.png)
Frameworks, core principles and top case studies for SaaS pricing, learnt and refined over 28+ years of SaaS-monetization experience.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Pricing agentic B2B products – such as AI copilots for legal, tax, or finance – is a critical strategic task for SaaS product managers. Get it right, and you unlock faster growth and stronger customer trust; get it wrong, and you leave money on the table or stifle adoption. This post presents a five-step pricing transformation framework (from Price to Scale by Ajit Ghuman and Jan Pasternak) to price these innovative products effectively. We’ll cover (1) Segmentation, (2) Positioning & Packaging, (3) Pricing Metric, (4) Rate Setting, and (5) Operationalization – with actionable guidance and real examples (like Harvey.ai and other AI copilots) at each step. The goal is a clear, practical approach that you can implement immediately to align price with value and scale your SaaS business.
Why a new approach? Agentic AI products often don’t fit neatly into traditional SaaS pricing models. Generative AI is shifting pricing from simple seat licenses to more complex usage- or value-based models. For example, companies are now charging premium add-on fees (often 30–110% above base prices) for AI features that dramatically boost productivity. Pricing must capture this new value while remaining fair and predictable for customers. The five-step framework below ensures you methodically design pricing that reflects your product’s unique value, cost drivers, and market expectations. Let’s dive into each step.
Step 1: Segmentation – Identify Your Key Customer Segments
Every successful pricing strategy starts with knowing who you’re pricing for. Different customer segments perceive value differently and have varying willingness to pay. Misaligned segmentation is where most pricing problems begin. If you try a one-size-fits-all price, you risk alienating some customers and undercharging others. Instead, define clear customer segments (by company size, industry, use case, etc.) and tailor your approach to each.
Start by analyzing your current and target customers. Group them by meaningful criteria, such as firmographics (e.g. SMB vs Enterprise, or industry vertical) or behavior (e.g. volume of usage, need for compliance). For an agentic AI product, one segment might be large enterprises that demand robust features and support, and another might be smaller firms or teams that need a lighter, affordable version. For example, in the legal AI space, Harvey.ai has focused on large law firms and corporate legal departments, while a competitor Paxton AI targets smaller law practices. Paxton offers transparent pricing ($159/user/month with a free trial) to appeal to cost-sensitive small firms, whereas Harvey doesn’t even publicize pricing and uses long-term contracts for enterprise deals. This stark contrast in go-to-market approach reflects deliberate segmentation – each company chose a different target segment and aligned pricing strategy accordingly.
The dangers of poor segmentation are illustrated by Narvar (a post-purchase SaaS platform). Narvar initially used a rigid one-size tiered model that lumped small and large clients together. The result was chaos: small businesses were forced toward an enterprise package 50–60% more expensive than what they needed. Sales reps had to heavily discount to close deals, hurting revenue, and customers ended up paying for “shelfware” features they never used. As Narvar learned, without segment-specific pricing, you’re “actively pushing customers away”. They fixed this by redesigning pricing with flexible, incremental options so smaller clients could buy add-ons without jumping to a huge bundle. The takeaway: define your Ideal Customer Profiles and segments clearly, and ensure your pricing structure can adapt to their distinct needs and budgets. Before you move on to packaging or numbers, double-check that you know your key customer groups and what each truly values.
Step 2: Positioning & Packaging – Tailor Your Offers to Each Segment
With your segments defined, the next step is crafting packaging – the specific bundles, tiers, or editions of your product – that aligns with each segment’s needs. Packaging is how you position the value for different customers. In practice, this means deciding what features or service levels to include in each tier, how many tiers to offer, and how to differentiate them. The goal is to make each package feel “just right” for its intended customer segment, avoiding both feature overload and feature gaps.
