How we check prices and score tools
CinchStack exists to answer one question for people running an online business: which software should I use, and what will it really cost me? This page explains exactly how we get the numbers, how we decide what to recommend, and what we do not claim. If anything here does not match what you see on a page, tell us.
Right now we track 22 tools from 578 vendor pages. The oldest price record was last checked on 26 September 2026 and the newest on 16 November 2026.
How we read a price
Every price starts on the vendor’s own pricing page. Prices are not copied from review sites, marketplaces or other blogs, because those go stale.
- We open the page in a headless Chromium browser (the open-source core of Chrome) set to a US location and save a plain-text copy of what it shows, with the date. Where a page has a monthly and a yearly switch, we save both views. We also save the vendor’s help articles on billing, add-ons and usage fees, because that is where most of the real cost hides.
- We record each plan, add-on and usage fee from those saved copies: the month-to-month price, the effective monthly price when you pay yearly, and whether it is charged per account, per seat or per site.
- Every figure must appear in a saved copy. A script checks every figure in a price record against the saved copy it cites and flags any figure that copy does not contain. If a price exists only behind a login, a sales call or an interactive calculator we could not capture, it does not become a number on this site. It goes on the page’s list of things we could not verify, in plain words.
- A second, separate AI-assisted pass re-reads every figure against the saved copies before anything is published, looking specifically for the common mistakes: a yearly price shown as monthly, a per-seat price shown as flat, a “from” price that assumes a minimum number of seats, or a plan the vendor no longer sells.
All prices are US list prices in US dollars. Some vendors charge different amounts in other countries; where they do, the tool’s pricing page says so. Taxes are not included.
How we work out the real monthly bill
A headline price rarely tells you what you will pay. So for every tool we work out the monthly bill for three standard business sizes: one person, five people and fifteen people. These are sizes of the whole business, not seat counts. Not everyone in a business uses every tool, so each tool is sized for its job within that business: a five-person business gets 3 CRM users, two seats in an SEO tool and five in a docs workspace. Each bill states the seats and volume it assumes, under the figure. We use the same assumptions for every tool that does the same job, so the totals can be compared side by side.
| Job in your stack | One-person business | Five-person business | Fifteen-person business |
|---|---|---|---|
| CRM and automation | 1 user, 1,000 contacts, 2,000 emails and 200 text messages a month | 3 users, 5,000 contacts, 20,000 emails, 1,000 texts | 8 users, 25,000 contacts, 100,000 emails, 5,000 texts |
| Selling (store, courses, funnels) | $3,000 a month in sales, 1 staff account | $30,000 in sales, 3 staff accounts | $150,000 in sales, 8 staff accounts |
| Website | 1 site, 1 editor, about 20 pages | 3 editors, about 200 content items | 8 editors, about 1,000 content items |
| SEO | 1 user, about 500 tracked keywords | 2 users, about 1,500 keywords | 4 users, about 5,000 keywords |
| Email marketing | 1,000 subscribers | 10,000 subscribers | 50,000 subscribers |
| Docs and projects | 1 seat | 5 seats | 15 seats |
For each size we pick the plan that business would actually need, checking the plan’s limits, and add every line that makes up the bill: the subscription, extra seats, contact tiers, required add-ons, usage fees and platform transaction fees. Each bill shows its lines, so you can see what is counted and change the assumptions in your head.
A few rules keep the totals honest:
- Month-to-month billing is used wherever it is offered, because that is what you pay before you commit. When a plan is sold only yearly, we use its monthly equivalent and say so.
- Card processing fees on selling platforms are listed separately and not added to the total, because every platform charges them at a similar rate.
- Each bill has a confidence level. High means every line came straight from the vendor’s page. Medium means a line depends on a usage estimate. Low means a needed figure could not be verified.
How we keep prices current
Every Monday an automated check re-opens each vendor’s pricing page and confirms that the prices we show are still there. When a price, plan or limit changes, we update the record and log the change, with the date and the source, on the price changes page. Every pricing page shows the date its prices were last verified.
Some vendors block automated visits. When that happens the check says so, and the page keeps the date of our last manual check rather than pretending to be newer.
How we score tools
Each tool gets one score out of five, based on six factors. The factors are about what matters to a small business paying the bill, not about feature counts.
| Factor | The question we ask |
|---|---|
| Real cost | At a typical small-business size, how far is the real monthly bill from the headline price once required add-ons and usage are counted? |
| Pricing honesty | Can a buyer work out the real cost from the vendor's own website, without booking a sales call? |
| Small-team fit | Can a small team set it up and run it without an agency, a developer or a week of onboarding? |
| Does the whole job | How completely does it cover its part of the stack, and how many other tools does it replace? |
| Freedom to leave | Can you export your data, cancel without a long contract, and move to something else without heavy cost? |
| Track record | Years in market, how often and how sharply prices have changed, and what is publicly known about reliability and support. |
The procedure:
- Two separate AI passes. Each pass is a separate run of an AI model (Anthropic’s Claude) over the same saved evidence. It scores all six factors from one to five, gives a one-sentence reason for each, and gives an overall score. Neither pass sees the other’s scores or reasons.
- Calibration. Both passes compare against every tool already scored, so a 4 means the same thing across the site. If the result would rank a tool above a rival in the same job that costs less and scores at least as well on most factors, we lower it, and that tool’s score panel says by how much and why.
- If a tool pays us a commission, or runs an affiliate program we have joined or may join, we publish the lower of the two scores. Otherwise we publish the average of the two, rounded down, unless calibration (step 2) lowered it.
- The overall score is a judgment, not a formula. It is not the average of the six factors, because the factors do not matter equally for every job.
- Thin evidence means no score. If we cannot find enough to judge a tool fairly, the page says “Not scored” rather than showing a made-up number.
Every tool page shows the six factor scores and the reason behind each one, under “Why we score it”.
How we choose the stack picks
Each stack page recommends one tool for each job a business of that kind needs done, and names alternatives with the situation in which each one is the better choice. We choose on fit and total cost for that business, not on commission. When a tool that pays us is recommended, the highest-scored option is shown beside it. When the right answer is a tool that does not pay us, or no tool at all, we say that.
What we do not claim
- We have not used every tool as a paying customer. Our prices come from the vendors’ published pages, and our scores come from that pricing evidence plus what is publicly documented. Where a page reflects our own hands-on use, it says so and gives the date.
- The monthly bills are estimates for the stated assumptions. Your bill depends on your usage and on the offer the vendor gives you when you sign up.
- We do not give financial, tax or legal advice.
How we use AI
AI does much of the work here, and we would rather say so plainly. AI (Anthropic’s Claude) drafts the pages from the saved copies of each vendor’s own pages, turns those copies into the price records, and runs the two separate scoring passes described above, which do not see each other. Every price in a record is checked by script against the saved copy it cites, and anything that cannot be traced to a dated copy is not published as a number. Rohail Nisar owns CinchStack and is responsible for its method and for everything published on it, including the recommendations and scores.
Corrections
If you find a wrong or outdated price, send it to us. We check it against the vendor’s page and, if we were wrong, correct it within the week and log the correction on the changes page. Vendors can use the same form; we treat their messages exactly like anyone else’s and correct only what their public pricing page supports.
How we are funded is explained on How we earn.
First published 26 September 2026 · Last updated