CTO personal branding guide: technical storytelling on LinkedIn
A CTO personal branding guide is a way to publish technical storytelling on LinkedIn: one decision the engineering organization will live with, written so a staff engineer, a peer CTO, or a customer champion can forward a single paragraph. The post names a constraint, the choice, the cost, a number with a window, and what is still unknown. The profile becomes a public record of how you reason. It is a standing exhibit of judgment, the same kind of exhibit a design review already demands inside the company.
CTOs lose the public version of that work in a predictable place. The interesting sentence is in a staff-meeting doc, an incident review, or a vendor email. The draft that ships is a vision paragraph about "building a culture of excellence," because the real sentence mentions a customer, an unreleased design, or a headcount the board has not discussed. The fix is a smaller artifact. Publish the tradeoff you can defend in public. Leave the hostname, the customer, the person's name, and the unreleased roadmap in the internal doc.
Boykli is the writing and queue layer for that smaller artifact. Create posts drafts from a note, a YouTube URL, an article, a document, or series source text, and the writer has to keep your nouns and drop statistics, studies, quotes, and anecdotes that are not in the source. Voice Agent does the same job from a live interview: three drafts, and only facts that appear in the transcript. The queue publishes an approved draft to the personal profile you connect with LinkedIn OAuth. It does not post to a company Page, and it does not comment, like, connect, or message. That limit is the same product decision described in building Boykli: the words start from something you brought, and you read them before they leave.
Employed engineers publish the change they personally made. That cut is the LinkedIn content playbook for software engineers. Founders who own the price and the refund count use the B2B founder personal branding playbook. A CTO often sits between those two. You can describe the architecture, the reliability bar, and the hiring bar. You often cannot publish revenue, a named customer's incident, or a promise the CEO has not made. This page is that middle seat. A CEO, CFO, or other officer whose public words are not an engineering decision uses the separate guide on C-suite and executive personal branding without ghostwriter agencies.
Logged-in visitors can browse Radar and Algorithm without a plan. Writing, images, carousels, Voice Agent, publishing, and scheduling need a paid seat. BYOK is $29 a month. Managed is $88 a month. Prices are the public list on boykli.com.
Three shapes of CTO writing
| Vision posts about engineering culture | Company Page announcements mirrored onto your profile | Technical storytelling from a decision you own | |
|---|---|---|---|
| What the reader is asked to trust | A mood: excellence, ownership, AI-first | A launch the company account already posted | A named constraint, a choice, a cost, and a number with a window |
| Who can write it | Anyone with a leadership prompt | Whoever owns the launch copy | The person who can be questioned about the decision. A model can draft. It cannot attend the review |
| Who forwards it | Rarely. Peers have the same prompt | People who already follow the company | A staff engineer sending it into a design review, or a champion explaining your architecture to their own CTO |
| Hiring signal | You can talk about culture | The company ships | You can state a tradeoff and say what you did not measure |
| Employer risk | Low, and also forgettable | Low when the copy is already public | High if you paste an unreleased design, a customer name, a security detail, or a board number. Publish only what policy allows |
| Where it lives | Personal profile | Page first. A personal mirror adds little | Personal profile. Boykli's queue has no company-Page destination |
| What you still have next quarter | A list of themes | A list of launches | A library of decisions a candidate can read before the loop |
The middle column of most "CTO personal brand" advice is the left column here: pillars, hooks, and a weekly theme about leadership. Pillars are filing labels. They become posts when each one names a fact from a decision you already signed. How a draft stays tied to that fact, and what a model is forbidden to invent, is the subject of LinkedIn thought leadership automation.
Draft one CTO post from a decision you signed
Create posts and Voice Agent sit on a paid seat. BYOK is $29/month with your keys. Managed is $88/month with platform models and monthly caps.
boykli.com →The five lines a technical story has to contain
Technical storytelling on LinkedIn is a compression of a decision record, not a origin story. Five lines are enough. If one of them is missing, the post is a slogan with your title under it.
- Constraint. The limit that made the choice real: team size, latency budget, data shape, contract, on-call load, a date you cannot move.
- Choice. The option you picked, named specifically enough that a peer could disagree.
- Cost. What that choice gave up. A decision with no cost is a preference.
- Evidence. One number, the window it covers, and the environment. If you cannot name the window, leave the number out.
- Unknown. What you have not measured, or the condition that would make you reopen the decision.
Example note. This is a shape, not a record from a Boykli customer, and not a suggestion to publish a figure your company has not cleared:
Checkout still writes to one Postgres primary at about 40k orders a day. Catalog reads go to replicas. We refused a second write store. Two engineers own the schema, and a dual-write would need a reconciliation job we have not staffed. p95 checkout write was 180ms over the last 14 days, excluding the Tuesday deploy window. We have not load-tested past 70k orders a day. We reopen this if p95 stays above 300ms for a full week.
