Here’s an invented scenario to frame the topic. A fictional mid-size company, unconnected to any real firm, runs Salesforce for 450 sales and service users with a team of 7 people. Last year’s roadmap listed 31 items, and by October only 9 had shipped. The CFO now wants a 12-month plan before approving next year’s budget, and she has asked for costs, expected benefits and what happens if estimates are wrong. The team lead wonders whether to hire roadmap consultants, and what a plan must contain to earn funding. This post lays out what Salesforce documents, what outside sources say, and what a careful buyer should test.

Scope and caveats come first. I read 9 pages. Salesforce supplied 3 of them, made up of 2 Trailhead modules and its Well-Architected overview. Another 4 were a consultancy’s roadmap post, a vendor’s technical debt guide, a 2020 Deloitte article and a 2012 McKinsey and Oxford study of IT projects. The last 2 pages held nothing usable on Salesforce roadmaps, so I set them aside. Consultancies and vendors sell planning help, so their advice favors having a plan. I found no study of Salesforce roadmaps and funding outcomes, and the statistics I cite are older and come from other kinds of projects. Confirm every number with your own finance team.

TL;DR

  • Salesforce documents the business case, not the roadmap. Its Trailhead modules describe a financial story built on assumptions, 4 value levers and phased delivery, and they give no prioritization method, funding process or statistics. The second module defers roadmaps to another unit, and the Well-Architected page I read has no roadmap guidance.
  • Planning advice comes from sellers, and the numbers are old. A consultancy suggests scoring work as value times confidence divided by effort and splitting capacity 60% to 70% planned work. A 2012 McKinsey and Oxford study of large IT projects found an average 45% cost overrun and 56% less value than predicted.
  • A fundable plan survives a stress test. In my invented model, a $648,500 plan with $1.5 million of planned benefit turns into a loss of $280,325 if cost runs 45% over and benefits fall 56%. Gates, tranches and a visible reserve are how leaders limit that exposure.

Why plans stall at the funding meeting

Most roadmap proposals fail for the same reason, and it’s rarely the technology. A plan arrives as a list of requests, each with a champion, and the funding body can’t tell which items matter most, what each costs, or how anyone will know they worked. Leaders then fall back on familiar tools. They ask for a smaller list, a lower number or a delay. Salesforce’s Trailhead material says the business case must make the investment and the projected benefits addressable at the presentation, and it admits that at this stage you’re making assumptions and giving estimates. Leaders can also say yes to a plan in parts far more easily than they can say yes to a plan in full, which is the idea behind the gates described later.

That framing is honest, and it suggests what a funding body wants: assumptions it can challenge, a sequence it can approve in parts, and a way to stop. A 12-month plan that reads like a catalog of wishes gives it none of those. A plan built around 3 to 5 business outcomes, a capacity limit and a few approval gates gives it each of those. The rest of this post builds that kind of plan step by step, using invented numbers where the sources are silent and flagging every gap I found.

What leaders ask, and what the sources offer

The table below lines up the questions a funding body usually asks with what the sources I read provide. In every row the evidence is guidance from a training module, a consultancy or a vendor, and none of it reports measured results for Salesforce roadmaps. Use it as a checklist when you prepare the proposal, and fill in the last column with your own data before the meeting. Add a row for any question your own leaders ask.

Leader’s questionWhat the sources offerSource typeGap to close
What outcomes will this deliver?Pick 3 to 5 outcomes and baseline indicators firstConsultancy postYour own baselines
What does it cost?Break costs down by department; model assumptionsTrailhead modulesA cost model with ranges
What’s the return?4 value levers: revenue, cost, productivity, strategic capabilityTrailhead moduleBenefit estimates with owners
What gets done first?Score value, confidence and effortConsultancy postA scoring session with sponsors
What if it goes wrong?Large projects overrun on averageMcKinsey and Oxford studyA stress test on your plan
Can we stop early?Intermediate checkpoints on long projectsMcKinsey and Oxford studyGates tied to results

What the training material covers

Salesforce’s Trailhead modules are training material, not research, and they describe how to present a case rather than what funding committees decide. The list below summarizes the 3 Salesforce pages I read and what each means for a roadmap proposal. None reports a statistic, and 2 of them are written around example scenarios, so they describe intent rather than proven practice. Use them to shape the pitch, and treat the rest as your own work.

