Managing large opportunities against the product roadmap
Most product organisations above a certain size run into the same situation. Sales brings in an opportunity that is large, strategically interesting, and conditional on building something that is not on the roadmap. Product either declines it or accepts it and absorbs the delivery cost quietly. Some months later there is general agreement that the call was wrong, and no clear record of who made it or on what basis.
It is tempting to treat this as a discipline problem, on the assumption that it would go away if product held the line or sales stopped making promises. The tension is structural. Sales is measured on revenue that can be closed now, product on capability that pays off later. Both functions are doing what they are asked to do, and the friction between them usually indicates where the business is stretched.
What can be changed is how the decision gets made. Organisations tend to lose money here through the absence of a process rather than through poor judgement. The commitment is made in a setting where its cost is not visible, and product finds out afterwards. The four mechanisms below make that cost visible at the point of decision.
One thing is worth saying first, because it is often lost when a company tightens up its process. Exceptions are fine. Some of the better product decisions a company makes are the ones that break its own plan, and a business that never adjusts its roadmap for a customer is probably not listening closely enough to its market. What follows is a way of making sure that when you say yes, you understand what you are taking on and make room for it properly. An exception taken with full sight of the impact, with the roadmap adjusted to absorb it, is a decision. The same exception taken because the number was large and the meeting was enthusiastic tends to surface as a problem a few months later.
1. A deal desk with real authority
A deal desk is a standing forum where non-standard opportunities are reviewed before anything is promised. It usually includes sales leadership, product, engineering and finance, with legal and delivery joining once deals reach a certain size.
Two things determine whether it works.
The first is the trigger. A desk that reviews everything becomes a bottleneck, and sales will find ways around it. Set explicit thresholds: deal size above an agreed figure, any commitment to a delivery date, any request for functionality that is not on the roadmap, any non-standard commercial term. Everything else follows the normal sales motion. In practice a small proportion of deals need the desk, and those deals carry a disproportionate share of the risk.
The second is authority. A desk that only advises will be overruled the first time a significant number is at stake. It needs the standing to say that a commitment cannot be made, or can be made only with an agreed trade-off attached. That authority has to be granted by the chief executive, stated openly, and then honoured when honouring it is inconvenient. Otherwise it is a meeting rather than a gate.
Speed matters too. A weekly cadence with an out-of-cycle route for urgent deals is usually sufficient. Sales will work with a gate that takes two days, and reasonably will not work with one that takes three weeks.
2. Capacity reserved rather than borrowed
A common failure here is arithmetic. A roadmap planned to a hundred per cent of capacity has no room for opportunistic work, so opportunistic work displaces committed work without anyone deciding that it should. The platform migration is not cancelled. It slips a couple of times and becomes next year's problem.
Reserve a share of delivery capacity explicitly for commercially driven work. The right figure depends on the business. An enterprise business with a small number of large customers will need considerably more than a product-led business with thousands of small accounts. What matters is that the figure is agreed in advance, visible, and finite.
The finiteness is the useful part. Once the reserve is full, the next opportunity has to compete with something already in it, and the conversation becomes concrete. Sales is then choosing between two revenue opportunities rather than debating priorities with product in the abstract, which is usually a shorter discussion.
Track how the reserve gets used. If it consistently runs well under capacity it is too large and delivery is being wasted. If it consistently overflows, either it is too small or the product is missing something the market keeps asking for, which is a strategy question rather than a capacity one.
3. Decision criteria beyond deal size
Contract value tends to dominate the discussion and tells you relatively little on its own. A structured assessment should cover at least four dimensions.
Strategic fit. Does this move the product towards where you have already decided to go, or sideways? Sideways is sometimes the right answer, and it should be a considered purchase rather than an accident.
Reusability. Will other customers want this, or is it specific to one account? A reasonable test is whether you would be comfortable putting it on the public roadmap. If not, it is bespoke work and should be treated and priced accordingly.
True cost to serve. Build cost is the visible part. Add the maintenance tail, the support burden, the additional testing surface, and the constraint it places on future architecture. Bespoke commitments have a tendency to become the reason a platform change is difficult several years later.
Opportunity cost. What does not get built. Name the specific items, since a general acknowledgement that there is a trade-off is not the same as having weighed one.
Score these, keep the scoring simple enough that people will actually use it, and record the outcome. The record matters more than the precision. After a year you can look back and see which large deals paid for themselves and which are still being carried. Most organisations that do this find their intuitions were wrong in at least one direction.
4. Communication that keeps everyone on the same side
The process work will not hold if the conversation stays adversarial. A few habits help.
Replace a flat refusal with a trade-off. 'We can do that, and it moves the reporting release to Q3. Is the deal worth that?' This puts the decision with the person who owns the revenue consequence, and it keeps product out of the role of permanent objector.
Give sales something to work with when the answer is no. A clear explanation of why the product does not do the thing, what the customer could do instead, and where the roadmap is heading, is more useful to a salesperson than a refusal on its own. Losing a deal well protects the relationship.
Involve product earlier. Many poor commitments are made because product was not present when the customer described the problem. Having a product person on two or three large opportunities a quarter is inexpensive and prevents a good deal of rework.
Close the loop once commitments are made. When something enters the roadmap because of a specific deal, record that in the roadmap. Teams work better on items whose purpose they understand, and the visibility makes it easier to notice when the volume of commercially driven work is getting too high.
Making an exception properly
Assume you will override the process several times a year, and that it will often be the right thing to do. What separates an exception that works from one that causes problems is largely what happens immediately after the decision.
State what it displaces. Name the specific work that moves and by how much. 'We will absorb it' is how a quarter gets lost. If nothing can be named, the plan was probably already unrealistic and the deal is about to make that apparent.
Update the roadmap the same day. The commitment goes in as a dated item with the deal attached, and whatever it displaces moves in the same edit. A roadmap still showing the old dates a fortnight after a major commitment is one that teams cannot plan against.
Price the exception where you can. Bespoke work has a cost, and the customer requesting it is usually well placed to cover some of it. A modest premium or a longer contract term changes the arrangement from a favour into a transaction.
Record the reasoning. A short note covering what was promised, what it displaced, who approved it and what you expected in return. It takes ten minutes and is the only thing that allows you to learn from the decision later. It also gives cover to the person who made the call, which matters if you want people to keep making difficult ones.
Set a review point. Bespoke commitments should carry a date at which you decide whether to generalise the feature, continue maintaining it as a one-off, or retire it. Without that date, anything built for a single customer is maintained indefinitely by default.
Handled this way, an exception costs a known amount and buys a known thing. The process exists to make that possible, and an organisation that never uses the override has probably built a gate that is too restrictive.
Two failure modes
Organisations tend to oscillate between two problems, and both are worth watching for.
The first is capture, where the roadmap becomes a queue of customer-specific commitments. Revenue holds up for a period and then the product loses its coherence. It can be spotted reasonably early: the proportion of work traceable to a single named account climbs steadily, and nobody can describe clearly what the product will be good at in two years.
The second is rigidity, where the process becomes a means of avoiding difficult decisions. Every large opportunity is declined on strategic grounds, the roadmap stays intact, and the company grows slowly while a competitor takes the market by being more useful to real customers. Sometimes a deal genuinely should reshape the plan. The gate exists to make that a decision rather than an accident, and it should let the good ones through.
The aim of all this is straightforward. Say yes to the unplanned opportunities that deserve it, and do so knowing what it costs, having agreed it is worth it, having made real space in the roadmap, and being able to explain afterwards who decided and why. That is a lower bar than getting every call right, and it is the one most organisations are currently missing.