Long mode can hold that note. Boykli's long band is the room for an objection plus a metric, labeled on the page as 1400–3000 characters. LinkedIn's own composer is plain text with a 3,000-character ceiling. Medium is about 100 to 250 words (400–1400 characters) and fits when the cost is one sentence. Short is about 30 to 75 words (45–400 characters) and fits a single refusal. YouTube, article, document, and series (2 to 8 posts) are the other modes. Tone, hooks, and the call to action are separate switches.
The draft should keep 40k, one primary, two engineers, 180ms, 14 days, Tuesday excluded, 70k untested, and 300ms for a week. It should not add a revenue impact you did not compute, a quote from a teammate who did not agree to be quoted, or a study about "companies that scale Postgres." If finance or comms says the order volume stays internal, delete it before you write. A constraint and a refusal without the rate is still a post. A post with a rate you were told to withhold is a problem the queue cannot fix, because the queue sends the text you approved.
When the decision is a sequence of fields a reader might screenshot, split it into a 2–8 slide PDF. One field per slide: constraint, options, choice, cost, evidence, unknown. The caption repeats the choice in sentences. Slide size, the 1080×1350 portrait frame, and the 220px safe zone are on the LinkedIn carousel dimensions page. Series mode is for a sequence of decisions, each with its own five lines. "Part 2 of 5" is not an opening by itself.
Seven post types a CTO can sign
Each type answers one question a stranger can ask after they read your title. The artifact is the minimum source. If you do not have it, or policy says you cannot publish it, skip the type this week.
| Post type | Who stops | Artifact you need | Mode that fits | How it dies |
|---|---|---|---|---|
| Architecture bet | Peer CTOs and staff engineers | Options, the constraint, the pick, the cost, the reopen condition | Long, or a 2–8 slide carousel | "We are a platform company" with no system |
| Reliability standard | On-call leads evaluating you | The bar, the window, what the number excludes, what you will not promise | Medium or long | "We take reliability seriously" |
| Build versus buy | People in the same vendor decision | What you tried, the limit you hit, what you kept, what you will not staff | Medium | A list of twelve tools with no limit |
| Hiring bar | Candidates and the hiring managers who inherit your loop | A criterion you can say in public. The exercise, the signal, what you ignore | Medium | A leaked question from last week's loop |
| Org shape | Engineers deciding whether to join | The topology, the ownership line, the meeting you removed | Medium | An org chart with names, or a post about "flat culture" |
| Incident you can tell | Peers who run similar systems | User-visible symptom, duration, the change, what you still do not know | Long | "We had an outage and learned to communicate" |
| Sunset | Users and internal teams who depended on the thing | What is going away, the date, what replaces it, what you will not support | Short or medium | A vague "we are sunsetting legacy" with no date |
These seven cover a year of executive writing. A compensation post, a layoff post, and a fundraise post are different jobs. They belong on the profile only when the number is yours to say, the window is named, and you accept the replies. The writer will not invent a salary band or a headcount. If the note does not contain the number, the draft should stay thin.
Architecture bet
This is the type another CTO forwards into their own review. They want the constraint that made one option win.
Keep the example note above as the standard. Name the store, the write path, the staffing limit, the latency, the window, and the reopen rule. Leave the cloud account, the customer who generated the peak, and the unreleased feature that depends on this schema out of the draft. A peer can disagree with "one primary until p95 crosses 300ms." They cannot disagree with "we are scaling thoughtfully."
Reliability standard
Candidates and enterprise buyers both ask what you will commit to. Publish the bar you already operate, with the exclusion.
Example note:
Page if checkout error rate stays above 1% for 5 minutes. We measure that on the API, not on the status page. Last quarter we paged 6 times. Four were deploys we rolled back inside 15 minutes. We do not publish a 99.99% promise. We have not staffed a follow-the-sun rotation.
Medium or long can hold it. The useful lines are the threshold, the clock, the measurement point, the quarter's count, and the promise you refuse. "Customer obsession" does not survive a procurement questionnaire. "We do not publish a 99.99% promise" does.
Build versus buy
CTOs are asked this in every sales cycle and every board meeting. The public version is the limit you hit, not a feature matrix.
Example note:
We ran a hosted feature-flag vendor for one quarter. Flag evaluation added a network hop on the request path, and our p99 budget for that hop was 20ms. The vendor's p99 in our region was above that on three of the four weeks we measured. We moved flags into the app config we already deploy. We will not staff a second flags team. We still pay the vendor for the experiment UI until the contract ends in March.
That note is a decision. A "top 10 tools I love" list is a different product, and it reads like a sponsored roundup even when nobody paid you.
Hiring bar
The public criterion is a gift to candidates and a constraint on your own loop. Publish the exercise and the signal. Leave last week's candidate out of it.
