Ahrefs belongs on a serious SEO shortlist when one team needs backlink and search research, technical crawling, rank tracking and competitor intelligence in the same workspace. Brand Radar now adds AI-answer discovery and scheduled prompt monitoring. The combination can reduce tool switching, but it also creates an unusually easy buying mistake: paying for the Ahrefs name before modeling which data, limits, add-ons and workflows the team will actually use.
> Distinctive strength: A search team can investigate the same competitor across Ahrefs' backlink and keyword datasets, then test how that brand appears in Brand Radar's sampled AI answers without reconciling separate research systems. > > Where it stops being an advantage: The connection matters only when the team needs both conventional search evidence and sampled AI visibility; localized pricing, add-ons and separate capacity limits can make that combined workflow expensive.
This is not a hands-on quality verdict. BenPicks has not validated Ahrefs' keyword estimates, backlink index, issue detection or Brand Radar coverage against a controlled independent panel. The useful question is narrower and more demanding: will the exact plan and add-ons turn your recurring SEO questions into actions at a predictable total cost, under data terms your organization accepts?
The decision in 60 seconds
| Your requirement | What it means for the shortlist |
|---|
| Backlink, organic, keyword and competitor research in one suite | Strong documented fit; test the exact reports, history and exports your team uses. |
| Technical crawling for owned sites | Plausible even at low cost because verified projects receive special allowances; model page count and JavaScript needs before choosing a plan. |
| Broad discovery of brand appearances in AI answers | Brand Radar is unusually broad, but it is a vendor-constructed sample—not every conversation or measured audience. |
| Tracking a focused set of commercial prompts | Available through check-based custom prompts; cost scales by prompt × platform × location × frequency. |
| A prescriptive content editor | AI Content Helper exists, but Ahrefs should not be purchased as a proven replacement for a specialist editor without a real brief test. |
| Agency collaboration with strict client separation | Mixed fit. Roles exist, but official documentation says some workspace data is shared among members; verify permissions with real test accounts. |
| Confidential AI inputs that cannot be used for model training | Potential hard blocker: current terms authorize training use of User Content supplied to AI Tools unless a separate written arrangement changes that boundary. |
| Public uptime remedy | Not established for ordinary plans; obtain the exact SLA in writing if it is a procurement gate. |
What are you actually buying from Ahrefs?
Ahrefs is a portfolio, not one workflow. The current suite spans Site Explorer, Keywords Explorer, Site Audit, Rank Tracker, Competitive Analysis, Content Explorer, Web Analytics, Bot Analytics, Report Builder, API v3, MCP and other tools. Brand Radar and content/project add-ons can sit beside the core subscription.
That matters because the buying unit changes from screen to screen:
- a monthly plan and user allowance;
- report credits on Starter and Lite;
- rows visible and rows exported;
- crawl credits and maximum pages per project;
- tracked keywords;
- API units and rows per request;
- prompt checks across models and locations;
- separately priced indexes, content capacity and project boosts.
A plan badge therefore cannot answer “What will Ahrefs cost?” The right answer is a workload bill.
How much does Ahrefs really cost for a working SEO stack?
The base subscription is only the first line of the budget. The public US-dollar pricing surface checked on 27 August 2026 listed:
| Plan | Monthly price | Selected published boundaries |
|---|
| Starter | $29 | 200 credits, one unverified project, 50 tracked keywords; no added users or report exports |
| Lite | $129 | 1,000 credits, 5 unverified projects, 100k crawl credits, 750 tracked keywords, 150 AI-prompt checks |
| Standard | $249 | Unlimited fair usage per active user, 20 unverified projects, 500k crawl credits, 2,000 keywords, 300 checks |
| Advanced | $449 | 50 unverified projects, 1.5M crawl credits, 5,000 keywords, 600 checks |
| Enterprise | $1,499 | Annual commitment; from 3 users, 100 unverified projects, 5M crawl credits and 2,500 checks |
Ahrefs advertised up to 17% savings on annual billing. Prices can be localized, taxes are not represented here, and checkout remains the authority for your account. BenPicks omits `Offer` schema rather than compressing regional, plan, seat, usage and add-on pricing into a misleading single amount.
