ROI is the word that opens and closes every Salesforce Enterprise conversation. The vendor's deck brings the numbers: seller productivity up 38%, average cycle reduced 24%, retention improved 27%. Real numbers, from real research — but drawn from a base of customers already using Salesforce, meaning a base with significant selection bias.
The question isn't whether Salesforce delivers ROI. It does, when the decision is right. The question is that the vendor's model projects revenue from above and license cost from below, then calls it TCO. It isn't TCO. It's the cost of entry. The cost that determines whether the investment actually works is what shows up after go-live.
What the vendor's deck doesn't include
A typical Salesforce commercial proposal lists three lines of cost: annual license, implementation, and initial training. It rarely includes the three lines that determine whether the numbers add up over 36 months.
Ongoing operations cost. A certified internal admin or a managed services partner. For mid-market companies (30–150 licenses), this cost ranges between R$ 8,000 and R$ 25,000 per month — depending on release volume, integrations that need maintenance, and Flows that need revision. This number doesn't appear in the proposal. It shows up on the second invoice from the partner.
Organizational adaptation cost. New process, behavior change, and legacy data rework. No Salesforce project skips all three. The figure we see in diagnostics: 1.5x the implementation cost, spread over 12–18 months. It's the cost of making Salesforce tell the truth — because at go-live it reflects what the company migrated, not what the company actually does.
Opportunity cost of the alternative. Hubspot or Pipedrive, properly configured, solve 60–80% of a mid-market operation's needs at 15–25% of the total cost. This opportunity cost should appear in the comparison. It rarely does. The vendor has a clear interest in not showing it — and the client rarely has an independent benchmark.
The cost of Salesforce isn't what's in the proposal. It's what shows up in operations two years after go-live.
The four-variable matrix
There's a structured way to decide whether ROI closes — without relying on the vendor's model. Four variables determine the outcome:
Average annual ticket per customer. Salesforce starts making financial sense when the average customer generates R$ 150k+ in annual revenue. Below that, the ratio between platform cost and generated margin rarely closes within 36 months. Low-ticket SMBs, as detailed in the analysis of when NOT to use Salesforce, become a bad equation by design — regardless of how many features the platform offers.
Process complexity. Number of sales stages, number of decision-makers at the customer, average cycle length in days, distinct pricing rules per product. Salesforce excels at high complexity — but carries operating costs proportional to the complexity you brought in. Simple process on a complex platform generates expense without equivalent return.
Team operational maturity. Certified admin (or hireable), defined release process, capacity to translate business demand into CRM requirements. Without that muscle, Salesforce isn't a platform — it's technical debt with a nice interface. Ongoing operational costs rise when the team lacks this structure, because every new customization generates rework the following quarter.
Acceptable payback horizon. Salesforce ROI closes — but rarely before 24 months of stabilized operations. A company that needs return in 12 months will experience frustration before ROI. The vendor's model typically projects returns starting at month 6, which is the go-live month, not the month when operations are actually running with clean data and real adoption.
How to use the matrix before signing
The exercise is straightforward: fill in the four variables with real numbers, not deck projections.
Calculate total cost over 36 months: license + implementation + ongoing operations + organizational adaptation. Conservative rule of thumb: the real cost of the first cycle equals 2.5–3x the annual license cost. If the implementation partner gives a lower number, ask for a detailed breakdown.
Calculate the alternative's cost over the same period: Hubspot Pro, Pipedrive Advanced, or the equivalent for your size. Include configuration and necessary customization. The difference between the two totals is the premium you're paying for Salesforce.
Calculate the differential gain that only Salesforce delivers in this scenario: automation the alternative doesn't cover, native integration that saves engineering time, unified data the alternative doesn't have. Be honest about what's a real differentiator versus what's a feature the team will never use.
Check whether the differential gain justifies the premium over the 36-month horizon — with a 20% buffer for project unknowns. If it closes comfortably: robust decision. If it barely closes: the project needs rigorous scope control. If it doesn't close: the process probably isn't complex enough to justify the platform.
As process mapping before Salesforce makes clear, the right time to run this exercise is before the RFP, not after. Once the vendor is at the table with a ready proposal, confirmation bias makes the numbers always work — because every uncertain figure is estimated on the favorable side.
Five questions your license rep won't bring to the meeting
Beyond the matrix, five questions that change the account:
What's the monthly ongoing operations cost with a certified admin, post-go-live? If the answer is "it depends," ask for a range based on comparable cases at your size. Don't close a contract without that number.
What's the cost of migrating legacy data — and who pays if the data is dirty? Dirty data is the rule, not the exception. A proposal that doesn't mention this cost is hiding the project's biggest risk.
What's the support model after go-live? Implementation partner and managed services partner are rarely the same contract. The Salesforce partner model has tiers and obligations the customer only understands when they need urgent support — and have already paid for the surprise.
What's the adoption metric that signals success at 90 days? If the answer is "user login," the project is going badly. Real adoption is registered process, data useful for decisions, and management using the pipeline — not sellers logged in.
What happens if the project is delayed by 3 months? License contracts typically start running at signature, not at go-live. Three months of delay means three months of license paid before delivery. Who absorbs that cost?
Honest ROI yields more than optimistic ROI
A company entering the project with calibrated expectations, a defined process, and a team with operational muscle generally extracts real ROI. Not the presentation ROI, but measurable ROI: seller productivity, revenue predictability, reliable data for decisions.
A company entering with the vendor's model and discovering the real cost two years later won't leave — because migrating is too expensive. They'll externalize the blame onto the product, cut investment in operations, and harvest the worst of both worlds: expensive license, poorly sustained operations, unreliable data.
The difference between a Salesforce project that pays off and one that bleeds rarely comes down to the product. It comes down to the quality of the diagnosis before signing.
The matrix doesn't guarantee ROI. It guarantees that the decision was made with the right numbers — and when the return is lower than expected, at least it isn't a surprise.