A common pitfall here is blindly using a “Good-Better-Best” tier structure without considering if it fits your market. Many SaaS companies default to three tiers, but if those tiers aren’t designed around real customer segments, it can backfire. Gainsight, a customer success platform, learned this the hard way. They had Good-Better-Best packages, but the middle tier didn’t align with mid-market customers’ needs. Many midsize clients found the mid package either too limited or too complex, and they either dropped to the lowest tier or demanded the highest – often with heavy discounts. The top “Best” tier became bloated with features clients didn’t use (shelfware), frustrating customers and blocking upsells. Gainsight’s response was to scrap the rigid tiers in favor of a flexible, modular packaging. Customers could then mix and match features they needed without overspending. This overhaul aligned pricing much more closely with actual customer usage and value, increasing satisfaction and average deal sizes.
The lesson: Design your packages based on customer value, not just a marketing template. If one segment only needs core features, create a basic offering for them. For power-users or larger enterprises, offer higher tiers or add-on modules with advanced capabilities and extra service. Ensure that each tier provides a compelling value proposition to its segment, and avoid “shelfware” – don’t pack a tier with features that that segment won’t use. Every feature in a package should drive value for the target customer; if it doesn’t, leave it out or put it in a different tier.
Real-world packaging strategies in agentic AI products reflect this principle. For instance, Thomson Reuters’ CoCounsel legal AI offers a basic plan versus a full-access plan, so law firms can choose the level that fits their workflow (the full plan is priced higher at $400/month, presumably offering broader capabilities). Another example: LexisNexis’s Lexis AI breaks its AI tools into separately priced modules – legal research, document drafting, contract analysis, etc., each with its own fee. A small firm that only needs AI legal research can pay $99 for that, while a larger firm might add the $250/month drafting tool. This packaging by use-case lets customers pay only for what they truly need, which increases perceived fairness and value.
When positioning your packages, also consider how you communicate the value of each tier. The naming and marketing for each package should resonate with its target segment. For example, you might label a tier for startups “Growth Plan” highlighting affordability and essentials, versus an “Enterprise Plan” that promises advanced analytics and dedicated support for corporate needs. In the agentic software realm, think of it as offering, say, “AI Assistant Basic – essential copilots for small teams” versus “AI Assistant Pro – advanced automation suite for enterprises”. The core product might be the same AI under the hood, but the packaging and messaging are tuned to different audiences.
In summary, create packages that map to your customer segments and clearly articulate the added value at each level. This may mean deviating from standard tier models – and that’s okay. The right packaging makes it easy for each customer segment to choose your product and see it as a perfect fit. If you find one of your tiers is rarely chosen or often requires custom tweaks, that’s a signal to refine your packaging. Done well, packaging becomes a strategic differentiator and upsell engine, rather than a straightjacket for your sales.
Step 3: Pricing Metric – Choose a Value-Based Charging Model
Arguably the heart of pricing design is selecting your pricing metric – the unit by which you charge customers. This is how you will bill them (e.g. per user, per transaction, per output, etc.), and it should align tightly with how customers derive value from your product. In the words of pricing expert Ajit Ghuman, your pricing metric is a direct reflection of how your product delivers value. Choose correctly, and customers feel your pricing is fair and flexible; choose poorly, and you create friction, confusion, or unexpected bills.
Traditional SaaS products often used simple metrics like per-seat (per user) licensing. But for agentic AI products, many companies are shifting away from pure seat-based pricing towards usage-based or outcome-based models to better capture the value delivered. Generative AI’s variable costs and variable outputs make this necessary. For example, OpenAI charges by the API call or token usage – you pay exactly for what you use, and heavy users pay more. Some newer AI SaaS firms charge per outcome delivered; an AI customer support bot might charge per ticket resolved or per conversation handled, aligning price with successful automation. When Microsoft introduced its AI Copilot for Office, it chose an add-on seat fee (about $30/user), but also reportedly considered usage elements (one report equated it to ~$4/hour of AI assistance). The key is that Microsoft tried to anchor Copilot’s price to the value of productivity gains, roughly 60–70% of the base Office suite price. This hybrid approach (flat per user plus potential usage caps) ties pricing to expected outcome – time saved – rather than just selling another software seat.
