Estates grow faster than they shrink and nothing retires on its own, so the renewal is the only easy removal moment until you build one that runs all year. Three knowledge checks along the way, and 3 clips from a senior licensing analyst.
This is a taught session, not a talking head. The instructor works through analyst grade slides, and three times the video stops on a question with four options on screen. Pause, commit to an answer, and the next slide explains which option is right and why each of the others is wrong. 3 times in the session the frame splits and a senior licensing analyst gives the view from inside real ServiceNow negotiations, and the instructor picks the clip apart when the slides return.
The full narration of this session, section by section, for reading and reference. Guest analyst clips are marked.
Welcome back, session thirty six, module eight opens, and we turn to the question that has been sitting underneath this entire course. Why does the waste keep coming back. Because here is the pattern we have found in every single module. New apps, new fulfillers, new credit packs, new modules, all arriving between renewals, and not one of them retires on its own. Growth is automatic and removal requires a negotiation. Which means the renewal is the only moment when removal is procedurally easy, because outside that window the path to cutting volume runs through a sales representative who is paid to keep it on the order form. That is not a criticism of anybody, it is a description of the machinery. Today is about building your own removal moment so you are not waiting three years for theirs. Three checks, homework, let's go.
Five objectives. First, name the asymmetry, which is that growth needs no approval and removal needs a negotiation. Second, build four gates, on fulfiller creation, table creation, module adoption, and consumption capability, four decisions that currently happen without anybody pricing them. Third, run demand management, so that a request for access gets answered with the cheapest correct option rather than the fastest one. Fourth, keep a standing baseline, which is session ten's document maintained quarterly rather than rebuilt in a panic twelve weeks before every renewal. And fifth, attach an action to every threshold, because a number with nobody's name on it is a dashboard, which is exactly the lesson session twenty three learned the hard way on the assist meter.
Four framings. Twenty to thirty five percent, the shelfware most estates carry, and every session in this course found a version of it arriving one reasonable decision at a time. One way, which is how the estate moves without governance, growth needing no approval and removal needing a negotiation. Sixty five and thirty five percent, the thresholds worth acting on, where consumption below sixty five percent of entitlement is a rightsizing candidate and below thirty five percent is a hard cut. And every three years, which is how often most organisations look at any of this properly, because the renewal is the only forcing function anybody has ever built. Now the note, because I want to be clear about what governance is for. The purpose is not tidiness. It is to make your next renewal start from a clean estate rather than from an archaeology project run in twelve weeks by people who all have other jobs.
Guest analyst clip. I have done the twelve week version of this many times and I want to describe what it actually feels like, because it is the thing governance is designed to prevent and most people have never seen the alternative. Twelve weeks out, somebody realises the renewal is coming. A person gets assigned, usually somebody capable who already has a full job. They start pulling exports. The exports do not match the order form, so they spend a week working out why. They find a module nobody can explain and start emailing around trying to find an owner, and the owner left in March. They find two hundred accounts with no activity and cannot tell whether those people left the company or simply never used it. And all of this is genuine detective work, done under time pressure, competing with everything else, and the output arrives about ten days before the negotiation with several caveats attached to it. And then somebody senior asks whether the numbers are solid, and the honest answer is mostly. Now compare that to the same organisation with a maintained baseline, where the answer to every one of those questions was written down at the time by the person who knew it. Same data, same estate, completely different position. The difference is not effort. The total effort is lower. It is just spread out, and done by people who knew the answers.
The total effort is actually lower, it is just spread out and done by the people who knew the answers at the time. So, the four gates, and where the decisions are being made today.
Four decisions that currently have no price attached. Creating a fulfiller, made today by an administrator solving an access request, and the gate adds a classification question, does this person work records or do they approve and request. Creating a custom table, made by a developer building something entirely reasonable, and the gate adds a count against the allowance, a named owner, and a review date. Adopting a module, made by a business unit with a genuine need, and the gate adds the unit it prices on, the population, and whether the platform already does this. Enabling a consumption capability, made by a team switching on something that arrived bundled, and the gate adds an entry in the burn report before it appears in the burn report. Now read the note carefully because it is the whole design principle. None of these gates should be an approval queue. Every one of them is a question asked at the moment of the decision, by somebody who now knows the question exists. That distinction is what stops governance from being routed around within a quarter.
Demand management, and the principle is to answer the request with the cheapest correct option. Five points. The request is almost never for a licence, it is for an outcome, see this queue, update that record, run this report, and several routes deliver that outcome at very different prices. Route occasional contributors, which is session nineteen's finding, because requester and portal flows serve a twice monthly update without a fulfiller seat and deflected work costs nothing at all. Check whether the platform already does it, which is session twenty one's build versus buy question asked at intake rather than at design review, where it is very much cheaper to answer. Make the cheap option the easy one, because if the fastest path to access is a fulfiller role then that is what people will request, and they will be entirely right to request it. And publish the price list internally, not to discourage requests, but so the people making them can actually see what they are choosing between, which most of them have never been shown.
