CartCited

Plain-language tool comparisons for small online stores. Prices last checked October 26, 2026.

Methodology · Last updated October 26, 2026

How CartCited compares tools

Launch pages are research comparisons. They use public prices and documentation. They are not hands-on reviews, and they do not carry a score.

What a launch page is allowed to say

  • A price, limit, trial length, or feature that appears on the vendor's site or help center, with the URL and the date we read it.
  • A figure we calculated from those published numbers, with the formula on the page. Cost per tracked prompt is one of those figures.
  • A conflict between two official pages, stated as a conflict rather than resolved by guesswork.
  • Who a published plan is aimed at, in plain language ("15 prompts is a small set"), without claiming we measured accuracy.

What we do not publish yet

Each tool block is labeled Research-based, not yet tested until a logged test exists. Until then we do not publish:

  • Prompt-tracking accuracy against a manual check of the same prompts.
  • How stable the numbers are from one run to the next.
  • Setup time, interface quality, or how useful the recommendations are.
  • Widget speed, email appearance, or import flows for review apps.
  • Screenshots, customer quotes, traffic numbers, or "we used this for 30 days" claims.

Numeric weights and star scores stay unpublished until that testing is logged. A weight that produces a score we did not measure would be a fake rating. The criteria we will weight later, once a test log exists, are: agreement with manual checks, whether the engines named on the plan match what a small store needs, setup time, and whether the prompt or order limit fits a small catalog. Price stays in its own column. It is not a quality score.

What "hands-on tested" will mean

A page can move off the research label only when the log includes:

  • The dates of the test and the plan that was actually turned on, including trials.
  • For AI visibility tools: the same fixed prompt set in each tool for at least 14 days, plus manual spot-checks, and a note of the location used.
  • For Shopify apps: the apps installed on a development store or a real trial store, with setup steps and what we could see in the widget and the review request.
  • Null results written with the same prominence as positive ones.

A free account or a short trial can be labeled "Trial only" with the dates. It is not the same as a full test. If a vendor gives us a free account, the page will say so in the disclosure next to that tool. The vendor does not get to approve the article. That follows the FTC Endorsement Guides, 16 CFR Part 255.

Affiliate links do not rank tools

Whether a company will pay us is not a criterion. Tools with no program we can join, including Profound and Judge.me, stay in the comparison. FTC Endorsement Guides, 16 CFR 255.4, Example 3, treats rankings that are paid for as deceptive. Our affiliate disclosure lists which links are placeholders and which links are ordinary.

Tables of paid plans are sorted by the verified entry price, or by a stated order such as order volume, and the page says which. They are not sorted by commission.

Small-store scope

The reader we write for runs a US online store, alone or with a team of roughly ten people, usually on Shopify. A five-product store does not need an enterprise prompt volume to start. We say that as a fit comment about published limits, not as a measured outcome.

Ahmed also runs a small store in Pakistan and plans to use it as a methods demo for AI visibility. That store is cash-on-delivery and priced in rupees. Its results will not predict what a US Shopify store sees. No case-study numbers are published, because the baseline has not been run. When it is, the raw table will be public, including weeks with zero mentions.

Freshness

Prices move. Each comparison shows a last-checked date. The date changes only when someone re-reads the source and edits the page. A changelog at the bottom of each article records what changed. The pricing index is the page we expect to update first.

Corrections

Email [email protected] with the URL, the sentence, and a public source. We date the correction in the changelog. Vendors may flag factual errors. They cannot buy a change in position. The editorial policy is the longer version of this rule.

Structured data

Article, breadcrumb, and FAQ markup matches the visible text. Comparison pages use an ItemList of the tools named on the page. We do not use aggregate ratings or review stars. Google's documentation says structured data must match visible content, and that AI Overviews do not require a special markup type beyond ordinary SEO.

Source: Google Search Central, AI features.