Representative cost scenarios
These are arithmetic scenarios from published inputs, not quotes.
1. Evaluate an owned site: $0. Ahrefs Free provides limited data for verified properties, 5,000 Site Audit crawl credits per project, one AI Content Helper document monthly and related tools. It is useful for checking an owned-site workflow. It is not a trial of unrestricted competitor research or the paid Brand Radar indexes.
2. Small operator with focused AI monitoring: $179/month. Lite ($129) plus the Basic custom-prompt package ($50) produces a $179 subtotal before tax. One additional published $40 Lite user without the prompt package makes $169; the user and Basic prompts together make $219. Lite still has report-credit, row, export and project limits.
3. Content and monitoring team: from $398/month. Standard ($249) plus Content Kit (from $99) plus Basic custom prompts ($50) equals $398. This excludes additional users, Project Boost and PAYG.
4. Broad AI visibility: $948/month. Standard ($249) plus all-platform Brand Radar ($699) equals $948. The all-platform purchase already includes 2,500 custom-prompt checks, so adding the Basic prompt package without a measured need would duplicate capacity.
5. Agency technical stack: $1,049/month. Advanced ($449), five additional users at the published $80 each ($400) and ten Project Boost Pro projects at $20 each ($200) total $1,049 before tax and PAYG. This illustrates why users and project-level add-ons can outweigh the base-plan difference.
Use this formula:
monthly cost = base plan + extra users + Brand Radar indexes + prompt checks/overage + content capacity + boosted projects + PAYG data + analyst/editor time + downstream reporting
Enabled PAYG can create automatic charges. Ahrefs says purchased report credits, export rows, crawl credits and API units last for three billing months including the purchase month. Custom-prompt overage is billed monthly. Cancellation takes effect at the end of the current period, and refunds are generally unavailable except discretionary requests for non-use.
Is Ahrefs Brand Radar measuring real AI visibility?
It measures a large, structured sample under Ahrefs' methodology—not every AI answer and not an observed audience count.
The current official Help article describes more than 405 million search-backed prompts across seven named AI platforms. Ahrefs derives questions from its keyword/People Also Ask data and submits them to supported AI interfaces. It says the collection uses no stored user history, personalization, pre-prompting, normalization or filtering.
That design has real value: a buyer can search existing data immediately instead of inventing every prompt in advance. It also imposes hard interpretation limits.
- Chatbot indexes refresh about monthly, not continuously.
- Prompt coverage is shaped by Ahrefs' search database and collection choices.
- Model versions and platform availability can change; Ahrefs currently notes a collection interruption for Grok.
- Private, authenticated, personalized and differently worded conversations are outside the guaranteed sample.
- A mention is counted once per response even if the name appears repeatedly.
- “Impressions” are modeled from Google search volume associated with the prompt's underlying People Also Ask keyword; they are not counted views of an AI answer.
Treat Brand Radar impressions as a prioritization proxy, not audience analytics. A chart can tell you where to investigate. It cannot prove how many people saw the answer or that one page edit caused a change.
What custom-prompt checks actually cost
One check is:
one prompt × one LLM × one location × one execution
Frequency then multiplies executions. For example, 20 prompts across three platforms and two locations use 120 checks per run. Daily tracking over a 30-day month requires about 3,600 checks; weekly tracking at four runs requires about 480. On the published packages:
- Basic: $50 for 2,500 checks, then $0.020/check;
- Growth: $100 for 7,000, then $0.015/check;
- Scale: $250 for 25,000, then $0.010/check.
At 3,600 monthly checks, Basic would use 1,100 overage checks, adding $22 for a $72 prompt subtotal. Growth would contain the workload at $100. Basic is cheaper in this scenario; Growth buys 3,400 unused checks and more headroom. Make the decision from the matrix, not from the tier name.
New tracked responses can take up to 24 hours to appear. Decreasing frequency reduces check consumption and may be the rational first test when the goal is strategic movement rather than daily alerting.
Can Ahrefs replace a content optimizer or publishing workflow?
Not safely by assumption. Ahrefs documents both content guidance and narrow publishing-adjacent tools, but neither proves a complete editorial system.