Knowledge check one. Should the fulfiller creation gate be an approval queue? A, yes, every fulfiller should require sign off from licensing. B, no, it should be a classification question asked at the point of the request, because approval queues get routed around. C, yes, but only above a monthly threshold. D, no gate is needed, the renewal audit catches it. Pause here, and ask what happens to a control that slows down access.
The answer is B, a question rather than a queue. A control that slows down legitimate access gets worked around within a quarter, and usually by somebody competent solving a real problem, which means you have not prevented anything, you have just made the data worse and annoyed people. So the gate has to be a question, does this person work records or do they approve and request. Answer D is the status quo that produces twenty to thirty five percent shelfware, because the renewal audit does catch it, three years and one compounding uplift too late. And session twenty two made exactly this argument about automation. Count it, own it, review it, and do not gate the thing the platform exists to do, because gating that is how you end up with a governance function everybody resents and works around.
The standing baseline, maintained rather than rebuilt. Session ten built the entitlement baseline as a renewal artifact, and governance turns it into a standing document, which changes both what it costs to produce and what you can use it for. Four cards. What it holds, which is what you own, what you use, and the gap, with every number traced to a source and a date, exactly as session ten specified. Quarterly rather than triennial, because one update a quarter is about an hour of work, whereas a rebuild twelve weeks before a renewal is a project competing with everybody's day job. It answers the audit too, because the same document that produces your negotiating position is the one that answers a compliance question, and session thirty eight will come back to that properly. And it survives people, because the baseline is the institutional memory that the account team already has and most customers simply do not, which is the single largest asymmetry in this relationship and the one this document closes.
Knowledge check two. A module sits at forty percent consumption against entitlement. What does the threshold discipline say? A, nothing, forty percent is normal adoption. B, it is a rightsizing candidate, because anything below sixty five percent qualifies, and below thirty five percent is a hard cut. C, cut it immediately. D, wait for it to fall below twenty percent. Pause here, and ask what number turns an observation into an action.
The answer is B, a rightsizing candidate. Published thresholds are what convert a number somebody noticed into an action somebody takes, and sixty five percent marks a candidate for review while thirty five percent marks a cut you should be prepared to defend. Answer C skips the review, and that review matters because it establishes whether the gap reflects a stalled rollout that is about to accelerate or a genuine over purchase, and those two need completely different answers. And answer A is how a forty percent module becomes a forty percent module for another three years, repriced upward at every renewal in between, which is session twenty five's arithmetic quietly doing its work while everybody agrees that forty percent seems normal.
Every number names an action and an owner, five of them. Fulfillers below the activity threshold, reviewed quarterly against the ninety day rule, owned by the platform owner, with the list going to the managers who can actually release the seats. Custom tables against the allowance, counted every quarter, owned by whoever holds the register from session thirteen, and expressed as a percentage of allowance rather than as a raw count because the raw count means nothing to anybody. Module consumption below sixty five percent, where a named business owner explains it or it becomes a renewal candidate, and the explanation is the point rather than the number. Assist burn projected over pool, which is session twenty three's monthly page with the attribution review triggered before capacity gets bought rather than after. And an annual retirement round, funded, scheduled, and owned, because switching things off is work that nobody is rewarded for and therefore work that nobody does spontaneously, ever, in any organisation I have seen.
Guest analyst clip. The funded retirement round is the recommendation I have to argue for most often, and the objection is always the same, which is that switching things off should not need its own budget line. And in a rational world that is true. But look at what you are actually asking somebody to do. You are asking a team to spend two weeks not building anything, confirming that a thing is genuinely unused, chasing people who left, and then turning something off, which carries a small but real risk of breaking something and no upside whatsoever if it goes well. Nobody gets promoted for that. Nobody puts it in a performance review. If it goes wrong they own an incident, and if it goes right absolutely nothing happens. So the question is not whether people are lazy, they are not. The question is why anybody would volunteer for that, and the honest answer is that they would not, and that is a completely rational response to the incentives you have built. Which is why I say fund it. Give it a name, a slot in the year, an owner, and a small budget, and treat the seats and tables it recovers as its return. The organisations that do this recover real money every year. The ones that do not recover it once every three years in a hurry, and pay uplift on it in between.
Nobody volunteers for work with no upside and a small chance of an incident, and that is a rational response to your incentives rather than a failure of character. So fund it, name it, and schedule it. Now, who holds what.