  • Business case as a financial story. The innovation module says the case rests on assumptions and estimates, involves finance, marketing and operations in the model, and should show benefits that far exceed the cost. It also pairs common objections with responses, such as phasing costs when the vision feels overwhelming. Use it for the shape of your pitch, then bring your own numbers, since the module supplies no benchmarks or formulas.
  • The value levers. The module names increasing revenue, reducing cost, improving productivity and increasing strategic capabilities as the drivers to model. That’s a useful filter for a roadmap, because an item that touches none of them is hard to defend. Tag every request with the lever it moves and the person who owns the number, and cut items where nobody will own the number. Review the tags with finance before the meeting so nobody is surprised.
  • Phasing and deliverables. The module recommends starting with a minimum loveable product and preparing 3 deliverables: a capability map, a technical architecture snapshot and an implementation plan with timing and resourcing. It uses a new product pitch as its example, so adapt it to an existing org. The module gives no funding sources or approval process, so you must design those yourself. Decide early which parts a partner would cover, since that choice affects cost, timing and the people involved.
  • Documenting the problem. A second module says a business case should break costs down by department and estimate how much money and time a new process saves. It says the timeline is high-level and doesn’t replace a project plan, and it doesn’t address prioritization. Roadmaps appear only as a link to another unit, so this page gives no direct guidance on building a roadmap. Ask finance for its own template before you draft your version.

Why 90 days and 12 months both matter

Fast Slow Motion, a Salesforce consulting firm, argues in a January 2026 post that annual roadmaps are too abstract and weekly backlogs too reactive, and it calls 90 days the sweet spot. It warns that planning too far out leads to overcommitting. That’s a fair caution, but it conflicts with how many finance teams work. Deloitte’s 2020 article notes that annual budget cycles and fixed allocations can limit agility and slow course corrections, yet the cycle exists, and a team that ignores it won’t get funded.

The practical answer is to plan in 2 layers. Commit in detail to the next 90 days, with named items, owners and measures, and describe the remaining 9 months as themes with ranges for cost and effort. Fast Slow Motion’s own sequence of monthly release themes, which it calls foundation, workflow expansion and adoption, fits that pattern. The funding body approves the year’s envelope, releases money in tranches, and reviews the themes at each gate. Neither the consultancy nor the training material addresses funding approval, so treat this as my synthesis.

Inputs to build before you score anything

Outcomes and baselines. Fast Slow Motion advises choosing 3 to 5 business outcomes and baselining their indicators before any prioritization. This is the most useful step in the post, because it gives the later argument something to measure. Pick outcomes that executives already track, such as lead response time, quote cycle time or case backlog, and capture today’s number. The post gives no benchmark values, so the baseline has to come from your own reports.

Inventory and tagging. The same post suggests listing all work and tagging each item as break and fix, run, change or a larger redesign, then cutting anything not tied to an outcome. In an org with 31 open requests, though the post gives no figure for how much it removes, so count that for yourself. The point is to see how much capacity routine work already consumes before new ideas arrive.

Scoring requests. The consultancy recommends scoring each request on value, confidence and effort from 1 to 5, and ranking by value times confidence divided by effort. Its examples give 10 for a lead-routing project and 1.5 for redesigning all page layouts, and it allows a risk flag that moves compliance or security items ahead. The scale and formula are the firm’s own choices, with no testing cited, so use them as a conversation tool rather than a verdict.