To determine the right metric, ask: What usage of our product correlates with customer success and value? Ideally, as customers get more value, that metric grows, and so does their bill (and they’re happy to pay it). For an AI document analysis tool, it might be the number of documents processed. For an AI coding assistant, it could be the number of code completions or projects. Mixpanel’s pricing journey provides a classic lesson here. Mixpanel originally charged based on “events” (every user action tracked) in its analytics platform. That made sense for small customers, but as enterprise clients generated millions of events, costs exploded unpredictably and customers didn’t perceive each extra event as added value. The result was busted budgets and customer churn. Mixpanel realized the disconnect and boldly switched to a new metric: MTUs – monthly tracked users. Instead of counting raw events, they charged by the number of unique users being analyzed, which aligned much better with the business value (understanding user behavior, not the raw clicks). This change not only stabilized bills, it also made sense to customers: more active users means more value from the analytics. Mixpanel even introduced tiered packages for different use cases (a package for engineers, one for marketers, etc.), each with its own relevant metrics and limits – further aligning price to how each persona sees value.
When choosing a metric for your own product, also consider your cost structure. Agentic AI solutions often have significant variable costs (API calls, GPU compute, etc.). If every use of your product costs you (in cloud resources), a pure flat fee could be dangerous to your margins. Usage-based pricing can protect you here – as usage increases, revenue scales to cover costs. A recent industry survey noted that nearly all AI startups face non-zero marginal costs and thus “lean into usage-based pricing” to avoid getting sunk by their own success. Even if you opt for a simple subscription, you might impose fair use limits or tiered usage bundles to hedge against outlier usage. The CloudZero team advises: whichever model you choose, know your unit costs to the penny so you don’t inadvertently underprice high-usage customers. If, for instance, each AI query costs you $0.05 of compute, ensure your pricing metric and rates account for that with healthy margin, or consider a separate usage charge for heavy consumption.
One more consideration: Outcome-based metrics (pay per result achieved) are compelling but tricky to implement. They directly tie price to value, which customers love in theory – e.g. “only pay $X when the AI actually finds a valid tax deduction for you.” However, you must very clearly define the outcome and prove it, otherwise disputes will arise (“the AI’s answer wasn’t actually useful, so why should we pay?”). Outcome pricing works best when success is objective and measurable. If you can nail that, it can differentiate your model (as in the earlier Zendesk example: charging per support ticket solved by the AI). But if not, a usage proxy or a hybrid model might serve better.
Actionable tip: If you’re unsure which metric will resonate, consider launching with a hybrid model or offering a choice. Some companies combine a base subscription (to cover fixed costs) with variable usage fees for the AI components. This provides revenue predictability while still scaling with value delivered. Hybrid pricing is common for AI add-ons now – e.g. Salesforce’s Einstein AI charges a platform fee plus credits for AI queries. Just remember that more complex models demand more explanation to customers and more sophisticated billing (we’ll address that in operationalization). Simplicity has its own value, so always weigh the benefit of a “perfect” metric against the ease-of-use of a simpler one. For example, GitHub Copilot chose a flat $19/user/month – no usage tracking needed – likely to minimize friction for developers adopting it. In contrast, an API-based service might have to be purely pay-as-you-go. Choose what aligns with both your value delivery and the practicality of billing and customer understanding.
In summary, pick a pricing metric that aligns price with value and scales reasonably with customer growth. Test it with customers if possible (“would you prefer if we charged by A or by B?”). And be prepared to adjust – if you see signs of customer pushback or unforeseen usage patterns, it might indicate the metric needs refinement (better to tweak the metric than to lose the customer).
Step 4: Rate Setting – Determine the Right Price Points
Once your structure is set – segments, packages, and metric – you face the big question: How much to charge? Rate setting is about translating the value you deliver into a specific dollar amount (or currency) that customers will pay. The challenge is to set prices high enough to capture value and sustain your business, but not so high that you deter customers or invite easy competition. The best practice here is value-based pricing: use your product’s value to the customer as the north star for pricing, rather than just costs or competitors. In practical terms, this means understanding your customer’s willingness to pay and the ROI your product provides.