Example note:
The staff loop includes a 45-minute design conversation about a system the candidate already ran. We ask for the option they refused and the number they used. We do not use a take-home. We do not score "culture" as a separate number. A strong design with no measurement is a mid-level signal on this loop.
If the question bank is confidential, publish the shape of the signal. "We ask what you refused and which number you used" is enough for a candidate to prepare, and it does not hand the exact prompt to a coaching site.
Org shape
People join or leave because of ownership lines. Describe the line. Skip the names.
Example note:
Platform is four engineers. Product squads own their schema migrations. Platform owns the cluster, the backup restore test, and the page for datastore incidents. We cancelled the weekly architecture forum. Design reviews happen on the change, with the two people who will operate it. We have not split platform by domain. We reopen that if the on-call rotation passes eight people.
An org chart screenshot with faces is a different post, and it ages the week someone leaves. The ownership line ages when you change the line, which is when you should write again.
Incident you can tell
Peer CTOs forward the mechanism. Legal and comms often allow the user-visible symptom and the duration long before they allow a customer name or a root-cause diagram that exposes an internal control.
Example note:
Checkout returned errors for 22 minutes on Thursday, 09:10–09:32 UTC. Buyers saw a retry message. A config push pointed writes at a replica. We reverted the push. Error rate in that window was 11% of checkout attempts, then under 1% in the next half hour. We have not finished the change that would reject a write against a replica at deploy time. We are not naming a revenue figure.
If the percentages are internal, publish the clock and the revert. The unknown line — the guard you have not shipped — is the part a peer respects. A lesson about blamelessness, with the mechanism removed, is the post everyone else already published.
Sunset
Sunsets are short. The reader needs the date and the replacement.
Example note:
The v1 export API shuts off on 30 November. v2 has been available since May. The difference is cursor pagination and a 90-day retention on the download link. We will not add fields to v1 between now and the date. Clients that still call v1 get a 410 after 30 November.
Short mode fits. A carousel fits when you need one slide for the date, one for the difference, and one for what you will not do.
A weekly workflow from note to queue
One signed post a week is a full cadence for a CTO who still runs reviews. Two is reasonable when two decisions already have records. A daily vision post empties the decision log and starts to borrow language from posts you did not write. Match the count to notes you can defend.
Block 90 minutes. The rest of the week is reading the replies yourself.
- Pick one decision from the last ten days that already has a written record: a design doc, an incident note, a hiring rubric, a vendor thread you can paraphrase, a sunset date in a changelog.
- Copy five lines into a source note: constraint, choice, cost, evidence with a window, unknown. If a line is missing, the decision is not ready to publish. Put it back in the internal doc.
- Delete customer names, hostnames, unreleased feature names, a named person's performance, security control details, and any number finance or the board has not cleared for public use. Read it once as if a customer, a candidate, and a reporter will each screenshot it. They might.
- Sign in on boykli.com and confirm a paid seat. Browsing Radar does not write. On BYOK, save the OpenRouter key before you write. Google AI Studio is only required for Voice Agent and for a YouTube watch when the video has no captions.
- Paste the note into Create posts. Use long mode for a single decision with a number. Use document mode when you want a 2–8 slide PDF of the fields. Use series only when each part has its own five lines. The post generator page lists the controls. Tone and hooks are switches. They do not supply a missing number.
- Read the draft against the note. Remove any statistic, study, quote, or anecdote that was not in the source. Each draft allows two refinements. "Cut the second paragraph" and "keep the 180ms line and the reopen rule" are refinements. "Make it sound more like a CTO" spends a refinement on a guess, because there is no stored profile of you. Managed accounts have a monthly cap of 130 post refinements, inside a broader cap of 210 posts, 150 carousels, 150 images, and 25 Voice Agent sessions. BYOK has no Boykli generation cap. You pay OpenRouter for chat and images, and Google AI Studio for Voice Agent. Platform keys are never used on a BYOK seat.
- If your company reviews executive posts, send this draft before it is queued. Edit in the Library until the comms note is done. The queue publishes the text you approved. It does not know which sentence legal asked you to remove.
- Connect the personal profile with LinkedIn OAuth. Save the weekdays and the one time of day. Preview the draft on a phone-width and a desktop-width feed before the slot. That check is the LinkedIn post preview. Schedule the document post the same way you schedule text. Boykli does not offer a company Page as a destination. Plan Page announcements as a separate, lower-frequency surface with their own copy.
- After the post is live, answer the first technical questions yourself. The product does not comment. A CTO post that sits unanswered teaches the opposite of the judgment you just described. How distribution actually works after the post leaves — retrieval, then ranking, in-network and out-of-network — is the LinkedIn algorithm guide. A specific decision gives a peer a reason to spend time on the post. A vision paragraph does not create that reason.
