How the ROI Conversation is Evolving in B2B Sales: The ROI Paradox

Sellers and marketers often mis-use ROI as a door-opener. Big dollar signs or multiples are the subject or headline in emails and drip campaigns, often supplemented with case studies proving the ROI you achieved elsewhere. The intent is to establish that the seller is worth the buyer's time.

What the buyer actually experiences is a claim with no anchor. There is no problem being solved, no goal being served, no explanation of how any of it would play out in their business. Lean on it too hard and it becomes actively discrediting — even claims such as "you don't pay us until we get to the outcome" miss the mark and open questions about how long that is going to take and why the guarantee is necessary at all.

Case studies fail in two symmetrical ways. Name a household brand and the buyer's first thought is "well of course they can do that because they're resourced to the hilt." Name a company the buyer doesn't recognize and the conclusion is "yeah but we're not that company." Either way the differentiation is thin — an 8.5x claim against a competitor's 8.0x is not a decision criterion, and both frequently confuse correlation with causation.

There is a genuine paradox here. Tell a buyer nothing about value and there is no reason to take the meeting. Tell them precisely how value will land in their business and the fair response is: how could you possibly know that already? This is much of why pre-sale value engineering exists as a function. But it also means sellers should expect to be accused of marking their own homework, every time.

• Never present ROI without customer context — the problem, the goal, and the specifics of their situation.
• Treat some level of value as table stakes rather than as the differentiator. If the offering weren't valuable, the buyer wouldn't be on the call.
• Choose reference examples for structural similarity, not logo size.

Assume the opportunity has progressed and the need is real. Now a business case has to be built for the CFO and everyone else who needs to have a say. This is where two failures appear.

Maximalist modeling. The calculator gets built on best-case scenarios that oversell the capability. Scope creep. Every additional use case, region, and shiny object gets a number attached to it, and the model grows tentacles until it is a gargantuan thing nobody can defend.

Both buyer and seller are complicit here, and both are being rational. The buyer wants to buy; they believe the way to win CFO approval is to make the number as large as possible. The seller hears more use cases and sees a bigger deal. The result is ROI tunnel vision on both sides of the table — perfecting the number instead of improving the decision.

Meanwhile the CFO is evaluating two entirely different questions: is this achievable and will it stick, and of everything this business could invest in, is this the right one? ROI addresses neither. And of all the buyers, the CFO is the one who is most skeptical — they have approved more purchases than anyone else in the org, and have seen the most failed implementations. Their gut reaction is to immediately discount the ROI by half. Not because they don't trust the math, but because they don't trust their own organization's ability to see it through.

One force is making this materially harder. Procurement's traditional qualification defenses have thinned. A working demo used to require real engineering investment; a proof-of-concept cost time and money. Both functioned as implicit qualification of the vendor. Now a credible-looking solution can be stood up in an afternoon, and the same AI tooling lets buyers pressure-test every assumption in a model, hunt for edge cases, and stress the scenarios. That puts enormous weight on assumptions — a real problem when the thing being modeled is genuinely new and there is no operating history to model from.

• Aim the model at the needle-movers, not at everything. The goal is a realistic number, not a big one.
• Resist scope creep even when the buyer initiates it. Big things are hard to pull off, and a bigger model makes the outcome feel less achievable, not more.
• Own the variables, not just the output. Knowing the trade-offs well enough to spot when something is drifting is as valuable to the buyer as the number itself.
• Don't hand the value conversation off to solutions or value engineering and step back. Bringing in a specialist doesn't remove the seller's role in it.

The business case looks good and the buyer still won't commit. They ask for more validation, another referral or case study, a discount, or a free trial. The seller has been hard-wired to lean into cost of inaction and starts turning the screws on the downsides of not moving forward.

This is the most damaging of the three, because the cost of inaction only works when the underlying objection is a value-based objection. But a buyer does not reach late stage without seeing the value. What is holding them back is the fear that they won't realize it — and, more often than not, the risk they're worried about is their own organization, not the vendor's capability. Telling that buyer they are losing millions a month says plainly that the seller has not understood the problem. Early-stage ROI missteps erode trust. This erodes the deal.

The JOLT research found that time-pressure tactics — including discounts with an expiration date — don't merely fail to accelerate decisions, they make the buyer less likely to decide. The mechanism is straightforward: this buyer's fear of missing out on value has already converted into fear or risk of messing up, and artificial scarcity stacks fear on top of fear. When a decision time is compressed, people feel more scrutiny and more chance of getting it wrong, so they are less likely to say yes.