Three groups, one number, one meeting, and this slide is the answer to a gap that has produced findings in every module of this course. The platform owner holds the instance, the usage data, and the monthly numbers, and can see absolutely everything and usually has no idea what the contract says. Procurement holds the contract, the clause set, and the dates, knows exactly what is negotiable, and usually cannot see what is actually happening. The business owners hold the value question, which no report can answer, and which decides whether a low number represents waste or a deliberately targeted deployment. And one meeting a month, which session twenty three called the cheapest control in enterprise software, and which is still the thing most estates do not have on anybody's calendar. Session thirty seven turns to the events that break all of this, mergers, divestitures, global entities, and restructuring an agreement around a company that is no longer the company that signed it.
Knowledge check three. Your governance forum reports usage monthly and nothing ever changes as a result. What is missing? A, more detailed reporting. B, thresholds with named actions and owners, because a number nobody must act on is a dashboard. C, more frequent meetings. D, executive attendance. Pause here. You have met this exact failure once already, back in session twenty three.
The answer is B, thresholds with named actions and owners. This is session twenty three's finding generalised, that reporting without a trigger produces awareness, and awareness does not remove a seat or retire a table. Answer A makes the report longer and therefore less likely to be read, which is the opposite of the fix and is also the single most common response organisations actually choose. Answer D helps only if that executive is the named owner of a specific threshold, in which case it is really answer B wearing a suit. And here is the test I would apply to any governance forum, including your own. Name the last thing that changed because of it. If nothing comes to mind, you have reporting rather than governance, and they are not the same activity.
Four habits that change the shape of the estate. One, gates as questions, on fulfiller, table, module, and capability, asked at the moment of the decision and never as an approval queue that slows down real work. Two, a maintained baseline, quarterly rather than triennial, so the renewal starts from a document rather than from a project. Three, thresholds with owners, sixty five and thirty five percent on consumption, ninety days on activity, projection over pool on the meter, each one with an actual name attached to it. And four, a funded retirement round, annual, scheduled, and owned, because nothing retires on its own and nobody is rewarded for switching things off. Do those four things and the renewal stops being the only moment when removal is easy, which changes both what you pay and how much of your year the renewal consumes. That second part matters more than people expect.
Guest analyst clip. I want to describe the difference governance makes, because it is not primarily about the money and I think that undersells it. Take two organisations with identical estates and identical contracts. The first has none of this. Their renewal consumes about four months of senior attention, most of it spent establishing basic facts about their own systems, and the negotiation itself is a compressed argument at the end conducted with numbers nobody fully trusts. The second has the four habits. Their renewal takes maybe six weeks, because the facts were already established, and the conversation is about price and terms rather than about what they own. Now, the second one does pay less, and that is real. But what strikes me more is what they get back, which is several months a year of capable people not doing archaeology. Those people work on something else. And when I ask organisations to fund governance, that is the argument I would lead with rather than the savings, because the savings are contested and speculative in advance, whereas anybody who has lived through the twelve week scramble knows exactly what it cost them and would quite like not to do it again. Sell it as time, and the money arrives anyway.
Sell governance as time rather than as savings, because the savings are contested in advance and everybody remembers the scramble. Session thirty seven takes the events that break every structure we have just built, mergers, divestitures, global entities, and restructuring the agreement when the company changes shape.
Three sentences. Estates grow faster than they shrink because new apps, fulfillers, credit packs, and modules all arrive between renewals and none of them retire on their own, which is why most estates carry twenty to thirty five percent shelfware. The four gates are fulfiller creation, table creation, module adoption, and consumption capability, and each one has to be a question asked at the point of decision rather than an approval queue, because queues get routed around by competent people solving real problems. And a standing baseline maintained quarterly, thresholds that each name an action and an owner, and one monthly meeting between the platform owner, procurement, and the business is what stops the renewal being the only moment when removal is easy.
Homework, about an hour, five items. Test the four gates, and for each of fulfiller, table, module, and capability, ask who decides today and whether anybody asks a licensing question at that moment, and I would expect four noes and that is normal. Date your baseline, and if the answer to when it was last updated is at the last renewal, then what you have is a triennial document rather than a standing one. Write down three thresholds, each with a number, an action, and a name, because three real thresholds beat a comprehensive framework that nobody uses. Find the monthly meeting, and establish whether anybody holds the usage number and the contract in the same room, because if not then setting that meeting up is the highest return action available to you this month. And name the last thing that changed because of governance in the last year, and if nothing comes to mind then you have reporting rather than governance.
Five guides. The license rightsizing playbook has the consumption thresholds, the exports that feed your baseline, and the explanation of why the renewal is currently the only easy removal moment. Fulfiller versus requester licensing explained sits underneath the first and most valuable gate, with the activity thresholds that make the classification defensible. App Engine licensing explained covers the custom table allowance behind the second gate and the boundary rules a table creation question should be testing. The 2026 cost creep report shows what ungoverned growth looks like priced out, including the shelfware that every uplift reprices upward. And avoiding true up surprises is the other reason governance pays, because the same baseline that produces your negotiating position is what answers a compliance claim. Next time, licensing events, mergers, divestitures, and restructuring. See you there.