AI Content Helper compares a page with competing content and scores topical coverage instead of prescribing keyword repetition. Its assistant can use the current document and competitor articles as context. All plans currently receive one new document monthly; additional capacity is sold separately or through Content Kit.
That is enough to run a representative test, not enough to conclude that it replaces Surfer, Frase, Clearscope or an editorial brief process. A score can increase while factual accuracy, differentiation or readability declines. The success metric should be correction time and publishable coverage—not the score itself.
Site Audit Patches are also easy to overstate. Current documentation limits patches to titles and meta descriptions. Publishing requires an eligible boosted, verified project and either a JavaScript snippet or Cloudflare Worker. Patches are described as temporary and reversible; they can remain active after a boost downgrade even when new publishing is disabled. Any user with project access can draft, while publishing requires 2FA.
IndexNow can notify supporting search engines about selected changed, new, redirected or removed URLs. It does not publish the page and does not guarantee indexing. Ahrefs is therefore not a general CMS replacement.
Are Ahrefs data, exports and API limits sufficient for an agency?
Possibly, but only after modeling properties, clients, report rows, exports and automation separately.
Verified owned projects are unlimited. Unverified projects—often the competitor/client research constraint—run from five on Lite to 100 on Enterprise. Site Audit publishes 100k/500k/1.5M/from 5M crawl credits for Lite through Enterprise and maximum project sizes of 25k/50k/250k/5M pages.
Only internal HTML pages returning 200 consume Site Audit crawl credits. Redirects, errors, resources and external URLs may still be crawled without the same charge, so total URLs crawled can exceed credits used. That distinction should be verified on the buyer's own site rather than inferred from page count alone.
Exports have two boundaries: the monthly allowance and what a given report/export exposes. Current core-plan allowances range from 500,000 rows on Lite to at least 10 million on Enterprise. Exhaustion can block further exports until reset, upgrade or PAYG.
API v3 adds another meter. Published included units are 100k/400k/1M/2M for Lite through Enterprise, with maximum rows per request of 100/250/500/unlimited. Paid requests generally cost at least 50 units, with the actual charge driven by rows and selected fields. The default rate limit is 60 requests/minute, and dynamic load throttling can also produce HTTP 429.
Free test queries and narrow field selection should precede automation. A script that requests every available field can consume materially more units without improving the decision.
Can confidential client data safely enter Ahrefs and its AI tools?
Do not assume so from the existence of a DPA. Review the exact content, workspace and integration path first.
The current terms say AI Content may be incomplete, inaccurate or offensive and requires human review. They give the user any available rights in qualifying generated output, subject to authorized inputs, but explicitly warn that output may not be unique and provide no non-infringement guarantee.
More importantly, the terms authorize Ahrefs to use User Content—including content submitted through AI Tools—to train machine-learning models and develop those tools. The retained public sources did not establish one general account-level opt-out. Until a written contractual exception says otherwise, unpublished strategy, client personal data, licensed material and confidential briefs should not enter an AI feature merely because the workspace itself is approved.
The privacy policy describes deletion of personal data when no longer necessary and when deleted within an account, with longer retention for legal, security, fraud-prevention or financial purposes. It does not give one universal deletion clock for every crawl, prompt, document, connected analytics dataset or derived aggregate.
Ahrefs publishes a DPA, EU SCC paths, security measures and a named subprocessor list. The checked Security Measures page says its systems had not yet been certified and that SOC 2 was planned. That is not SOC 2 certification.
Workspace roles also need a real test. Owner, Admin and Member levels are documented, while Enterprise adds Guest access. Official shared-workspace guidance says workspace data—including search history and keyword lists—is shared among members. For an agency, “has roles” is not proof of client isolation.
When should a buyer choose something else?
Exclude Ahrefs or postpone purchase when:
- the business case treats modeled AI impressions as actual people reached;
- a specialized content editor is the main job and Ahrefs has not beaten it on a real brief;
- confidential AI inputs cannot be used under the current training term;
- workspace sharing cannot meet client-separation policy;
- export, API, crawl, project or prompt economics exceed a focused alternative;
- an independently verified certification or public contractual uptime remedy is mandatory;
- the team wants a CMS rather than temporary metadata patches and search-engine notifications;
- nobody owns the action after a report is produced.