Begin by researching how your target customers perceive the value. Talk to some customers (or prospects) to gauge what outcomes or features matter most and what they currently spend (or would spend) to achieve those outcomes. For instance, if your AI copilot saves a tax analyst 10 hours of work a month, and that analyst’s time is worth $100/hour, that’s $1,000 of value created monthly. Pricing anywhere below that ensures the customer is getting a positive return. Microsoft 365 Copilot’s $30/user/month price was reportedly set in light of huge productivity gains – up to 50% faster tasks in early tests. Microsoft essentially priced Copilot at a fraction of a knowledge worker’s salary cost, making it a no-brainer ROI for many firms. Similarly, if a legal AI tool can replace or augment a junior lawyer (costing say $8k/month fully burdened), a price of $1k or $2k per month per user for that AI can be justified by the labor cost savings. In fact, rumors around Harvey.ai’s pricing suggest it is positioned as a premium product – some sources indicate $1,000+ per user per month for Harvey. That high price reflects the mission-critical value it provides to top law firms (and perhaps its early mover advantage). In contrast, competing products aimed at smaller firms (like Paxton AI’s $159/user) price much lower because those customers won’t pay enterprise-level fees. Both have presumably done research to find those sweet spots for their respective segments.
It’s wise to also look at alternatives and competitors. How are potential customers solving the problem now, and what do they pay for that? Your pricing can be anchored against the status quo or competing solutions. If your AI finance tool replaces a consultant engagement that costs $50k, pricing your software at $10k/year is a compelling deal. If competitors sell per-seat licenses at $100/month, you need a story if you want to charge $300 (perhaps you deliver 3x the value, which you’ll need to prove). That said, do not just copy competitors’ pricing blindly – ensure it fits your positioning. As a reminder from Step 2, copying pricing models (like Good-Better-Best tiers) without context was a mistake Gainsight and others had to correct. Instead, use competitive pricing as one input among many.
Testing and feedback are your friends here. In high-growth SaaS, pricing isn’t static – you can iterate. Consider techniques like a beta pricing program or A/B tests if feasible: offer different customers slightly different packages or prices and learn from their choices. You can also use formal methods like willingness-to-pay surveys or conjoint analysis to pinpoint optimal price ranges. For example, you might survey prospects with a Van Westendorp price sensitivity questionnaire to find at what price they’d consider your product “too expensive” versus “a good deal.” These methods can reveal a price window that customers find acceptable. If data shows your target SMB customers won’t pay more than, say, $49/month, you either price at $49 or need to change which segment you’re targeting.
Don’t overlook the psychological aspect of pricing. There’s a reason many SaaS companies price at $99 instead of $100 – it just feels smaller. Rounding, tier naming, and including/excluding certain features can all influence perceived value. Ajit Ghuman’s work references cognitive pricing – framing prices to match customer psychology. In practice, this could mean highlighting the cost of not having your solution. For instance: “On average, our AI saves 40 hours of work, roughly $2,000 of labor, for a cost of $500 – a 4x ROI for you.” This helps customers justify the price internally.
Finally, consider a tiered pricing point strategy: have entry-level pricing for lower segments and premium pricing for top tiers. Many SaaS products do this (e.g., a free or cheap starter tier to capture wide users, and a pricey enterprise tier for deep-pocket clients). In the agentic software arena, LawGeex (an AI contract review tool) exemplifies this with plans from $399/month for small businesses up to $2,799/month for large firms. The large firm tier is higher because those customers have more at stake and a greater ability to pay – and they presumably get more value (and features) from the product. Your rate setting should mirror the value segmentation: higher segments can have higher price points. Just be sure the value scales, too, not just the price.
In summary, set your prices based on value and validate them with data. If possible, start a bit high – it’s easier to offer discounts or adjust down than to realize you underpriced and try to raise rates later. Many startups err on the side of underpricing initially out of fear; don’t sell yourself short if you have a strong value proposition. As one pricing truism says: if nobody complains that your product is too expensive, you probably priced it too low. Aim for a price that some will push back on, but your core target will accept for the value they get. And remember, pricing is not permanent – revisit and refine your rates as your product’s value and market positioning evolve. The best companies treat pricing as an ongoing experiment, not a one-time decision.