Capacity limits. The post suggests 60% to 70% of capacity for planned roadmap work, 20% to 30% for break and fix and small requests, and 10% to 15% for platform health. Those ranges can add up to anywhere from 90% to 105%, so reconcile them to 100% for your team. The percentages are unsourced, which makes them a starting hypothesis. Measure your own ticket history for a quarter before you fix the split.

An invented scoring table

Everything in this section is invented, and I recomputed it in code. The table uses the consultancy’s formula on 6 requests from the fictional company, with value, confidence and effort each scored from 1 to 5 by an imagined steering group. The hours are invented estimates. The ranking shows how the formula favors cheap, confident items and pushes large, uncertain ones down the list, which is useful for conversation and misleading if treated as precise.

RequestValueConfidenceEffortScoreInvented hoursRunning total
Lead routing and response-time rules4428.0900900
Duplicate cleanup and field retirement3527.57001,600
Quote approval simplification4334.01,4003,000
Service case triage3333.01,3004,300
Forecast dashboard rebuild2442.01,5005,800
Page layout redesign2241.01,2007,000

Capacity comes before the wish list

A roadmap that ignores capacity is a wish list, and finance teams spot it quickly. In the invented company, 6 full-time equivalents with 1,800 productive hours each give 10,800 hours a year. A 65% planned work share is 7,020 hours, a 25% break and fix share is 2,700 hours and a 10% platform health share is 1,080 hours, which totals exactly 100%. All 6 requests above need 7,000 hours, leaving 20 hours of slack, and that’s a plan with no reserve.

That result changes the proposal. Rather than ask for funding on all 6 items, fund the first 4 at 4,300 hours and hold the last 2 as options for a later gate, leaving about 2,700 hours of planned capacity unassigned. The unassigned time is a reserve for estimated errors and surprise work, and leaders can see it. Estimates for software work are often wrong, and the McKinsey and Oxford findings on cost overruns below show how large the error can be on big projects. Show the reserve as a line in the plan, with the hours and the rule for spending it, such as release only on sponsor approval.

Cost, benefit and the stress test

Cost side. Separate internal time from cash. The invented plan uses 4,300 hours at a loaded $95 an hour, which is $408,500, plus $180,000 for a partner and $60,000 for licenses, so the total is $648,500. Internal labor is usually already in the payroll budget, so the incremental cash ask is only $240,000. Leaders will want both figures, because the first shows the full cost of the choice and the second shows the new money required. State which figure you mean every time.

Benefit side. Assume an invented $1.5 million of annual benefit across productivity, revenue and cost avoidance. Every benefit needs an owner who agrees to the number, a baseline and a date for measuring it. Salesforce’s training material tells you to model 4 levers but gives no benchmarks, so the estimates must come from your own data. Avoid counting the same hour twice, such as productivity gains that also appear as revenue, since finance will check for overlap.

Reference-class stress. McKinsey and Oxford’s 2012 study found that large IT projects, defined as those with initial price tags above $15 million, ran 45% over budget on average and delivered 56% less value than predicted. Your plan is probably smaller and different, so the figures don’t apply directly, but they show what finance leaders have seen. Applied here, cost rises to $940,325 and benefit falls to $660,000, a net loss of $280,325. At plan, the net is $851,500. The study also suggests a short diagnostic before committing money, which compares a project with outcomes from similar past work, and you can mimic it by listing your own last 3 projects with planned and actual cost.

Break-even. At the planned cost, the roadmap needs 43% of the planned benefit to break even. If cost runs 45% over, it needs about 63%. Those are the numbers to put in front of the funding body, along with the gates that limit exposure. A plan that says it breaks even at 43% of the benefit and stops spending if early results trail is easier to approve than a plan that promises full benefit. Keep the arithmetic on a single page, with every input labeled as measured, estimated or assumed, so a skeptic can challenge any line without rebuilding the model.

How finance leaders tend to decide

Deloitte’s January 2020 article, written by its global CIO research director, offers the clearest view of how finance and technology leaders handle investment, though its data come mostly from 2018 and 2019. The list below summarizes what it says and what each point implies for a Salesforce roadmap. Many of its statistics have thin citations, and the survey populations differ from yours, so read the points as direction rather than benchmarks.