Alternatives should follow the job. Compare Semrush when conventional SEO and AI visibility need different packaging, Surfer SEO when the primary job is content optimization, or start in the AI SEO category to separate research, monitoring and editorial workflows. Use Compare only after declaring the workload and hard gates.
A product-specific Ahrefs evaluation protocol
Use one owned property, one competitor and one market. Freeze 20 commercial questions and 20 technical issues already known from other evidence.
- Verify ownership so free/verified-project allowances are separated from competitor usage.
- Run Site Audit with a declared scope. Record internal 200 HTML pages, total URLs, crawl credits, JavaScript settings, duration and known defects missed.
- Compare Site Explorer and Keywords Explorer findings with Search Console and a second independent source. Record disagreements rather than choosing the preferred number.
- Export the reports the team actually delivers. Measure visible rows, exported rows, truncation and cleanup time.
- Configure five commercial prompts across two AI systems and two locations weekly. Calculate expected checks before saving; compare with actual account use.
- Preserve prompt, platform, location, timestamp and response. Repeat controlled prompts before interpreting trend movement.
- Trace ten Brand Radar citations or mentions to accountable actions. Reject a report whose only output is another dashboard.
- Test AI Content Helper on one real brief. Count factual errors, missing concepts, unsupported additions, revision time and publishable output against the existing process.
- If API or MCP matters, start with free test queries. Record fields, rows, units, 429s, retries, latency and projected monthly cost.
- Invite accounts with the intended roles. Verify project visibility, shared histories, keyword lists, exports, billing controls and offboarding.
- Review AI training use, DPA, subprocessors and connected-data scope before entering confidential material.
- Calculate the full cost, including users, add-ons, PAYG and operator time.
- Apply gates declared before testing: minimum known-issue recall, maximum cleanup time, maximum monthly cost, acceptable data boundary and minimum actionable findings.
BenPicks has not executed this protocol. It is a reproducible reader test, not a disguised claim of hands-on experience.
What remains unknown
The retained public pricing, legal, security, Help Center and developer sources do not establish one generally applicable contractual uptime SLA and service-credit schedule for self-serve plans. Ahrefs' terms instead disclaim a guarantee of continuous, uninterrupted or error-free operation.
If availability is material, request the precise uptime percentage, measurement boundary, exclusions, incident classification, claim procedure, support response and credits in the signed agreement. Until then, the field remains Unknown.
Verdict
Ahrefs is strongest when its search index, technical audit, competitor research and AI visibility feed decisions owned by the same team. It is weaker when bought as a universal answer to content, reporting or AI visibility, because every workflow retains its own meter and evidence boundary.
Start with Ahrefs Free for an owned property or the smallest plan that exposes the required competitor workflow. Add Brand Radar only after a prompt-and-action test proves that broader discovery changes a decision. Before confidential AI use, resolve the training-use term in writing. Before automation, model rows and API units. The right Ahrefs plan is the cheapest configuration that passes those gates—not the most complete plan the pricing page can assemble.
Official sources
- Ahrefs plans and pricing ↗ — plans, users, projects, data limits, Brand Radar, add-ons, billing and refunds.
- Plan comparison ↗ — detailed plan feature and limit differences.
- Ahrefs Free ↗ — verified-site evaluation scope.
- Brand Radar overview and methodology ↗ — coverage, freshness, collection and packaging.
- Custom prompt tracking ↗ — check formula, frequency and allowances.
- AI visibility metrics ↗ — mention, citation, impression and share-of-voice definitions.
- AI Content Helper ↗ — workflow and plan allowance.
- Site Audit crawl-credit accounting ↗ — billable and non-billable crawled URLs.
- Always-on Audit ↗ — monitoring cadence and Project Boost speeds.
- Site Audit Patches ↗ — publishing boundary, deployment and access safeguards.
- Workspace users ↗ and shared workspaces ↗ — roles, access and shared data.
- API v3 introduction ↗ — eligibility, unit costs and rate limiting.
- Terms of Service ↗ — AI review, output, training use, cancellation and warranty boundaries.
- Privacy Policy ↗ — personal-data retention and deletion.
- Security Measures ↗, DPA ↗ and Subprocessors ↗ — security, transfer and procurement evidence.