Methodology

How we score

Every verdict combines what each option costs over 3 years with the risks the price doesn’t show. Stack Scan then applies those verdicts to what you actually pay.

1. Prices

We read each vendor’s pricing page by hand and record the monthly price on annual billing (and on monthly billing when shown). Every page shows the date we checked, October 26, 2026, and links the source. 39 of 54 tools were confirmed on the vendor’s own page. Where a vendor showed regional prices or hid them, the page says so and we use figures that recent reviews agree on.

2. The cost of buying

Plan price × seats × 36 months, plus usage fees where they apply (for example Intercom’s charge per AI resolution).

3. The cost of building

Build hours × your hourly rate + upkeep hours per month × 36 × your hourly rate + running costs × 36 + one month of an AI app builder

  • Your time defaults to $30/hour. It is never free, even when nobody pays you for it.
  • Build hours are our estimate for one person using Lovable or Claude Code, including testing, for a version scoped to typical small-team use, not a full clone.
  • Upkeep covers bugs, API changes, dependency updates and “it broke” messages.
  • Running costs cover hosting, database and services like email or AI calls.
  • AI app builder: one month at $25, the list price of Lovable Pro.

4. The verdict

  • BUILD: building costs at most 75% of buying over 3 years, the risks are low or medium, and the core is simple data in and out.
  • SWITCH: building doesn’t pay, but a near drop-in alternative is free or at least 30% cheaper for the same use.
  • HYBRID: build the part that drives the bill (heavy automations, AI answers) and keep paying for the hard part.
  • BUY: the subscription is cheaper, or a high risk (payments, legal validity, deliverability, security, real-time video, infrastructure) outweighs the savings.

Every verdict page also says when the verdict would flip, because the right answer depends on how you use the tool.

5. How Stack Scan builds your plan

  • Finding subscriptions. We match each statement line against the descriptors 54 researched tools and 100+ other vendors use on card statements. Charges we don’t recognise count as subscriptions only if they repeat on a steady rhythm (weekly, monthly, quarterly or yearly) at a steady amount, and they start unticked for you to check.
  • Cut: tools you say are rarely used. Saving = what you pay a year.
  • Consolidate: two paid tools that do the same job, or a tool your Google Workspace or Microsoft 365 plan already covers (like Zoom next to Google Meet). We keep the one your team uses most. Saving = the other tool’s yearly cost.
  • Switch: a researched alternative that costs at least 20% less for your seats. Saving = the difference.
  • Downgrade: when your charge exactly matches a higher plan and seat count, the difference to the next plan down. If the charge could match more than one plan, we assume the typical one and suggest nothing.
  • Build: for BUILD and HYBRID tools, the yearly difference between buying and building at your spend and hourly rate.

Each tool gets one main suggestion (the biggest saving that fits its verdict), so savings are never counted twice. Other options are listed underneath.

6. Instant AI verdicts

For tools we haven’t researched by hand, an AI research assistant reads current pricing on the web and applies the same rubric. Those answers are clearly marked as AI drafts, list their sources, and are cached so everyone sees the same result. The most-asked tools get a full, hand-checked verdict.

Our honesty rules

  • No company can pay for a verdict or change one.
  • We’re not affiliated with any tool we review.
  • We never show a saving we can’t explain with a number you can check.
  • If a price is wrong, we fix it and update the date. Tell us on X.