Step 5: Operationalization – Implement and Manage Your Pricing
Congratulations – by this stage you’ve designed a solid pricing model on paper. Now, the real work begins: making it a reality in your business. Operationalization is the often-overlooked step that can make or break your pricing strategy. It’s about integrating the new pricing into all facets of your operations – billing systems, sales processes, customer communication, and metrics tracking – to ensure the strategy actually delivers results. A brilliant pricing scheme means little if your sales team can’t sell it or your billing system can’t charge for it properly.
Update your billing and tooling: Modern pricing (especially usage-based or hybrid models) can be complex. You must ensure your systems can handle it. If you’re moving from a simple subscription to, say, a base fee + usage model, can your billing platform meter usage and generate accurate invoices? Do you need to implement usage tracking in the product and connect that to your billing app or Stripe? Many companies underestimate this effort. In fact, hybrid pricing models “require sophisticated systems to track and manage both fixed and variable charges effectively.” If you rely on manual invoicing or an outdated system, you risk errors, revenue leakage, or customer frustration. Invest in a proper subscription billing solution or adapt your CRM/ERP to handle the new model. For usage-heavy products, you might need to build an internal usage metering service. This can take time – some startups spend months to get billing in order – but it’s fundamental infrastructure. Don’t go live with a pricing model you can’t bill correctly. For example, when introducing a per-document pricing component, ensure every document processed is counted and billed; if offering tiered overage rates, make sure those kick in at the right thresholds automatically.
Sales enablement and policies: Your sales and customer success teams need to deeply understand the new pricing. Train them on the value rationale behind it, so they can confidently explain it to customers. Each rep should know which package a given prospect should consider, how to handle common objections (e.g. “Why usage-based? I prefer flat fee”), and the flexibility (or lack thereof) in the pricing. Update your sales playbooks and scripts with the new packaging structure and pricing metric. If you created an enterprise tier with outcome-based pricing, for instance, the sales team should be able to articulate the ROI and how outcomes are measured.
Critically, establish clear discounting and approval guidelines. Nothing can torpedo a well-crafted pricing strategy faster than ad-hoc deep discounts given by overeager salespeople. Decide on a discount policy: e.g. perhaps sales can offer up to 10% discount for annual deals, anything beyond requires VP approval. Many SaaS companies use “guide rails” for sales with defined floors and approval levels. For instance, a sales rep might have authority to give a small discount, a sales manager a bit more, and anything larger needs CFO sign-off. This prevents one-off exceptions that undermine your pricing integrity. It’s also a good idea to document the reason for any discount (volume, strategic logo, etc.). If you arm your team with a clear value story, the need for discounts should diminish, but be prepared with a process regardless. In the context of agentic B2B products, be wary of heavy discounts especially early on – it could signal that customers don’t see the value. Stick to your value-based guns as much as possible.
Customer communication: If you are rolling out new pricing to existing customers (or changing their plans), manage this proactively. Communicate changes clearly, highlighting benefits. For example, if you’re shifting them from unlimited use to usage-based, explain how this could save them money if they use less, or how the new model ensures future sustainability of the service. It’s often wise to grandfather existing customers on their old pricing for some period, or give plenty of notice, to maintain goodwill. Consider offering migration incentives (e.g. extra credits, or locking their current rate for a year) to ease the transition. The introduction of AI features and pricing can actually be an upsell opportunity if framed right: “We’ve added powerful new AI capabilities. Existing customers can try them free for this month, and thereafter add them to your plan.” Always keep the tone customer-centric – emphasize how the new pricing or packages help the customer succeed (more flexibility, more choices, etc.) rather than just how it helps your revenue.
