Voice drift is the silent failure mode of LinkedIn ghostwriting. The engagement numbers look fine. The posting cadence is steady. The client is not complaining. But somewhere around month 4, the posts stop sounding like the founder and start sounding like a founder — generic, polished, optimised for the platform rather than the person.
We have seen this pattern in roughly 38% of ghostwriting engagements we have audited that ran longer than 90 days. The drift is slow enough to miss week-to-week and obvious in a side-by-side read of month 1 versus month 4. By the time the founder feels it, the brand pattern in the feed has already shifted.
This is how we detect voice drift early, why it happens, and the four-step recalibration we run when a feed starts to lose the founder.
What Voice Drift Actually Sounds Like
Drift does not look like bad writing. It looks like competent writing that could belong to anyone in the category.
Three textures appear at the same time:
- Sentence rhythm flattens. Month 1 had short punches, parentheticals, the occasional one-word paragraph. Month 4 reads in even 14-18 word sentences. The ghostwriter has unconsciously normalised to a "good LinkedIn cadence."
- Specific numbers get replaced with category language. "We cut TACoS from 18.4% to 11.9% on a $6.2M brand" becomes "We help brands lower TACoS significantly." The proof points soften into claims.
- Opinion gets sanded down. Month 1 named names, took positions, called out anti-patterns. Month 4 talks about "challenges" and "opportunities" without telling anyone they are wrong.
If you read 5 month-1 posts and 5 month-4 posts back to back and could not tell which founder wrote them, the voice has drifted.
The Three Drift Signals We Track
We audit voice every 60 days on long-running engagements. Three quantitative signals catch drift before the founder notices.
Signal 1: Specificity ratio
We count proper nouns, dollar figures, percentages, and named scenarios per 200 words. The founder baseline locked in week 2 of the engagement becomes the benchmark.
On healthy engagements the specificity ratio holds within 20% of baseline across 6 months. On drifting engagements it drops 40-60% by month 4. Generic content is faster to produce and harder to mess up, so under deadline pressure the ghostwriter unconsciously trades specificity for safety.
Signal 2: Opinion density
We count load-bearing claims per post — sentences that take a position someone in the category could disagree with. Baseline is usually 3-5 per long-form post. Drift shows up as 0-1 per post, replaced by descriptive setup and "here is what we are seeing" framing.
A post with zero load-bearing opinion is not a founder post. It is a category-trend post. Those get reach but not pipeline.
Signal 3: Save-to-like ratio
Saves are the truest engagement signal for founder content because they signal the reader wanted to return. We documented why in a recent post on the LinkedIn saves ranking signal.
On voice-aligned content the save-to-like ratio sits around 1:6. On drifted content it slides to 1:14 or worse — the posts still get likes from the algorithm-aware audience, but ICP readers stop saving them because the content has lost the specific operator detail they would return to.
When all three signals move together, voice has drifted. One signal alone can be a content mix issue.
Why Voice Drifts in the First Place
Four mechanics drive almost all of the drift cases we have seen.
The ghostwriter stops doing fresh founder calls. The first 30 days have a heavy capture cadence. By month 3 the rhythm settles into "send the doc, we will write it up." The well of fresh phrasing, fresh stories, fresh frustrations runs dry, and the ghostwriter falls back on what they have already heard.
The post archive becomes the training set. Once 40-50 posts exist in the founder's feed, the ghostwriter starts writing from the archive rather than from the founder. It feels efficient. It is actually averaging the founder against themselves — a regression to the mean of every post that performed well, which means the edges get cut off.
Reviewers become permissive. Early in an engagement the founder edits aggressively because the voice is still being calibrated. By month 4 they are skimming and approving. The ghostwriter loses the signal on what is on-voice versus off-voice.
Algorithm chasing gets dressed up as voice. Hooks get rounder. Sentences get more rhythmic. The post structure converges on what is working that quarter on the platform. None of that is wrong on its own — but it pulls the writing toward the platform median and away from the specific person.
The Four-Step Recalibration
When a 60-day audit shows drift on two or more signals, we run a recalibration cycle rather than a rewrite of the brief.
Step 1: Re-capture, do not re-document. We book a 60-minute founder call with one job — get unposted material. Stories that did not make it into earlier posts. Opinions the founder has been holding back. Anti-patterns they noticed on a recent client. We are refilling the well, not refining the strategy.
Step 2: Pull the 5 highest-saved posts from months 1-2. Not the most-liked. The most-saved. Those posts are the truest voice signal — they are the posts the audience returned to because the specificity rewarded a second read. Re-read them with the ghostwriting team and name what is in them: cadence, structure, the type of detail, the type of opinion.
Step 3: Reset the specificity floor. Every post for the next 30 days has to hit baseline specificity ratio. We track it per post, not per month. If a draft comes in under floor, it goes back before founder review. This is uncomfortable for a few weeks and then stops being a constraint because the writer has recalibrated.
Step 4: Schedule a permanent re-capture cadence. Every 6 weeks, a fresh founder call with the explicit goal of new material. Put it in the calendar as a fixed appointment, not a "when there is time" call. The cadence is the only structural defence against drift returning.
We typically see save-to-like ratio recover within 3 weeks of starting the recalibration. Specificity ratio recovers in week 1 because it is mechanical. Opinion density takes the longest — usually 4-6 weeks — because the ghostwriter has to relearn how far the founder is willing to go on positions.
What This Means for Founders Working With Ghostwriters
If you are 4+ months into a ghostwriting engagement and the posts feel fine but you cannot remember the last one you forwarded to a friend, that is voice drift.
Three quick checks before your next ghostwriter call:
- Pull 5 posts from your first 60 days and 5 from your last 30 days. Read them in a single session. Note whether they sound like the same person.
- Count the specific proof points (numbers, named clients, dollar figures) in each batch. If the recent batch is half or fewer than the early batch, drift has happened.
- Ask yourself which posts from the last month you would actually forward to a peer. If the answer is fewer than 1 in 5, the content is reaching but not converting.
The good news: drift is a process failure, not a talent failure. The same ghostwriter who wrote the strong month-1 content can write strong month-12 content. They just need the inputs refreshed.
If your LinkedIn content is starting to feel generic and you want a second pair of eyes on whether it is drift or a different problem entirely, book an audit with our team and we will read your last 60 days against your first 60 and tell you what we see.