If the decision is still in your head after a review, replace steps 5 and 6 with one Voice Agent session. Speak the five lines, including the number and the refusal. The session transcribes both sides and returns Storytelling, Educational, and Opinion drafts from that transcript only. It does not read your old posts. Queue the draft that still contains the number and the unknown. Leave the other two in the Library or delete them. The session workflow is on the Voice Agent page. On Managed, a saved session counts toward the 25. The three drafts do not also draw the 210-post allowance.
The headline under your name is a different object. It is capped at 220 characters and should say the system you are responsible for and who it is for. The week's post can be a sunset or an incident without the headline changing. Rewrite the headline when your role or the system changes. The difference between that standing line and a post opening is on the LinkedIn headline page.
What stays off the profile
Technical storytelling fails in public when the artifact is too large. Use this filter before the draft exists.
| Leave in the internal doc | Safe to publish when you own the decision |
|---|---|
| Customer name, contract value, a logo you do not have permission to use | The shape of the workload: orders a day, a latency budget, a team size you are willing to say |
| Unreleased feature names, dates on a private roadmap | A choice you have already made and can defend if a customer asks |
| Hostnames, account IDs, security control gaps, exploit detail | The user-visible symptom, the duration, the revert, the guard you have not shipped yet, in words comms has cleared |
| A named engineer's performance, a promotion, a departure | The ownership line: which team holds the schema, the page, the restore test |
| Board slides, runway, a hiring freeze you have not announced | A bar you already operate: the page threshold, the loop you run, the promise you will not make |
| A teammate's sentence they did not agree to publish | Your own refusal, in your words |
The right column is still subject to your employment agreement and your comms policy. Boykli does not review posts for securities-law, customer-confidentiality, or security-disclosure rules. You do that before the queue.
FAQ
Should the CTO profile or the company Page carry technical storytelling?
The Page distributes announcements to people who already follow the company. The CTO profile carries decisions a stranger can question. Boykli schedules the personal profile only, after LinkedIn OAuth. Write the Page launch in the company's voice, on the company's tools. Write the decision in the first person, with the cost included. Mirroring the Page post onto your profile adds a second copy of an announcement and none of the judgment.
Will this collide with the CEO's posts?
It collides when both of you publish the same launch. It separates when you publish the technical cost of a promise the CEO already made in public. If the CEO says the product now supports a region, your post can say what you refused to staff in order to make that region real, using numbers you are allowed to share. Agree on which decisions are yours to narrate. One decision, one author, is enough.
Can I describe an incident?
Yes, at the altitude comms and legal allow: user-visible symptom, clock, what you changed, what you still do not know. Leave customer names, affected-account counts you were told to withhold, and any detail that would help someone repeat a failure. A timeline without rates is still useful. A root-cause diagram of an internal control is a different document.
How often should a CTO post?
One decision a week is a complete cadence. Zero is the correct count in a week with no artifact you can defend. Daily posting pushes the five lines out of the draft, because you will not have five real lines a day. Put the approved draft on a clock so the post does not depend on you opening LinkedIn at the slot. The queue is the clock. The note is still yours.
Can an AI ghostwriter run this guide?
It can draft from the note or from the transcript. It cannot attend the review. Paste the five lines, or speak them into Voice Agent, then delete anything the draft added. There is no archive upload and no persona that later posts inherit. Tone is a choice on the draft in front of you. An edit changes that draft. It does not retrain the next one. A brief that says "sound like a category-defining CTO" produces the left column of the first table.
What does Boykli cost for this workflow?
Both seats use the same writer, the same queue, and the same personal-profile limit. BYOK at $29 a month uses your OpenRouter key for chat and images and your Google AI Studio key for Voice Agent, with no Boykli monthly generation cap. Managed at $88 a month uses platform models inside the caps above: 210 posts, 150 carousels, 150 images, 25 Voice Agent sessions, and 130 refinements. An AppSumo lifetime code redeemed at /redeem is a BYOK seat, one code for one account. An existing paid seat blocks redeem. Pick BYOK when you already pay a provider and you write often. Pick Managed when you want the platform bill to be the whole bill. Prices are the public list on boykli.com.
Do candidates actually read this?
The ones you want will read one specific post if a recruiter or a future teammate sends it. They are checking whether you can name a constraint and a cost. They are not scoring your posting streak. A single architecture bet with a reopen rule does more in a hiring process than a quarter of culture posts. Write that post so it can be forwarded into a loop without a preface.
Where does the carousel sit?
As a proof format for a decision you already wrote. If the five lines do not exist, the carousel is eight slides of leadership advice. Produce the text first. Then put one field on each slide so a reader can stop on the cost. The caption still has to stand alone, because many people never swipe. Schedule that PDF on the personal profile the same way you schedule the text post.
Queue one decision on your profile
Write from a note you can defend, then schedule it on the personal profile you connect. BYOK is $29/month. Managed is $88/month.
boykli.com →