  • Portfolio over fixed budget. The article argues for managing technology as a portfolio of investments instead of a fixed annual budget, and it says fixed budgets encourage silos and a project mindset. For a Salesforce team, that suggests presenting the roadmap as a small portfolio with different risk levels, not a single block. Label each item as a keep-running, improve or new-capability investment, and show the mix. Show the mix on a single page.
  • Credible business cases. It says every investment should have a calculated, preferably tangible return, and that finance and technology should jointly assess cloud costs. It reports that 37% of surveyed organizations used defined processes or business case templates, citing a 2018 source. Bring finance into the model early, and use their template if finance keeps a standard form, so the numbers don’t get rebuilt at the last minute. Ask finance what return threshold it expects for this kind of spend.
  • Shared ownership. A figure in the article says joint IT and business ownership of investments was more common among leading organizations, at 60% versus 52% for a baseline group, though the text gives no endnote for it. Treat it as suggestive. Name a business owner for each roadmap outcome, and ask that owner to speak at the funding meeting instead of the Salesforce team alone. Their presence signals commitment and shows leaders who answer for the result.
  • Measuring impact. The article reports that 20% of organizations had a dedicated finance staff member for IT investment value, from a 2018 source, and 14% didn’t measure investment impact, without a source. Offer to report results on a fixed schedule, and ask finance to help define the measures. A promise to measure and report is cheap to make and often persuasive. Agree on the reporting format before the first gate so nobody argues about definitions later.

An invented funding ask in 3 gates

The table below turns the invented plan into 3 tranches with gates, so leaders approve money in parts rather than all at once. It’s invented, and the split is a design choice, not a sourced standard. Tranche 1 covers the first 2 requests at 1,600 hours plus $60,000 of licenses, tranche 2 covers quote approval with 1,400 hours plus $90,000 of partner work, and tranche 3 covers service triage with 1,300 hours plus $90,000.

GateItems fundedInternal hoursCashTotal at $95 an hourRelease condition
Gate 1, months 1 to 4Lead routing; duplicate cleanup1,600$60,000$212,000Baselines captured and signed by owners
Gate 2, months 5 to 8Quote approval1,400$90,000$223,000Gate 1 outcomes within agreed range
Gate 3, months 9 to 12Service case triage1,300$90,000$213,500Gate 2 outcomes within agreed range
Total4 items4,300$240,000$648,500Reserve of about 2,700 hours stays visible

What the evidence says, and where it stops

The strongest independent evidence I found is the 2012 McKinsey and Oxford study, which drew on more than 5,400 IT projects and found average cost overruns of 45%, schedule overruns of 7% and value shortfalls of 56% on large projects. It also reported that 17% of projects had overruns above 200%, and that each extra year of scheduled duration added 15% to cost overruns. Its authors recommended a well-defined business case and a short diagnostic that compares a project with reference-class outcomes, and they described a health care case where leaders halted a long project and asked for intermediate checkpoints.

The evidence has 3 limits. First, the study is more than 14 years old and covers large projects, mostly outside Salesforce. Second, I found no data on Salesforce roadmaps, funding approval rates, or the effect of hiring planning consultants. Third, Fast Slow Motion’s figures and Deloitte’s survey data are either unsourced or dated. Those limits don’t mean planning fails. They mean the buyer should treat every number as a hypothesis, record it with a date, and compare it with their own results after the first gate. Treat any claim in this post, including mine, as something to test at your first gate.

Platform health as a line item

Leaders often cut maintenance first because it produces no visible feature. Gearset, a vendor of Salesforce deployment tools, published a technical debt guide in October 2025 that offers a measurement idea and a ranking method. The list below summarizes it and what it means for a roadmap. It’s vendor guidance with no original data, so use it as a method to try rather than proof of savings. Treat the guide as a starting method.