Monitor key metrics and feedback: Once in market, closely watch the data. Important indicators include: conversion rate (do fewer prospects convert now due to pricing complexity or level?), average revenue per customer (ARPA), churn rate, and product usage patterns under the new pricing. If you introduced a usage-based model, are customers engaging as expected or holding back usage due to cost? If you see anomalies – e.g. a spike in churn or a popular tier that everyone is gravitating to – investigate quickly. Pricing is not “set and forget.” The best companies treat it as a living system. Establish a regular cadence (say quarterly) to review pricing performance. Look at support tickets or sales feedback about pricing: Are customers frequently confused about the metric? Are sales reps requesting a certain discount repeatedly, indicating a possible mis-price for a segment? Use that insight to adjust if needed. For example, if your mid-tier package is rarely selling, it might be either unnecessary (so simplify and remove it) or it lacks enough value relative to price (so either add value or drop price).
Finally, assign ownership for pricing. As you scale, it’s wise to have a product manager or pricing specialist responsible for ongoing pricing strategy and operations. They can coordinate between product, sales, and finance to tweak packages or prices as the market evolves. This person/team should also stay on top of competitive moves (if a new rival undercuts your price or changes their model, you may need to respond or at least be ready to justify why you’re superior). Pricing operationalization also means aligning incentives internally: for example, if you moved to a subscription model, ensure the sales compensation plan rewards quality of sale (customer retention, upsells) and not just initial contract value. You might adjust sales commissions to discourage excessive discounting or overselling features the customer won’t use. All parts of the organization should be rowing in the same direction with the new pricing.
In short, make the new pricing part of your company’s DNA. Technically, get your systems and processes in place to handle it. Culturally, get your team onboard and fluent in it. And practically, keep measuring and refining it. Many great pricing strategies fail not because they were wrong in concept, but because they weren’t executed and managed properly. Don’t let that be you – treat operationalization with the same rigor as you did the design phase, and you’ll reap the rewards of your well-crafted pricing in the form of smoother sales cycles, predictable revenue, and happy customers who understand exactly what they’re paying for.
Conclusion: Summary Checklist for Pricing Agentic SaaS Products
Pricing is a journey, not a one-time task. By following this five-step framework, you ensure that your pricing strategy is grounded in market realities and executed with discipline. To recap, here’s a quick checklist of practical steps when pricing your agentic B2B SaaS product:
- Define Your Segments: Identify distinct customer groups by size, industry, or needs. Ensure your pricing strategy targets your Ideal Customer Profiles (ICP) – no one-size-fits-all. (Do your small clients and enterprise clients have different options? They should.)
- Tailor Your Packaging: Design tiers or packages that match each segment’s needs and willingness to pay. Avoid bloated packages with unused features (“shelfware”). Each segment should naturally gravitate to a package built for them. (Check that each package’s features align with its target customers.)
- Pick the Right Pricing Metric: Choose a billing metric that scales with customer value – whether per user, per API call, per transaction, or outcome. Make sure it’s understandable and tracks with both the value delivered and your internal costs. (Ask: will customers feel this pricing grows in fairness as they use more?)
- Set Value-Based Price Points: Research your customers’ willingness to pay and the ROI your product provides. Set initial price levels that capture your value but remain competitive. Don’t be afraid to charge for real value – but be ready to adjust based on market feedback. (Would your customer pay this price confidently after seeing the value for a month?)
- Operationalize Thoroughly: Implement the pricing in your tools and teams. Update billing systems for any usage tracking. Train sales on the new model and enforce discount guardrails. Communicate clearly with customers, especially if transitioning from an old model. Monitor key metrics (ARR, churn, usage) and iterate as needed. (Do you have the systems to bill and the people to sell this pricing correctly?)
By ticking off each of these steps, you’ll significantly increase your chances of pricing your AI-powered product for scalable growth. Pricing agentic SaaS solutions can be complex, but with a structured approach, you can align price with value in a way that benefits both your business and your customers. Remember, effective pricing is as much an art (understanding customer psychology and value perception) as a science (using data and process). Use the framework, stay attuned to your customers, and be willing to learn and adjust. With that mindset, your pricing can become a powerful lever for your product’s success rather than a hurdle. Good luck – and happy pricing!