Moz Tripled AI Visibility Quotas. Track Fewer Prompts Anyway.

On 5 October 2026, pricing trackers logged a quiet packaging change from Moz: AI Visibility quotas on Moz Pro tripled, and the list price did not move. Standard went from 1 dashboard and 50 tracked prompts to 3 dashboards and 150 prompts. Medium went from 2 and 100 to 6 and 300. Large went from 4 and 200 to 12 and 600. Moz’s product notes describe the same step-up and say existing subscribers already have the new limits. No upgrade click. No new SKU.

The same week, Ideogram raised priority credits at unchanged prices: Plus from 1,000 to 2,400 a month, Pro from 3,500 to 7,500, Team from 1,500 to 3,600 per user. Vendors are stuffing more AI units into plans you already pay for. The margin is still in the seat.

For a small SaaS, the scarce resource stopped being the dashboard cap. It is which questions you watch, and whether a mention ever shows up as a trial.

What the extra quota buys

Moz AI Visibility tracks whether named brands appear in answers from ChatGPT, Gemini, AI Mode and Perplexity, against prompts you choose. A dashboard holds a brand, competitors and that prompt list. Share of mentions, rank inside the answer, and cited pages are the outputs that matter. An empty result often means neither you nor the competitors were mentioned. That is a finding, not a bug.

A Standard seat used to watch 50 prompts. That was already enough to fool yourself. Fifty vague prompts such as “best project tool” produce a vanity chart. One hundred fifty prompts, used the same way, produce a larger one.

A two-person SaaS does not need twelve dashboards. It needs one that survives a Monday review. The quota increase is room for a control group: prompts you expect to lose, prompts a rival should win, and prompts tied to a page you can edit this month.

A worked example

LedgerLane is a fictional invoicing tool for freelance studios in the UK and Ireland. The annual plan is £180. The founder already pays for Moz Pro Standard to check the three landing pages that produce trials. She skipped AI Visibility because 50 prompts felt like a second job.

A useful first set fits inside the new 150 and would have fitted badly inside the old 50.

• Fifteen commercial prompts she hopes to win: “invoice software for freelance designers”, “simple invoicing app UK”, “alternative to a named incumbent for studios”.
• Fifteen prompts she expects the incumbent to win: “enterprise billing platform”, “VAT invoicing for 200-person finance teams”.
• Ten support prompts matching pages she already has: “how to chase a late invoice politely”, “what to put on a freelance invoice UK”.
• Ten branded prompts: “LedgerLane reviews”, “is LedgerLane legit”, “LedgerLane vs the incumbent”.

That is 50. She leaves the other 100 empty. Empty quota is the ability to add a prompt the week a customer email reveals a new phrase.

The review question is binary. Did a cited URL on her domain appear, and did a trial that week mention the same phrase in the signup survey? If the chart moves and the survey does not, the dashboard is entertainment.

Why the meter got wider

Answer-engine tracking used to justify a higher tier. A 50-prompt cap is easy to outgrow on paper and hard to defend once other suites offer a similar chart.

Tripling the allowance without a price change does three things. It removes the objection that blocked activation: Moz tells current Standard, Medium and Large accounts the limits are already updated. It makes add-on maths less embarrassing. Public notes have listed extra dashboards around $20 and extra prompts around $50 per 100. A customer who needed 150 prompts last month was looking at an add-on. This week she is inside the plan. Ideogram’s credit bump is the same move on a generation meter. Read quota increases as competitive pressure, not as a target to fill.

What not to do with 150 prompts

Do not paste a keyword export into the prompt field. A keyword is a string people type into a box. A prompt is a sentence with a country, a team size, or a named alternative. “Invoicing” and “what should a two-person design studio in Manchester use to chase late payers” are different instruments.

Do not let your brand name fill half the list. Branded prompts are a canary for reputation and for invented pricing pages. They are a poor discovery measure. If 80 of 150 prompts contain your name, share of mention looks healthy on a week with no trials.

Do not open a dashboard per rival. Standard allows three. Use one for the product, one for a rival you lose deals to, and leave the third for a segment you might enter.

Do not treat a cited domain as a ranking. Models cite roundups, docs and forum threads in the same answer. A pricing-page citation is a different event from a listicle that mentions you in paragraph nine.

Eight steps for Monday

1. Pick one product line and one country. LedgerLane tracks UK and Ireland studio language, because that is who pays £180.

2. Write 15 prompts from the last 30 support tickets and sales emails. A phrase that never appeared in a real message does not earn a first-batch slot.

3. Add 10 prompts you should lose, so a flat line is interpretable. Winning enterprise prompts means the sample is confused. Do not celebrate.

4. Add 10 branded prompts, including one that asks for a feature you do not sell. If models invent that feature, publish a clarification page.

5. Record the cited URL, not just the mention. Each week, open the three cited URLs and mark whether you can update them this month. A docs page with an old price is a liability the tool has surfaced.

6. Join the mention log to one field you already trust: a signup survey, a tagged landing path, or the Gemini campaign parameter on some outbound AI links. A mention with no downstream trace is a hypothesis, not a channel.

7. After four weekly reads, drop prompts that never cite a rival and never match a customer phrase. Spend freed slots on lost-deal language. The larger cap makes the cull painless.

8. Once a quarter, price the seat against the decision it changed. Standard is $99 a month, or about $79 billed annually. If it did not change a page, a comparison block, or a support macro, drop the AI habit even if you keep rank tracking.

Keep it off the rank tracker

Rank tracking answers where a URL sits on a results page you can reload. AI visibility answers whether a model named you and which source it used. A page can sit in position three and never be cited. A help article can be cited and have no ranking you recognise.

Keep the notes separate. One blended score hides a failing page behind a flattering mention rate. Rank tracker for URLs that already convert. Prompt list for sentences a buyer asks before they know your name, plus a short branded canary. Moz also simplified dashboard setup around the start of October. That lowers the cost of a first run. It is not a reason to track more than you can review.

A mirror for your own meter

If you sell a metered AI feature, buyers now expect quota bumps without price changes once a feature stops differentiating. If the meter is the only reason someone is on a higher tier, a competitor will widen it. Put the upgrade on a workflow they cannot get by waiting: a role, a history limit, an approval step, an export an accountant needs. A prompt cap is a weak paywall in October 2026. Moz showed that by declining to charge for the triple.

FAQ

Did Moz raise Pro prices when it tripled AI Visibility?
No. Trackers on 5 October 2026 recorded the increases with prices unchanged. Current Standard, Medium and Large accounts have the new limits.

How many prompts should a solo founder track?
Start around 40 to 60: customer phrases, prompts you expect to lose, support prompts tied to editable pages, and a branded canary. Leave the rest empty.

Is a mention the same as a referral?
No. A mention is a model naming you. A referral is a trial you can join back to that demand. Report them separately.

Should I upgrade for more dashboards?
Not for this feature. Three dashboards cover a product, one rival, and a segment test. Larger plans buy users, campaigns and crawl limits.

What if the dashboard finds no data?
The brands were probably not mentioned. Tighten the prompt, or fix a page instead of adding another competitor.

Does the Ideogram credit increase change this?
Only as context. Both meters rose in early October without a list-price change. Use units that change a page. Do not treat a larger allowance as a target.