Methodology

ToolMargin compares the known cost and verified capabilities of agency software. The calculations are deterministic: the same inputs always give the same results. No affiliate or commercial data is an input. Pricing checked against official sources Sep 18–24, 2026.

Four stages, one set of rules

Each stage answers a different question, but all four use the same catalogue, the same evidence and the same calculations.

• Start plans a new agency's first stack and prices three states separately: launch day, ready for the first client, and your target size. Later subscriptions are never added to today's cost, and a free plan is shown as $0.
• Build keeps what you already pay for and adds the lowest-cost changes that close your Required gaps. When nothing is missing, the answer is that no purchase is needed.
• Grow follows your stack from today's team and client count to your stated target. It separates required changes, optional cheaper alternatives and automatic price changes, and says what triggers each one.
• Optimize compares your current stack with the lowest-cost and most consolidated valid stacks, lower plans and removable products. It says so when your stack is already the lowest-cost valid option.

Verified products and plans

The catalogue holds 19 products and 37 plans. Each plan has its own price, billing options, limits and capabilities, taken from official vendor sources: pricing pages, pricing documents, help-center articles and vendor publications. Every source is recorded with its address and review date, and the results link to the evidence behind each plan.

• A plan's capabilities are never assumed from another plan of the same product.
• Free plans are permanent free tiers with documented limits, never trials.
• Temporary promotions, legacy prices and quote-only plans are kept separate and never ranked.
• Where official sources disagree, the conflict is recorded and the current, consistent statement is used; nothing is averaged or guessed.

Required and Preferred jobs

You describe what your software must do as business jobs, such as a sales pipeline, proposals or client reporting, and mark each one:

• Required: the stack must cover it with a verified plan. Required jobs are the only ones that can justify a purchase.
• Preferred: nice to have. A Preferred job never triggers a purchase on its own; results show whether the stack happens to cover it.
• Not needed: ignored.

Each job has a depth (Basic, Full-featured or Specialist). Only capability claims backed by verified evidence count, and partial or ambiguous claims are treated as not covered. For example, guest access to a project tool is not treated as a client login portal. A stack's coverage of a job is its single strongest product: two basic tools do not add up to a full-featured one.

Known and unknown costs

• Known recurring cost is the monthly equivalent of verified subscription charges. On annual billing it is the prepaid year divided by 12.
• First-year known cost is twelve months of recurring cost plus required one-time fees such as onboarding.
• A price that is not verified for your billing period or usage is shown as not priced. It is never treated as free and never counted as zero.
• A plan you are not sure of is kept as unknown: its capabilities are not counted and its price is not guessed. Tools outside the catalogue count only as you describe them and are never priced.
• A cost difference is shown only when every component on both sides is priced. Otherwise the comparison is marked incomplete, with the reasons, and no savings figure is calculated.

Monthly and annual billing

A price is used only for the billing period it was verified for. Monthly rates are never derived from annual discounts, or the reverse. If a plan has no verified price for the billing period you choose, it is shown as not priced for that period, and you can switch billing to compare.

Plan limits and usage

Plans have limits such as users, clients, contacts, dashboards, automation volume or storage. Your stated team size, client count and usage are checked against each limit. Where you have not stated a quantity, a documented default derived from your team and client count is used, and results say which figures are assumed. A plan that cannot serve your usage is not recommended for it. Where a limit is finite and meaningful, results show how much of it you use. Usage beyond the ranges a vendor publishes is not modeled.

How stacks are chosen

The search is exact: it gives the same answer as checking every combination of verified plans, with no sampling or shortcut that could miss a cheaper valid stack. It keeps only stacks that cover all your Required jobs, fit your usage and follow the product rules, such as using at most one primary CRM.

It then removes dominated stacks: those that another stack matches or beats on every job you asked for, on known cost and on number of products. A stack that is stronger on something you asked for is never removed for costing more.

