Choosing a Value Metric That Scales with the Customer

The distinction between how you price and how much you charge is easy to blur. The table below separates the two layers and shows the trade-off each metric family carries.

The businesses that get this right usually layer more than one: a base subscription for access, a primary metric for scale, and optional add-ons that absorb variation Robin Landy.

Per-user pricing is still the default in B2B software because it is intuitive to buyers and produces predictable expansion as teams grow, but it also encourages seat sharing and stops working when value is not tied to people Stripe. That last clause is the whole decision.

Run each candidate through these four checks and write down the answer for each. A metric that fails one test is not necessarily fatal, but you need to know which one you are trading away.

Value alignment. Does the metric rise when the customer gets more of what they bought? The library analogy is useful here: two libraries can serve the same readers, one charging per book borrowed and the other per rental period, and the choice of unit decides which one earns more from heavy readers Chargebee. If your metric is storage, and nobody buys you for storage, your revenue will drift away from your value story Sixteen Ventures.

Predictability. Can the buyer forecast next month's bill within a reasonable band? Pure pay-per-use shifts cost risk to the customer and can produce bill anxiety; a hybrid with included usage plus overage keeps a floor for you and a ceiling for them Lago.

Measurability. Can you count the unit from data you already emit, without a new instrumentation project or a monthly argument with the customer's finance team? This is where outcome metrics live or die, because success criteria must be agreed and auditable in the contract The SaaS CFO.

Scalability. Does the metric stay sane when the customer's use grows tenfold? More on this below, because it is the test most teams skip.

For the price level itself, survey methods can anchor the number once the metric is chosen. Van Westendorp's Price Sensitivity Meter asks four questions about too cheap, bargain, expensive and too expensive, then plots the intersections to find an acceptable range and an indifference point Wikipedia. The technique works with small samples, which makes it practical for in-depth interviews Conjointly. One caution: asking directly what someone would pay tends to produce lowball answers, which is why the indirect four-question format exists Forbes.

Take a project management tool priced at $20 per seat per month. A 40-person customer pays $800 per month. The metric passes the alignment test: more collaborating people, more value. It passes measurability and predictability easily.

Now the customer reorganizes. Ten people become heavy daily users, thirty log in occasionally. Procurement proposes paying for ten seats and sharing logins. Your revenue halves while delivered value does not. This is the shelfware problem in reverse: the buyer is paying for a headcount that no longer reflects use Sixteen Ventures.

The fix is rarely to abandon seats. It is to add a second dimension, such as a project or workspace count, so the metric tracks both collaboration and the size of the portfolio. Slack solved a version of this by charging for active users rather than provisioned seats, removing the friction of paying for people who never log in Paddle.

Run the numbers on the fix before you ship it. If the 40-person account drops to 10 paid seats at $20, you collect $200 and lose $600 a month. A workspace metric at, say, $15 per active workspace with 40 workspaces recovers $600 even if seats fall, and it keeps rising as the customer adds projects. The arithmetic is simple; the discipline of doing it before you announce is what separates a pricing change from a pricing mistake.

These three product types share a property: the machine, not the person, generates the value, so seats are the wrong unit.

APIs and infrastructure. Charge per measured unit, usually requests, compute or storage, and consider a committed floor. A prepaid or credit-based structure gives the buyer a ceiling to budget against while keeping your revenue tied to consumption. The risk is volatility in both directions, so most infrastructure businesses pair the usage metric with a minimum commitment.

Marketplaces. When your product sits inside the money flow, a percentage of the transaction is the most aligned metric available, because it only grows when the customer earns. Percentage pricing requires visibility into the money, which is why it suits payments and creator platforms and rarely fits software that cannot see its customers' revenue Lago.

AI agents. This is where the outcome unit becomes practical. Intercom's Fin agent charges per outcome, defined as a resolution or a completed procedure handoff, billed once per conversation even if several questions were answered Schematic. HubSpot's customer agent moved to a per-resolved-conversation and per-recommended-lead structure, charging when the agent completes the task rather than when it produces output The SaaS CFO.

The decision rule for AI products is a single question: does the agent finish the job end to end, and can both sides verify it from the same data? If yes, an outcome unit aligns you with the buyer's KPI. If no, you have usage pricing wearing an outcome label, and the invoice disputes will find you. Build in failure forgiveness, meaning you do not charge for outcomes that did not happen, as a conversion tool rather than a billing edge case The SaaS CFO.

Before you commit, model what happens when your best customer gets ten times bigger. The exercise is mechanical and takes an afternoon.

1. Pick the customer's current metric value. Say 500,000 API calls a month or 40 seats.
2. Multiply by ten.
3. Apply your price per unit with the volume discounts you plan to publish.
4. Ask three questions: Is the resulting invoice still a plausible fraction of the value the customer receives? Can the buyer explain the number to their finance team without calling you? Does your gross margin hold after the cost of serving that volume?

If the invoice becomes absurd, the metric is wrong, not the price. If the buyer cannot forecast it, add included usage or a committed tier. If your margin collapses because per-unit delivery cost does not fall with scale, the metric is hiding a cost problem.

Here is the check in code form, which you can run against real account data:

def stress_test(units_today, price_per_unit, volume_discounts, serve_cost_per_unit):
    units_10x = units_today * 10
    rate = price_per_unit
    for threshold, discounted_rate in volume_discounts:
        if units_10x >= threshold:
            rate = discounted_rate
    revenue = units_10x * rate
    cost = units_10x * serve_cost_per_unit
    return {
        "units": units_10x,
        "revenue": round(revenue, 2),
        "gross_margin_pct": round((revenue - cost) / revenue * 100, 1),
        "price_vs_today": round(revenue / (units_today * price_per_unit), 1),
    }

# 500k calls today at $0.02, dropping to $0.01 above 2M
print(stress_test(500_000, 0.02, [(2_000_000, 0.01)], 0.002))

The output tells you whether a tenfold customer pays ten times, or something else. If it pays three times while consuming ten times the resource, you have built a discount cliff that punishes your best accounts.

Two patterns usually survive the test. First, tiered discounts that step down gently rather than in a cliff. Second, a hybrid: a platform fee that covers fixed costs plus a usage component that scales. Hybrid structures let you keep predictable baseline revenue while pricing scales naturally with the customer Stripe.

The single metric trap. If one number determines the whole bill, an experienced procurement team has one target to attack. Multiple levers, such as tiers, add-ons and support fees, keep the negotiation balanced Robin Landy.

The commodity unit. Storage, message counts that do not matter to the buyer, or a feature flag that nobody tracks will not carry a value story. If customers do not care about the number, pricing on it invites a discount conversation Sixteen Ventures.

The unmeasurable outcome. If your definition of a resolution takes a paragraph of caveats to explain, it will be renegotiated at every enterprise deal. Keep the product definition tight and negotiate a separate enterprise definition on the order form The SaaS CFO.

Mistaking variance for value. A metric that swings wildly month to month is not tracking value, it is transferring budget risk. Watch retention and expansion alongside the metric; neither seat nor usage pricing guarantees strong net revenue retention on its own, because each ties expansion to a different driver Lago.

If you are building a new product and need the metering, billing and packaging to match the metric you choose, a build partner can take that off your plate: describe the app or plan you need and a specialist picks it up with you.