  • A debt ratio. The article defines a ratio of remediation cost to development cost, multiplied by 100, and says a rising ratio shows that maintenance is consuming more capacity. That gives a leader a single number to track each quarter, though the article offers no target value. Calculate your first ratio from last year’s time records, and agree with finance on how remediation work will be labeled. Recalculate it every quarter and share the trend.
  • Impact and effort scores. It suggests scoring each debt item from 1 to 5 for business impact and 1 to 5 for effort, and tackling high-impact, low-effort items first. This resembles the consultancy’s formula, so a single scoring session can cover both. Keep debt items on the same list as features, so leaders see the trade-off and can decide with full information. Agree on the scale definitions up front so scores mean the same thing to everyone in the room.
  • No budget percentage. The article gives no share of time to reserve for debt, and it folds debt work into the normal delivery rhythm. Fast Slow Motion’s 10% to 15% for platform health is unsourced. Pick a starting share, write it into the capacity plan, and review it at each gate based on your ratio and your incident history. Whatever share you choose, label it as an assumption and defend it with your own incident history.
  • Unsourced claims. The article says a small share of debt drives most deployment problems and that AI-generated code makes debt compound faster, with no data for either. Don’t repeat those claims to a funding body without your own evidence. If you can count your own incidents by cause, that count will persuade more than a general claim. Treat vendor claims on any side of this debate with the same caution, including the ones in this post.

Where roadmaps fail

Overcommitment. The most common failure is a plan that fills every hour. The invented example used 7,000 of 7,020 hours, which leaves no room for break and fix work that always arrives. Fast Slow Motion warns that planning too far out leads to overcommitting. Hold a visible reserve, fund only the items you can finish, and let leaders see the list of options that wait for a later gate. Review it monthly with the sponsor.

Benefits that nobody owns. A benefit without a named owner and baseline can’t be defended a year later. Ask each outcome owner to sign the estimate and the measure, and ask finance to confirm how the number will be calculated. If an owner won’t sign, drop the benefit from the case. A smaller, owned number is more credible than a larger, orphaned figure. Revisit each owner’s commitment at every gate, and record any change.

A roadmap that nobody owns. Consultants can write a plan in weeks, but the team has to live with it for a year. If the product owner didn’t help build the scoring, the plan becomes someone else’s document. Keep the owner in every workshop, and have the team present the plan to leaders, with the consultants in support. That also tests whether the team understands the trade-offs. Document who decides what, and when.

No stopping rule. Plans that can’t be stopped keep absorbing money, and leaders sense it. McKinsey and Oxford describe a case where top management asked for intermediate checkpoints rather than approve a long project outright. Build gates with explicit release conditions, such as baselines captured and outcomes within an agreed range. Say in advance what happens if a gate is missed, so a miss triggers a conversation instead of a surprise.

Hiring help without losing ownership

Roadmap consulting can save time when your team lacks the capacity to run workshops, collect data and draft the case. It can also produce a polished document that nobody owns. Fast Slow Motion’s post ends with a call to book a roadmap session, which is a reminder that firms in this field sell the next engagement. That doesn’t make their advice wrong. It means you should ask for deliverables you can reuse without them, and decide in advance what the engagement must produce and when it ends. Write the end date and the handover into the statement of work, and pay for deliverables as they arrive rather than all at the end.

Useful deliverables are specific and checkable: a baseline of today’s indicators, a scored inventory, a capacity plan with a reserve, a cost model with ranges, a benefit register with named owners, and a gated funding schedule. Ask for the working files in a format your team can edit. Request a fixed scope and a fixed end date, and keep your product owner in each session. I found no independent evidence on how well roadmap consulting performs, so treat claims of outcomes as unproven until the firm shows results from a customer similar to you.

Questions to ask a roadmap consultant

The questions below turn the gaps in this post into requests you can send before you sign. Ask for written answers, attach them to the statement of work and keep them in the project file. If a firm can’t answer, record that, since it tells you how much of the work your team will carry. Send the list to 2 or 3 firms at the same time and compare their answers side by side.

