A growing share of marketers are pulling back on AI visibility tools after months of inconsistent outputs, according to Digiday's reporting on the category. The software — which scores how often a brand surfaces in ChatGPT, Perplexity, Claude, and Google AI Overviews — drew heavy spend from agencies and in-house teams. Buyers now say the numbers drift too much to defend in budget reviews.

The category is real. The question is whether the tools work.

What do these tools actually do?

Vendors pitch three core deliverables:

• Prompt-level citation tracking across major LLMs
• Share-of-voice scores inside AI answers
• Competitive benchmarking against named rivals

Annual fees typically run from low five figures for self-serve seats to six- and seven-figure enterprise contracts, with pricing tied to prompt volume and brand coverage.

Why are buyers pushing back?

Marketers running the same prompt set on competing platforms report different share-of-voice leaders for the same client in the same week. Identical audits return materially different rankings within 24 hours. Citations shift without any change to the underlying content or media spend.

That breaks the basic procurement test. A tool that cannot produce repeatable measurements does not survive a Q4 budget review, especially at six figures or more.

What are agencies and brands doing instead?

Several large advertisers are moving the work in-house. Internal teams cite lower cost, full control over the prompt taxonomy, and the ability to share raw outputs with brand and legal teams without licensing restrictions. The trade-off is headcount: someone has to maintain the prompt set and run audits on a recurring cadence.

For teams already running SEO operations, the lift is manageable. For teams that don't, the math gets harder.

What should buyers ask vendors?

Three questions separate credible platforms from the rest:

1. Can you share the exact prompts used in your scoring?
2. What is the variance when audits are re-run on the same prompt set within 24 hours?
3. Will you let a third party reproduce your results on a defined prompt set?

Vendors that decline any of the three are telling buyers something. Vendors that publish methodology, expose raw prompt logs, and accept independent verification stand a better chance of holding their book through the next renewal cycle. Vendors that cannot defend their numbers will not.

What happens next?

Expect more procurement pushback through year-end as annual contracts come up for renewal. Agencies that already paused renewals are asking vendors to disclose methodology, raw prompt logs, and the sampling logic behind their scores. Some vendors have responded with methodology documents. Others have not.

The category will not disappear. AI-generated answers are a real surface for brand discovery, and someone has to measure them. The vendors that survive will be the ones that treat repeatability as a feature, not an obstacle.