbye?
Visual · Text onlyNo image, or two small lines: containment going up, repeat contacts going up with it.Check before postingUse a case a leader really raised at your roundtable, anonymised, with their OK. Keep their metrics, not these.
The Agent Permission CheckLead magnet: the four roundtable questions turned into a one-page check. Modelled on Kevin Lau’s workbook for CS leaders.
Hook
When should AI be allowed to act on behalf of your customer?That was the question at our last DCX AI Roundtable.The room kept coming back to four questions:→ What does the customer need to agree to?
→ What context do your systems need to act safely?
→ Who owns the outcome when an action crosses teams?
→ When should a person take over?Nobody had a perfect answer. But the leaders who could answer all four for one journey were the ones moving fastest.So I turned the four questions into a one-page check you can run on any journey: refunds, bookings, payments, cancellations.It takes 20 minutes with your team, and it shows you where the decision has no owner yet.Want it? Comment “PERMISSION” and I’ll send it to you.
Visual · Single imageThe one-page check: the four questions down the side, three journeys across the top, cells left empty.Check before postingPut your Netwarm Reveal link behind the one-pager so you can see which CX leaders download it.
Free forever: unlimited scheduling through LinkedIn’s official API, with images, video and PDF carousels.
06
LinkedIn lead generation: lead magnets that work to target CX leaders
Reactions show interest. A download shows intent. The fastest way to find out which CX leaders are serious is to give them something useful and see who takes it.
In customer experience and AI, these formats get the most requests:
Format 1The workbookWorksheets built around the decisions a leader has to make, filled in for their own team.
Format 2The template libraryThe plans, reports and frameworks clients pay for, as templates.
Format 3The one-page cheat sheetThe frameworks leaders have heard of, on one page they can print.
Our pick for you: “The Agent Permission Check”Built from the four questions the advisor already put to the roundtable (customer consent, context, ownership, human takeover), as one page to run on a single journey. Offer it in post 5, and use it as the first step of the paid experience assessment.With a Reveal link, every person who downloads it is shown by name, with their role and how warm they are. Your first Reveal link is free.Create your free Reveal link
07
From reactions to clients
Content brings CX leaders to the profile. It doesn’t say when to reach out. Most never comment, so they stay invisible, and a leader reading every agent story in October may have hired another advisor by January.
On LinkedIn, the right moment to start a conversation is when someone keeps coming back. We call that Hot: 5 or more points over 90 days, where a reaction counts 1, a repost 2, a download or a reply 4.
Hot right now5 CX leaders · not contacted yetNetwarm
5 CX leaders are Hot. See who to message first.See who’s Hot
First messageAn opener for each Hot person, based on the posts they readNetwarm
Hi Dana, you’ve been reading the posts on personal AI agents. Has your team decided what an agent may do in your service journeys yet?Hi Chris, I saw you reacted to the containment post. Are you tracking repeat contacts next to it?
A first line for each of them, ready to send.See the openers
Netwarm sends nothing for you. It tells you who and when, and you write the message. It also works inside Claude and ChatGPT: ask “Who’s Hot today?” and get the list.
08
FAQ
Here, posts about AI agents acting for customers brought 1.5x more CX leaders than other CX stories. Open on a named company and a number, run the test yourself, and end on the decision a leader has to make.
For this advisor, Thursday brought the highest share of CX and business leaders (47.5%, across 6 posts) and the most in total. Saturday brought the lowest share (26.2%).
This advisor posts about 7 times a week, and 15 leaders came back 3 times or more in four weeks. Daily works when most posts give leaders a decision to think about.
Look at who reacts, not how many. Group reactions by role, measure the share of your ideal clients per post, and follow the people who keep coming back. A free Netwarm scan does this for you.
Yes. Netwarm only reads public reactions to your own posts. It never logs in as you, never asks for your password, and never messages anyone for you.
How we made this playbook
We read the public reactions on 27 posts published between 7 October and 2 November 2026, and sorted the 380 people by role. Three duplicate posts were left out. Comments were not counted, so engagement figures are a floor. The 2 November post was too recent to count as an example.
Niche rules come from top posts by CX and AI advisors over the last 3 months. Role groups are approximate, and some leaders sell to the same audience; a Netwarm scan matches people to your own ideal client.