TopicQuestionA satisfactory answer includes
DeliverablesWhat will we own at the end, in what format?A list of editable files and templates
MethodHow do you score and rank requests?A written method our team can rerun
CapacityHow do you reserve and break and fix work?A capacity plan built from our ticket history
BenefitsHow do you test benefit estimates with owners?Signed estimates with baselines and dates
Stress testWhat happens to the case if cost and benefit are off?A sensitivity table with break-even points
GatesHow do you design release conditions?Conditions tied to measured outcomes
ReferencesWhich similar customer can show before and after numbers?A reference call and documented results
HandoverWho presents the plan to leaders?Our product owner, supported by the firm

FAQs

What should a 12-month Salesforce roadmap include?

A fundable plan starts with 3 to 5 business outcomes and their baselines, a tagged inventory of all work, a scored list of requests, a capacity plan that includes break and fix and platform health, a cost model, a benefit register with named owners and a gated funding schedule. Salesforce’s training material covers the business case but not these planning steps, so build them yourself and test them with finance before the meeting.

How do we prioritize requests?

A single consultancy recommends scoring value, confidence and effort from 1 to 5 and ranking by value times confidence divided by effort, with a risk flag for compliance or security items. Those choices are the firm’s own and untested. Gearset suggests scoring technical debt by impact and effort in a similar way. Use the formula to start a conversation with sponsors, and keep the final decision with people who own the outcomes.

How much capacity should we reserve for non-roadmap work?

The consultancy suggests 60% to 70% for planned roadmap work, 20% to 30% for break and fix and small requests, and 10% to 15% for platform health, with no source. Those ranges can add up to anywhere from 90% to 105%, so reconcile them. Measure a quarter of your own ticket history, set a starting split, and review it at each gate. A visible reserve also protects the plan from surprises.

What do overrun statistics tell us?

A 2012 study by McKinsey and Oxford of more than 5,400 IT projects found that large projects, with initial price tags above $15 million, averaged 45% over budget and delivered 56% less value than predicted. Your roadmap is probably smaller, and the data are old, so don’t apply them directly. Use them to design a stress test, and show leaders the break-even point if cost and benefit both miss. Keep the assumptions visible.

How do we make benefits credible?

Name an owner for every benefit, capture a baseline, agree on the measure with finance and set a date for checking. Salesforce’s training material names 4 value levers, revenue, cost, productivity and strategic capability, and gives no benchmarks. Avoid double counting across levers, and drop any benefit nobody will sign. A smaller number with an owner is easier to defend than a large number with no owner. Check the figures again at each gate, and report any drift.

Should we fund the whole year at once?

I found no evidence on approval outcomes, but the sources point toward staged funding. Deloitte argues for managing technology as a portfolio with periodic review, and McKinsey and Oxford describe a case where leaders asked for intermediate checkpoints. A gated schedule with release conditions lets leaders approve the year’s envelope while limiting exposure. Design the gates with finance, and state what happens if a gate is missed. Document the agreed conditions in writing.

Do we need roadmap consultants?

It depends on your team’s capacity and skills, and I found no independent evidence on consulting outcomes. Consultants can run workshops and draft the case faster, but the plan must belong to your team. Ask for reusable deliverables, a fixed scope, a fixed end date and a reference customer with before and after numbers. Keep your product owner in every session, and have the team present the plan. Document the arrangement in writing.

What does the 90-day view add?

A consultancy calls 90 days the sweet spot, arguing that annual plans are too abstract and weekly backlogs too reactive. A 90-day commitment gives leaders detail they can check, while the remaining months stay as themes with ranges. Measure results at 30, 60 and 90 days as the firm suggests, then update the plan at each gate. The consultancy gives no data for its cadence, so test it against your own pace.

By

Leave a Reply

Your email address will not be published. Required fields are marked *