A gap analysis produces two numbers and the distance between them. Most teams start planning how to close that distance without checking whether either number is real. Skip that check, and the output looks rigorous on a slide but falls apart when someone asks where either measurement came from.
I watched a professional services firm run a gap analysis on its client service capability. The planning team compared current service delivery scores, drawn from annual manager surveys, against an industry benchmark published by a consulting group. The gap looked clear: staff lacked the skills to meet rising client expectations. The firm funded a six-month training program across three offices.
Twelve months later, client satisfaction scores had barely moved. When someone finally observed actual service delivery, the constraint was a handoff between two internal systems that added three days to response times. Staff had the knowledge. They lacked the tooling. The "current state" had been measured by what managers believed, not by direct observation of work. The "desired state" had been a benchmark drawn from firms with a different client mix and a different operating model.
Both sides of the gap were assumptions dressed as measurements. The training program closed a gap that did not exist, while the real constraint sat in a process nobody had examined. The format of the gap analysis made the distance look factual. The format does not test whether either side is true.
- Rewrite both sides of the gap as testable claims. "We lack digital maturity" becomes "We believe our digital capability is below the level required for our operating model."
- Score each claim on influence and confidence. High influence + low confidence = the side that needs investigation before anyone plans a closure.
- Investigate the high-risk claims. One data check per side.
- Redefine the gap using only verified positions. The Walk runs you through testing whether each side is measured or assumed, before the action plan commits budget to closing a gap that may not exist.
- Assign ownership and a review trigger. The gap owner names the condition under which the gap should be remeasured.
After a gap analysis, the standard next step is an action plan to close the gap. The step worth taking first is to test whether both sides of the gap rest on evidence or assumption.
The standard next step: plan how to close it
After gap analysis, most planning guides recommend an action plan. Identify the interventions that move the organisation from current state to desired state, assign resources, set timelines. The logic is straightforward: the gap is the problem, the action plan is the solution.
The concept has deep roots. Parasuraman, Zeithaml, and Berry formalised the gap model in 1985, defining service quality as the distance between customer expectations and customer perceptions. Their SERVQUAL framework gave the gap a measurable shape: survey both sides independently, compute the difference, act on the largest gaps first. The model worked because both sides were surveyed, not assumed.
The appeal is obvious. A measured gap gives a team a target. An action plan gives the target a schedule. Together they produce a narrative that sounds analytical: we know where we are, we know where we want to be, and here is the plan to get there. When both sides are genuinely measured, that narrative holds. The problem arises when one side, or both, is a guess wearing the format of a measurement.
What the closure step adds
Gap analysis forces a comparison. Without it, teams describe aspirations without anchoring them to a starting position. The comparison itself is the contribution. It creates a distance that can be prioritised, resourced, and tracked. A situation analysis that includes a gap step is more actionable than one that stops at description.
McKinsey's 2020 workforce survey found that 87 per cent of companies reported either facing skill gaps already or expecting them within the next five years. The skill gap has become the dominant frame for workforce planning. It structures hiring decisions, training budgets, and reorganisation initiatives across nearly every sector.
The closure step takes that frame and turns it into a project. It converts the distance between two positions into a set of interventions with budgets and owners. That is a genuine improvement over leaving the gap as a slide in a strategy deck. The tool does useful work when the inputs are sound.
Rewrite both sides of the gap as claims and test them before the closure plan commits resources to a fictional distance. Start the Walk →
Where the standard playbook breaks down
The closure step inherits a problem from the analysis itself. It treats "current state" and "desired state" as settled measurements. In practice, neither side is usually as solid as the gap format implies.
Parasuraman and colleagues' original SERVQUAL model worked because both sides were independently surveyed. Customer expectations were measured through structured interviews. Customer perceptions were measured separately. The gap between them was calculated, not asserted. Most organisational gap analyses skip that discipline. "Current state" comes from a manager survey, a self-assessment, or last quarter's KPI dashboard. "Desired state" comes from an industry benchmark, a competitor's published metrics, or a target carried forward from the previous planning cycle.
McKinsey's own skills gap figure illustrates the pattern. The 87 per cent came from executive self-reporting. Managers said they perceived skill gaps. But perception surveys measure belief, not capability. A team that recently adopted a new system may report a skills gap when the actual constraint is process design. A team whose tools are outdated may report no gap at all, because the staff have adapted and managers no longer notice the limitation. Closing a perceived gap without checking whether the perception matches reality is the same error the firm in the opening story made. A gap worth closing is one where both sides have been examined.
The NHS Digital Maturity Assessment shows a different version of the same problem. The assessment measures digital capability against a national framework called What Good Looks Like. Ercole's cross-sectional analysis of 111 acute trusts found that trusts in the highest maturity quartile operated at 98.0 per cent of their production frontier compared with 93.2 per cent for the lowest quartile. But the framework measures what is deployed and available, not how those systems are experienced in practice. A trust scoring well on the benchmark might still have clinicians working around software that slows them down. The desired state, as defined by the framework, measures inputs. The actual desired state, for a clinician, is a system that does not add steps to patient care.
I have seen this repeatedly across nearly fifty years of working with planning teams. A board adopts a target from an industry report. A planning team measures the organisation against that target. The gap is presented as a finding. The finding becomes a budget line. At no point does anyone ask whether the industry target applies to this organisation, or whether the internal measurement reflects what staff actually do rather than what a dashboard reports. Several documented cases show what happens when neither side of the analysis has been verified.
The step to take first
Before an action plan can close a gap, both sides need to survive a simple question: what are we assuming here?
Take the current-state description and rewrite it as a claim. "Our service delivery scores are below industry benchmark" becomes "We believe our service delivery scores, as reported by managers in an annual survey, are an accurate reflection of client experience." Then ask two things about that claim: how much influence does it have on the decision to invest, and how confident are we that it reflects reality? If the entire closure plan rests on a score that nobody has verified against operational data, the claim needs investigation before it generates a project.
Apply the same test to the desired state. "Industry benchmark is 4.1" becomes "We believe 4.1 is the right target for an organisation with our operating model, client base, and regulatory constraints." If the benchmark came from a survey of firms in a different market, or if it measures inputs rather than outcomes, the target may be wrong. A gap measured against a wrong target is not a gap worth closing. The question of how much due diligence is enough applies here as much as anywhere: one check per claim, not a consultancy project.
The planning team from the opening would have listed "staff lack the skills to meet client expectations" as their gap statement. Rewritten as a claim: "We believe staff skills, as perceived by managers in an annual survey, are the primary constraint on service delivery quality." Influence on the decision: high, because the entire training budget rested on it. Confidence: low, because nobody had observed actual service delivery or compared perceived skill gaps against process bottlenecks. That claim needed verification before it became a six-month program.
When I work with teams after a gap analysis, the first thing I do is sort the entries on each side into evidence and assumption. The sorting takes ten minutes. It regularly halves the number of gaps that survive into the action plan, because several turn out to rest on one assumed current state or one inherited target.
This is the step between completing a gap analysis and committing resources to close it. Test both sides before funding the bridge. The Universal Decision-Making Method calls this "recognise assumptions" and places it before any commitment, because an action plan built on an untested gap is not a plan.
You could measure both sides and still close a gap between a guess and an inherited target.
Work through your decisionNo sign-up. Just pick your decision and start.
Grant Purdy is the co-author, with Roger Estall, of Deciding (2020), and the architect of the Universal Decision-Making Method.