Standing vs Scrunch AI
A pure-play with an agency tier and a focus on how AI agents experience a site.
We make Standing, so read this accordingly. Every claim about Scrunch AI below comes from their own published pricing, documentation or help centre, verified August 6, 2026. Where they have not published something we say so rather than guessing, and the first section is where they beat us.
Where Scrunch AI is better
- Agent-experience analysis of how AI crawlers see a site, which is a genuinely different angle.
- An agency tier and multi-client structure we do not have.
- A larger effective basket than ours even after discounting variants.
The prompt basket
Scrunch AI: Core $250/mo for 125 prompt variants across 4 LLMs, Agency Core $500 for 250 variants.
Prompt variants, not prompts, and their own documentation concedes that every seed prompt is multiplied by the number of platforms it runs on. So 125 variants is roughly 31 distinct questions. This is the only rival whose real basket is the same order of magnitude as ours, and it is still about 2.5 times larger.
Standing: a fixed basket built from your category, priced per domain rather than per prompt. It is the smallest basket in the category and the only fixed one, which is the trade: fewer questions, but the same questions every scan, so a change in the score is a change in the answers rather than a change in what we asked. It also means two customers in one category are directly comparable.
How many times each question is asked
Scrunch AI: Not published. Across their methodology FAQ, accuracy FAQ, pricing FAQ, metrics guide and prompt strategy guide there is no run count anywhere, and brand presence is defined as present over total without ever defining what constitutes total per cycle. Their documented answer to variance is temporal aggregation, which cannot separate change from noise at any single point in time.
Standing: 5 runs per prompt per engine on paid scans, 3 on the free check, published. Roughly 300 responses per paid scan. This is the number that decides whether a score is a measurement or a single draw from a distribution.
What each publishes about uncertainty
Scrunch AI: None. They substitute a waiting period for a threshold, advising users to monitor trends in two to three week windows before calling a change a signal, which names no magnitude and no test. Their own blog states the failure mode precisely: that AI responses vary from session to session, so a single check does not tell you whether you are appearing consistently.
Standing: a 95% confidence band on every score, shown in the product, with alerts gated on whether a move exceeds that domain's own historical drift. An engine that fails more than half its calls reports that it did not respond rather than a measured zero. Our own weakness, stated: the band covers within-scan sampling and does not cover platform drift between scans, which is why the model version is on the report.
Where Standing is better
- We publish a run count; they do not, despite their own blog naming single-check unreliability as the problem.
- A confidence band and a significance threshold rather than a two-to-three week waiting period.
- $99 against $250, and prompts counted as questions rather than as variants.
Which to pick
Choose Scrunch AI if
You want agent-experience analysis and are running client work at agency scale.
Choose Standing if
You want a magnitude threshold rather than advice to wait three weeks and squint.