94% of the shortfall sat in 10 of 40 locations, and the media budget increase was avoided
A multi-unit US consumer services group, roughly 40 locations under a portfolio of regional brands
The Brief
The operator was growing and, at the portfolio level, beating its budget. That was exactly the problem: a healthy topline was hiding an unhealthy middle.
The arithmetic is worth stating plainly. Fourteen of the forty locations finished the month behind plan. The five strongest locations, on their own, produced a surplus equal to roughly 91% of the entire below-plan shortfall, and the rest of the above-plan set covered it a second time over. So the portfolio closed 2.8% ahead of plan while more than a third of the estate was losing ground, and nothing in the standard reporting pack made that visible.
Leadership could see that some locations were underperforming, but not why, and the instinct in the room was the usual one: spend more on marketing. Three things made that instinct dangerous.
The reporting could not tell demand from conversion. Marketing spend and leads were reported at cluster level for several brands, not by location. Conversion meant a phone call or a form fill, not a booked, paying customer. So cost per lead was known, but true cost per acquired customer was not. A location could look like it had a demand problem when it actually had a booking problem.
Diagnoses were being assigned by anecdote. Weak locations were attributed to churn, or to a soft local market, without a measure of repeat behavior anywhere in the data set.
The obvious answer was the wrong answer. With no attribution linking a lead to a booked customer, adding budget on top of a broken funnel would have produced more leads, more cost, and no incremental revenue, while looking like activity.
| Measure | Value | What it means |
|---|---|---|
| Locations finishing below plan | 14 of 40 | More than a third of the estate losing ground |
| Shortfall carried by the 10 softest | ~94% | Concentrated A targeted problem, not a portfolio-wide one |
| Offset by the 5 strongest alone | ~91% | Enough to hide the gap before the other 21 are counted |
| Reported portfolio result | +2.8% | Ahead of plan, and telling leadership nothing useful |
The Approach
The governing principle was diagnose before spend, and its corollary: separate what the financials prove from what the evidence merely suggests, and label the difference out loud. Four commitments shaped the work.
Anchor to one clean, closed financial month. All variance work was done on a single closed month so every figure tied to the same source and could be re-derived by the client’s own finance team. No blended periods, no forecasts dressed up as findings.
Concentrate, do not spread. Rank every location by revenue variance to budget, isolate the below-plan set, and test whether the shortfall was broad or concentrated. If concentrated, the fix is targeted and cheap. If spread, it is structural and expensive. That question had to be answered before any recommendation was made.
Assign a root cause, an owner, and a confidence level to every priority location. Each gets a named constraint, the specific next decision, the person who owns validating it, the data needed to confirm it, and an explicit confidence level, so leadership can see which findings are proven and which are still hypotheses.
Diagnose at the service line, not at the location. A location total is an average of several different businesses running under one roof. Every priority location was broken into its individual service lines and measured against each line’s own plan, because that is the level at which a demand problem and a capacity problem stop looking alike.
What we Did
Rebuilt the variance picture from the source financials. Ranked all 40 operating locations against budget for the closed month and isolated the below-plan set. The finding that reframed the engagement: the shortfall was highly concentrated. The ten softest locations carried about 94% of the total below-plan revenue. That single fact converted an expensive portfolio-wide problem into a targeted eleven-location problem.
An eleventh location, healthy and closest to plan, was added to the priority set as a control rather than as a recovery target, so every diagnosis had something known-good to be measured against.
How the operating estate was narrowed to the priority set. Each pass had to be answered before the next was worth running.Decomposed every priority location into its service lines. Locations within a few points of each other on revenue attainment were missing completely different parts of the business. One was soft in its core line while every attached line ran at or above plan. Another held its core line at plan and lost the entire gap in two attached lines. A third was well above plan in its highest-margin line and behind in the line that feeds it.
Those three call for three different responses, and nothing in the location-level view distinguishes them.
Each priority location by service line, as a percentage of that line’s own plan. Service lines are generalized and location names removed.
Concentration of the below-plan revenue by location, coloured by root cause. Location names and absolute revenue figures removed.Diagnosed each priority location individually, not by category. Financial variance alone cannot tell you why a location is soft, so each was tested against service line detail, cluster-level marketing efficiency, and available operational volume data.
The single worst location by attainment turned out not to be a performance problem at all. At 54% of plan it was the newest site in the group, and its operational volume had climbed steadily month over month since opening. It had been budgeted to a run rate it had not yet reached. Diagnosis: a first-year ramp, a timing gap, not a failure.
One location reclassified from failing to a normal first-year ramp. Volume is indexed and calendar months removed.