• Lowest cost: the lowest first-year known cost. Ties go to lower recurring cost, then fewer products, then more Preferred jobs covered.
• Most consolidated: the fewest products, then the lowest known cost. It is shown only when it differs from the lowest-cost stack.
• Specialist options: stacks that are stronger on the jobs you asked for, each with its extra cost. No single "best" stack is picked for you.

What is not modeled

The catalogue is a defined set of products, not the whole market. A product that is not in the catalogue is not judged, and results routinely say so rather than filling the gap.

Some software categories are not modeled at all: agency website, website hosting, landing pages. ToolMargin does not currently model these categories, so you can mark them as needed but no product is recommended for them, and results show them as not modeled. An explicit "not modeled" answer is deliberate: it is preferred over a recommendation the evidence cannot support.

Exclusions

Prices exclude taxes, currency conversion, negotiated discounts, migration and implementation effort, cancellation terms and optional add-ons. Usage-based charges are not included unless explicitly priced. Examples are messaging, phone and AI consumption, extra automation credits, and overage beyond the priced tier. Each result lists the exclusions that apply to it.

Price verification dates

Prices are not live quotes. Each plan's price was checked against official sources on its own review date, shown with the evidence in the results; across the catalogue, review dates run from Sep 18, 2026 to Sep 24, 2026. Vendors change prices and plans, so check the current price with the vendor before you buy. If you find a price that has changed, the contact page explains how to tell us.

Recommendation independence

Rankings use only verified prices, verified capabilities and your inputs. There are no sponsored placements or paid rankings.

The site may earn affiliate commissions from some vendors in future. Affiliate status, tracking links and program rules are kept in a separate commercial module that the pricing data, the calculations and the results model cannot import; an automated test checks the whole import graph and fails if that changes. That module can only change where a vendor link points and whether a disclosure is shown beside it. It cannot add, remove or reorder a recommendation, and results routinely include products with no affiliate relationship. Links to official pricing and evidence sources are never affiliate links. See the affiliate disclosure.

The classic optimizer

The classic optimizer is the original comparison of ten products, one package each, kept so that earlier shared links keep working. For a review of your current stack, Optimize uses the full catalogue. The classic optimizer follows the same pricing, capability and independence rules described above, with its prices checked Sep 18, 2026.

The classic optimizer evaluates combinations of those ten products against capability requirements rather than business jobs. It also recalculates results at 5, 10, 20 and 50 clients, scaling dashboards and data sources with your per-client assumptions while team size and other usage stay fixed, so that view is not a growth forecast; Grow is.

Known evidence gaps in the classic optimizer

These are open as of Sep 18, 2026. Where one applies, the product is shown as not priced; nothing is estimated.

• Teamwork: Accelerate monthly rate is not published in reviewed official sources; monthly comparisons exclude Teamwork. Annual comparisons use the verified annual rate. Optimize profitability pricing is custom.
• monday.com: monthly seat rates are verified (Basic 12 / Standard 14 / Pro 24). The 3,000-credit pack price (30/month) was observed only in the annual view. Under monthly billing the pack is excluded from known cost and shown as unpriced; its monthly price needs verification.
• Make: Pro (21) and Teams (38) are verified monthly only; their annual prices are not stated in a readable official source. Credit tiers above 10,000/month are unpriced (requires-verification).
• DashThis: capacities and rates above 25 dashboards / 100 connected sources were inconsistent across official-page reads; no higher tier is modeled or extrapolated.
• HubSpot: annual-prepay additional-contact rates above the included 2,000 marketing contacts are not published; annual comparisons above 2,000 contacts are unavailable. Credits and messaging overages are excluded.
• Zapier: task tiers above 100,000/month, pay-per-task overage beyond the selected tier and additional operator/team access are not modeled.
• HighLevel: the web Client Portal and client document upload are documented, but no official source ties them to the Unlimited plan; those claims stay partial and cannot satisfy requirements.
• Client-financial capability depth (Teamwork Accelerate strong = budgets/retainers/time-based invoicing only) remains an editorial rubric.

Choose your stage