Voice is difficult to measure, which is one reason most organic-social analytics avoid it. It is also difficult to change, which is the other reason. The industry's default advice — publish more, iterate faster, adapt to the platform — implicitly encourages a brand's voice to drift, both because more posts means more chances to drift and because faster iteration means less time per post to hold the line. The result is that most brand voices, on inspection, have drifted more than the brands themselves realise.
This letter is about what happens when a brand doesn't drift, and what the data says about what that consistency is actually worth.
How we measured voice
We ran a lexical consistency analysis across the last three years of organic-social output for each of the 214 brands in our sample. The technical implementation is not particularly novel — we tokenised each brand's posts, produced a semantic embedding per post, and measured the pairwise cosine similarity across posts within brand across time. The measure of "voice consistency" for a brand is, in essence, how similar the brand's posts are to each other, over rolling twelve-month windows, over a three-year period.
The measure is imperfect. It picks up on choice of vocabulary, syntactic patterns, opinion positions, and framing. It does not fully pick up on tone in the sense a copywriter would use the word. But it correlates well enough with human judgement — we blind-tested it on a subset of the sample and it agreed with our editorial team's classification 84% of the time — that we're willing to lean on it.
The finding
Voice consistency is, across the sample, the single strongest single-variable predictor of weighted engagement rate. Stronger than posting frequency. Stronger than production budget per post. Stronger than platform tenure. Stronger than follower count (which, on inspection, is a downstream variable rather than an upstream one).
The strength of the relationship is such that a brand in the top quartile of voice consistency, on our measure, has, on median, roughly 3.1x the weighted engagement rate of a brand in the bottom quartile, holding other measured variables constant. The relationship is broadly linear across the middle of the distribution and appears to accelerate slightly at the top — brands with unusually consistent voice for three-plus years show engagement rates disproportionate to their other characteristics.
The reverse also holds. Brands that pivoted voice materially in the last three years — a rebrand, a change of creative agency, a shift in strategic direction, a founder change — show, on our measure, a persistent decline in weighted engagement rate that takes, on median, 18-24 months to recover from. The recovery, when it happens, produces performance that is broadly consistent with the new voice's consistency going forward. The intervening 18-24 months, however, are almost always worse than either the previous voice or the eventual new steady state.
Why the market undervalues this
Voice consistency is undervalued, on our reading, because the incentives around it are structurally difficult.
The person maintaining voice consistency at a brand is typically a mid-level social manager or content lead. The person changing the brand's voice is typically an incoming CMO, a new creative agency, or a founder responding to short-term pressure. The person maintaining consistency has less institutional power than the person driving change. The change, when it happens, is presented as strategic. The consistency, when it is maintained, is presented as caution. The rewards flow to the changer; the risks flow to the consistent.
The evidence, over rolling three-year windows, does not support this distribution of rewards and risks. The brands whose social managers held voice against change pressure show, on aggregate, materially better performance than the brands whose new creative directions were implemented with executive enthusiasm. But the social manager who held the line is rarely credited for the performance, because the performance manifested slowly. The CMO who drove the pivot is rarely blamed for the subsequent 18 months of decline, because by the time the decline manifests, the CMO has usually moved on.
"The person who most reliably improves an organic-social account, over a five-year window, is the person who was there for all five years and refused to change what was working. That person almost never gets credit for it. The industry structure rewards the pivoter and punishes the persistent, and the platform's engagement math rewards the reverse."
What to do with this
If you are running an organic-social function, three practical implications.
First, treat voice as a permanent asset, not a campaign variable. Document what your voice is, in specific terms, at a level of detail that would let a new writer produce work in it without further instruction. Review the document annually with your best writer or editor. Update the document only when the underlying brand has genuinely changed.
Second, when a new stakeholder — a new CMO, a new agency, a new founder — proposes a voice pivot, the burden of proof should sit with the pivoter, not with the person maintaining continuity. The default should be continuity. The pivot should require an articulable strategic reason, evidence that the current voice is failing on measurable dimensions, and a specific plan for how the new voice will preserve what was working while changing what was not. Most proposed voice pivots do not clear this bar.
Third, when the pivot does happen — for legitimate reasons or otherwise — plan for the 18-24 months of degraded performance and communicate the expectation clearly to the executive team. The pivot is not free. The pivot is expensive. The expense is worth paying only if the strategic reason is genuine and the eventual new voice is materially better than the old. Both of those conditions are met less often than proposed voice pivots would suggest.
A closing note
The brands in the top quartile of voice consistency in our sample are, almost without exception, brands whose social manager has been in role for at least four years. The correlation between social-manager tenure and voice consistency is roughly 0.68 in our data. This is not a coincidence.
The retention of the social manager, in other words, is one of the most consequential decisions the brand's leadership makes on the organic-social function. Losing a good social manager is not a personnel change. It is a brand-voice change, whether or not anyone frames it that way, because the incoming manager will almost inevitably reshape the voice in their own direction — and even if they try not to, the audience will notice the difference.
Compensate your social manager appropriately. Retain them. Give them the authority to hold voice against pressure from elsewhere in the organisation. This is, on the data, the single most valuable thing a leader can do for their organic-social function. Almost none of them do it.