Elsewhere, one cluster generated hundreds of leads but converted only 9% of them into customers, at the highest acquisition cost in the portfolio, while its sibling location on the same marketing ran near plan. That is funnel leakage, and adding budget would have made it worse. And one highly profitable location at 94% of plan was carrying the highest marketing spend and the lowest return on ad spend in the portfolio: the location most likely to have received more prospecting budget, and the one that needed it least.Tested attainment against profitability, and found almost no relationship. The most profitable location in the set was below plan. The location carrying the largest revenue gap was solidly profitable. Four locations sat almost on top of each other on both measures with different root causes underneath. Ranking on the revenue gap alone, which is what the existing reporting invited, would have aimed the response at the wrong locations.
Each priority location plotted on revenue attainment and operating income at once. Bubble size is the revenue gap; absolute figures removed.Sized the prize honestly, and said so. The eleven-location gap was converted to an operating income figure using the portfolio’s own operating margin as a proxy, and labeled explicitly as directional sizing, not a forecast. Two sentences in the deck exist purely to prevent the number from being over-read later.
Named the attribution gap instead of working around it. We mapped the customer journey stage by stage and marked each one visible, partial, or missing in the current reporting. This made the core defect legible to a CFO in one slide: cost per lead is not cost per acquired customer, and until a lead can be traced to a booked, paying customer, no channel decision is defensible.
Cost per lead versus true cost per acquired customer, indexed. The gap is the conversion loss the current reporting cannot see.
The customer journey, each stage marked visible, partial, or missing in the current reporting.Wrote the what-not-to-do list. Six explicit anti-recommendations, including do not run a broad portfolio-wide media push, do not add spend to the locations with a broken funnel, and do not optimize to cost per lead. In a diagnostic, the spend you prevent is as valuable as the spend you direct.
Designed a controlled test instead of a budget increase. For the one genuinely healthy, scalable market, we specified a contained, single-market test with matched control markets, a primary KPI of incremental booked customers (not leads, not impressions), pre-launch validation gates including physical capacity to absorb demand, and a decision rule stating that spend scales only if booked customers and revenue move.
The criteria were scored, not asserted. The selected market converted at the better end of the portfolio while acquiring customers at a fraction of the portfolio’s cost, and booked roughly five times the new customers of the next candidate. One efficient-looking alternative was excluded outright because a major campaign was already scheduled there inside the flight window and would have contaminated the read.
Every candidate cluster scored on conversion and true acquisition cost. Market names removed and acquisition cost indexed.
Results in Detail
This was a 30-day diagnostic. The output is decisions, not yet realized revenue, and the case is stronger for saying so. Every figure below is drawn from the client’s own financial and marketing source files and was reconciled to them line by line.
Shortfall isolated to 11 locations out of 40. The ten softest account for approximately 94% of total below-plan revenue. The problem was reframed from portfolio-wide to targeted.
Media budget increase avoided. The immediate marketing recommendation was a creative and targeting refresh inside the existing budget, with no incremental media spend requested. The only new money asked for was a small, contained, control-matched test.
This is the result that pays for the engagement. The instinct in the room was to spend more; the diagnostic showed that on a funnel this leaky, more spend buys more leads and no more customers.
Two locations pulled back from the wrong intervention. The worst-attaining location was reclassified from failing to on a normal first-year ramp, changing the decision from remediation to funded, ramp-benchmarked growth. The most profitable soft location was reclassified from scale prospecting to test repeat behavior.
Every hypothesis shipped with a confidence level. Three diagnoses at high confidence, five at medium-high, three at medium, published rather than implied, so leadership could see at a glance which findings would survive scrutiny and which had to be validated first.
Six decisions put in front of leadership, each tied to a specific finding and a named owner, so the diagnostic closed with approvals rather than with a reading list.
The mask was quantified rather than just described: 14 of 40 locations sit below plan, with the five strongest offsetting 91% of the shortfall. A directional operating income opportunity was sized from the priority-location gap using the portfolio’s own margin, carried with an explicit sizing-not-forecast caveat. Funnel leakage was identified with high confidence in a high-volume cluster converting 9% of leads at the portfolio’s highest acquisition cost. An attribution build was specified, turning a vague data-is-messy complaint into a defined next-phase scope with named data requirements and owners. Service line decomposition separated diagnoses that looked identical at location level, and attainment and profitability were shown to be nearly unrelated, so priority was set on constraint and confidence rather than attainment alone.
What the diagnostic prevented
- A portfolio-wide media push against a problem that lived in a quarter of the estate.
- Prospecting budget for the location whose constraint was retention, and more leads for the cluster already failing to book.
- Optimizing to cost per lead, the one metric the reporting could produce and the one least connected to revenue.
Tools & technologies
What the build runs on
verified