Semrush AI Writing Assistant: 2024 Feature Updates and What Changed for SEO Teams
A changelog record covering the significant AI writing assistant updates Semrush shipped through 2024 — what features changed, which workflows are affected, and what SEO content teams need to know before relying on the current toolset.
Semrush's AI Writing Assistant sits inside the ContentShake AI and SEO Writing Assistant tools — two overlapping but distinct products that both got meaningful updates across 2024. If you set up a content workflow using either tool before mid-2024, several of the defaults and output behaviors have shifted enough that a review is worth your time.
This record documents what actually changed, when, and what it means for teams using these tools in production SEO and content workflows.
What Semrush Shipped in 2024
The most substantive 2024 changes landed in three areas: the underlying model powering ContentShake AI, the SEO Writing Assistant's real-time scoring integration, and the addition of brand voice configuration. Each of these has different implications depending on how your team uses the tool.
ContentShake AI: Model and Output Changes
ContentShake AI moved to a more capable underlying model during 2024, with the most visible effects on long-form article generation. The tool now produces longer first drafts with more structured heading hierarchies by default. That sounds like an improvement, but it creates a specific problem: the outputs tend to be more generic precisely because the model is trying harder to cover all the angles a broad keyword might warrant.
Teams that were happy with shorter, tighter drafts in 2023 may find the new defaults produce more content to cut. The word count expansion is real — drafts for competitive informational keywords routinely come out at 1,800–2,400 words before editing, up from the 1,200–1,600 range that was more typical before.
SEO Writing Assistant: Live Scoring and Keyword Integration
The SEO Writing Assistant's real-time content score was updated to account for semantic keyword coverage more heavily than raw keyword frequency. This is a meaningful shift. Before the update, teams could optimize the score by inserting target keywords repeatedly; the new scoring penalizes over-repetition and rewards natural usage of related terms.
The readability scoring component was also recalibrated. Content that previously scored in the 7–8 range may now score lower if sentence length variance is poor, even if keyword placement is fine. This has caught some teams off guard — their existing content templates, optimized for the old scoring model, now produce lower baseline scores.
Brand Voice Configuration
Brand voice was added as a configurable layer in ContentShake AI during 2024. The setup flow asks you to paste in sample content from your brand, then generates a voice profile that influences subsequent drafts. In practice, the influence is moderate — it affects tone and formality more reliably than it affects structural preferences or specific vocabulary.
The voice profile doesn't persist across all generation modes. If you switch between the full article generator and the inline "Write with AI" tool, you may need to reapply or recheck whether your brand voice settings are active. This is a workflow friction point that Semrush hasn't fully resolved.
Change Summary Table
| Feature area | What changed | Effective period | Workflow impact |
|---|---|---|---|
| ContentShake AI model | Upgraded to more capable LLM; longer default drafts | Mid-2024 | High — draft length and structure changed significantly |
| SEO Writing Assistant scoring | Semantic coverage weighted more; frequency penalized | Q1–Q2 2024 | Medium — existing templates may score lower |
| Readability scoring | Sentence variance and flow recalibrated | Q2 2024 | Low–Medium — affects teams with rigid content templates |
| Brand voice configuration | New setup flow; voice profile applied to drafts | Q3 2024 | Medium — inconsistent persistence across generation modes |
| Keyword clustering in briefs | Brief generator now groups related terms by topic cluster | Q4 2024 | Medium — brief structure looks different; cluster logic may differ from your own |
What This Means for Active Workflows
The scoring recalibration is the change most likely to affect teams without them realizing it. If you've been using the SEO Writing Assistant score as a pass/fail gate before publishing — say, requiring a score of 8 or above — you should retest your existing top-performing content against the new model to establish a new baseline. A score of 7.5 under the new system may actually represent better-optimized content than an 8.5 did under the old one.
The brand voice feature is worth setting up if you haven't already, but don't expect it to replace editorial review. It's more useful as a rough tone guardrail than as a precision style control. Teams that have detailed style guides will still need a human pass.
Pricing and Plan Access
ContentShake AI is a standalone subscription separate from the core Semrush plans, priced at $60/month as of this record's verification date. The SEO Writing Assistant is included in Guru and Business tier Semrush subscriptions ($249.95/month and above), but the AI generation features within it require ContentShake AI or a compatible add-on.
Teams on the Pro tier ($139.95/month) can access the SEO Writing Assistant's scoring and recommendations but not the AI draft generation. This distinction matters when evaluating whether the tool fits a budget-constrained workflow.
Known Limitations After the 2024 Updates
- Output length inflation: The upgraded model generates longer drafts by default. For transactional or short-form content, this requires more editing time, not less.
- Brand voice persistence gaps: Voice profiles don't carry consistently across all generation surfaces within the tool.
- Score threshold drift: Teams using score thresholds as publishing gates need to recalibrate — old thresholds don't map cleanly to the recalibrated scoring model.
- Cluster logic opacity: The brief generator's topic clustering is a black box. You can't inspect or override the clustering logic directly, which creates friction for teams with established topical architecture.
- No version history for AI drafts: ContentShake AI doesn't maintain generation history. If you regenerate a draft, the previous version is gone unless you've manually saved it.
Who These Changes Affect Most
The 2024 updates are most disruptive for teams that built semi-automated publishing pipelines around Semrush's AI tools. If a human reviews every draft before it goes live, the changes are manageable — just recalibrate your editing checklist. If you were relying on score thresholds or default outputs as a near-automatic publish signal, the scoring changes and output length inflation both require workflow adjustments.
For smaller teams using ContentShake AI as a starting point for blog content, the brand voice feature and the improved semantic scoring are genuine improvements. The tool produces more contextually rich drafts than it did in 2023, even if they require more trimming.
Agency teams managing multiple client accounts should note that brand voice profiles are account-level, not workspace-level — each client account needs its own profile configured separately.
Comments
Join the discussion with an anonymous comment.