Discounting also has a second-order problem. Having spent the entire cycle justifying a high, defensible return, moving on price at the last moment suggests you don't fully believe in the value yourself. Put another way: if you could genuinely prove a 10x return, the rational move would be to raise the price. The reason that isn't happening is that the return hasn't been proven in a defensible way. Discounting to buy trust is margin erosion standing in for evidence.

Even a win obtained this way carries meaningful risk and baggage. A buyer who signs under pressure that they didn't feel good about walks in the door on a transactional footing — which is the account your customer success and expansion teams inherit.

If ROI pressure, discounting, and free trials are the wrong tools, these are the better ones. Trials and POCs deserve one note of their own: they are never actually free from the buyer's side — they cost time and internal resource — and if they are made genuinely free, they stop being representative of the real thing.

This sounds obvious and happens far less than it should. It was one of the most important variables in the JOLT model — built from roughly 8,000 variables across 2.5 million recorded sales conversations. Instead, what works is a number that is co-created rather than delivered: the buyer understands the number, understands what it will take, understands the trade-offs required to get there, and finds it practical, believable, and achievable. Clarity into the number matters more than the size of it.

Be explicit about the few levers that actually move the outcome, and about the financial mechanics behind them — where in the business value gets created, what the inputs are, what assumptions have been made, what the current baseline is, and what a realistic improvement looks like. Then name the non-quantifiable outcomes separately rather than forcing them into the model. Risk reduction is the clearest case: proving you prevented a cyber attack that never happened is close to impossible. Model what you can genuinely defend, land on a credible multiple, and present the unquantifiable outcomes alongside it — in many deals those are what the buyer is really buying anyway. The things that are easy to quantify and deliver become the things that financial-minded buyers can use to rationalize or justify the investment, and the things that are harder to quantify become the icing on the cake. You can feel good about investing in something you can't prove, because the things you can prove are what makes it pay for itself.

Once you've told a buyer what they have to bring to the table, the next move is making that feel smaller and safer. Success plans, named CSMs, executive sponsors. Creative use of services — staging onboarding, implementation, and rollout around the constraints the customer actually has. And, where genuine risk remains, one-off exceptions. Those should be the exception, not the rule: do the first two well and you need them far less often.

In the absence of validation the seller provides, buyers will source their own — analyst quadrants they can show a board, LLM tools to justify one vendor over another, reference calls. Reference calls have a known ceiling: the buyer assumes they will only be introduced to customers who had a good experience, and never to one where something went sideways for reasons that may have had nothing to do with the vendor.

This section reflects the approach described by James Davies of Guaranteed Outcomes during the session. It is presented as his account of his firm's methodology, not as a SellingInnovations service or endorsement.

The mindset shift here is to treat vendor delivery as a risk-modeling problem rather than a marketing problem — mapping every implementation a vendor has completed across their customer base and presenting the result as a risk profile. Three layers were described:

• Certification. An in-depth value-engineering review of a vendor's actual performance across their customers — not a survey — producing what a prospective buyer can realistically expect. The useful byproduct: bottom-quartile implementations almost always trace back to specific things the customer did or didn't do, which surfaces customer-side best practices as well as vendor-side ones.
• Third-party adjudicated money-back guarantee. If the vendor does what they committed to, and the customer does what they committed to, and it still fails, the buyer recovers their money plus cleanup costs. The distinguishing feature is the adjudicator: buyers have heard vendor-issued guarantees before and know who marks the homework.
• Underwritten outcomes. Where a vendor can measure the delivered value precisely — the example given was industrial heating and cooling equipment with measurable energy-cost reduction — the value can be written into the contract and treated as an asset.

The reported floor for building an insurable ROI model was roughly twenty delivered implementations, though earlier-stage work is possible using value-engineering techniques without an installed base.

One point worth carrying past the specifics: this only works if both sides' obligations are spelled out, which makes it inherently a multi-threaded, CFO-level conversation rather than a champion-level one. De-risking is an enterprise problem. Understanding what removes risk for your champion is not the same as understanding what removes it for everyone else who has to live with the decision.

SellingInnovations is a B2B sales research, training, and advisory firm co-founded by Matt Dixon, Ted McKenna, Dave Anderson, and Rory Channer. Dixon and McKenna are the authors of The JOLT Effect, drawn from a study of 2.5 million recorded B2B sales conversations; Dixon also co-authored The Challenger Sale and The Challenger Customer.

SellingInnovations was spun off from DCM Insights and focuses exclusively on B2B sales. The two remain sister companies under shared ownership.

Value and de-risking are core to how we work with sales teams — the mindset shift that lets sellers hold these conversations without needing to be financial engineers, and the tactics that make an outcome feel achievable to a buyer. Both sit inside our JOLT Effect and Fearless Buyer work. Related reading in our research library, including Stop Losing Sales to Customer Indecision (Harvard Business Review, June 2022).