How to work out if you should reduce or extend opening hours
The decision that gets made for the wrong reasons
Opening hours are one of the highest-impact decisions a contact centre takes and one of the worst-analysed. Most operations don’t actually decide their opening hours — they inherit them, then either defend them when finance asks why they cost so much or extend them when customer experience asks why they’re so short. The decision drifts; the analysis happens in retrospect; the numbers used to defend whichever side prevails are usually thin.
A defensible opening-hours decision needs eight pieces of data, two scenarios modelled honestly, and a clear view of what happens to the demand that shifts. This article walks through how to do that properly, the common traps, and the operating-model considerations that the headline numbers don’t capture.
The first principle — demand doesn’t disappear
The most common mistake in this analysis, by an order of magnitude, is treating reduced opening hours as if the demand inside those hours simply vanishes. It doesn’t. The customer who would have called at 8:30am calls at 9:05am instead. The customer who would have called at 7:30pm calls the next morning. Some of that demand shifts into your operating hours and lifts the cost of your peak. Some shifts into self-service if the self-service is good. Some goes to a competitor. Some becomes a complaint when the customer realises you’re closed. A few customers genuinely give up. The mix of those four destinations is the single most important variable in the analysis, and the one most operations don’t model.
The mirror principle applies to extending hours. The new demand you capture in the extended window isn’t all new. Some of it is demand that would have arrived inside your existing hours and simply shifted earlier or later. That portion doesn’t generate new revenue or save existing customers — it just spreads the same demand across more hours and lifts your cost-to-serve.
The eight inputs you need
A workable model has eight data inputs. Most operations have six of them readily available; the two harder ones are worth the work.
1. Volume curve by 15-minute interval. Across the full 24 hours, not just your operating hours. Demand outside your hours is captured by voicemail volume, IVR-abandoned attempts, callback-request volume, after-hours email arrival, and the “contact us” volume that comes the next morning. None of these capture it perfectly; together they triangulate.
2. Cost-to-serve by hour. Fully loaded cost per agent-hour, including unsocial-hours premiums, supervision ratios, technology overhead, and the marginal cost of opening (lighting, heating, security if site-based). Cost-to-serve at 9am is not the same as cost-to-serve at 9pm; the difference can be 30–50%.
3. Abandonment by hour of day. The bookend hours (just after opening, just before closing) usually have substantially higher abandonment than the middle of the day. That signals demand the operation is failing to serve at the boundary.
4. Customer demographic and segment. Vulnerable customers, older customers, and customers in different time zones interact with opening hours differently. A B2B operation may have most of its demand during business hours; a B2C operation will have substantial evening and weekend demand. Demographic-level segmentation matters here.
5. Channel substitution. What proportion of customers will use self-service if voice isn’t available? What proportion will email? What proportion will wait until you re-open? This is the hardest to measure and the one with the biggest impact. Operations that have run actual experiments — closing for a fortnight and watching the channel mix — have real data; operations that haven’t are guessing.
6. Competitor opening hours and complaint volume. If your competitors are open when you’re closed, the substitution risk is real. If they’re closed too, the customer has fewer options and the demand is more likely to wait. Complaint volumes linked to opening hours (“couldn’t get through outside business hours”) are a leading signal.
7. Regulatory constraint. Some sectors have explicit duty-to-respond requirements. UK financial services under Consumer Duty (see Consumer Duty for planners), regulated utilities under Ofgem, betting and gambling under UKGC. These constraints can make reducing hours below a threshold non-compliant regardless of the economics.
8. Agent appetite. Are agents willing to work the proposed extended hours? Voluntary uptake is usually high in the first month and lower after six. The schedule has to be sustainable, not just possible. See can work-life balance coexist with scheduling for customer demand.
The two scenarios you need to model
The reduce scenario. Take the lowest-volume hour at each end of the current operating window. For each, model: cost saved (cost-to-serve at that hour × hours removed), revenue lost (customers in that window who don’t return), peak uplift cost (proportion of displaced customers who land in your peak the next day, multiplied by the marginal cost of that peak demand), and substitution loss (customers who go to competitors or give up). Total = cost saved − revenue lost − peak uplift − substitution loss. Most reduce scenarios deliver less than half the headline cost saving once the other three lines are honest.
The extend scenario. Take the hour immediately before opening or after closing. For each, model: net new contacts captured (demand × close-rate or resolution-rate), cost incurred (cost-to-serve at that hour × staffing required), revenue retained (customers who would otherwise have churned or moved channel), substitution from existing hours (demand that shifts from peak into the extended window — this is value not created, just relocated). Total = new contacts captured × value − cost incurred − the substitution credit (because that demand was already being served). Most extend scenarios deliver more substitution than genuinely new demand, and the cost is direct while the benefit is partial.
Common mistakes
Treating reduced-hours demand as lost. Most of it isn’t lost — it shifts. If your model assumes 100% loss, the case to keep extended hours looks artificially strong. If your model assumes 100% absorption into peak, the case to reduce looks too easy. Use historical data from prior changes, or run a controlled experiment.
Ignoring the cost shape. A 7am hour and a 7pm hour aren’t the same cost-to-serve. Unsocial-hours premiums, supervisor coverage, building costs, security — the cost line varies meaningfully across the day. Most opening-hours analyses use a flat cost-per-agent-hour and overstate the benefit of extension.
Ignoring the volume shape. The first hour after extension is rarely as busy as you expect; customers take a quarter or two to adjust their behaviour. The first hour after reduction is busier than expected for the same reason. Don’t model steady-state from day one.
Forgetting vulnerable customer dependence. A small but important fraction of your customer base depends on the hours they can call — carers calling from home in the evening, people whose employers don’t allow personal calls during the day. Cutting evening hours often disproportionately affects vulnerable customers; the regulatory exposure is real.
Letting the noisiest internal voice win. The agent who hates Saturday shifts is louder than the customer who calls on Saturday. The CFO who wants the cost out is louder than the brand team who cares about access. The decision should be defended by the data, not by who’s in the room.
What good operations do when the numbers genuinely don’t tell you
Sometimes the analysis comes out marginal — the reduce case saves a small amount and risks a small loss; the extend case captures a small amount of new demand at a small cost. The honest answer is “we don’t know.” The leading operations have one of two responses.
Run an experiment. Pick a quarter. Change the hours. Measure what actually happens to volume, abandonment, complaint volume, channel mix, and revenue. Reverse the change if the data goes against. A quarter of experimental data beats six months of analysis-by-spreadsheet on a marginal call.
Default to access. If the numbers are marginal, the operations that retain the longer hours tend to have stronger long-term customer loyalty than the operations that cut them. The marginal-cost saving of cutting an evening hour rarely shows up in any single financial year; the customer-loyalty signal of having been open when needed does, slowly. When in doubt, keep the access. The asymmetry favours customers.
The leadership conversation
The opening-hours decision is one of the few where the CFO conversation and the customer-experience conversation collide visibly. The honest leadership framing isn’t “we can save £X by cutting an hour” or “customer experience suffers if we do.” It’s “here’s the cost-to-serve at each hour, here’s the demand at each hour, here’s where they cross over, and here’s what we think happens to the demand we’re no longer serving.” That framing keeps the conversation in the data and out of the personalities.
Conclusion
Opening hours are too important to decide by inheritance or by whichever internal voice is loudest. The defensible decision starts from the eight data inputs, models both scenarios honestly with the displacement effects included, and acknowledges when the numbers are too marginal to be confident. The operations that handle this well treat opening hours as an annual review, not a once-a-decade conversation, and don’t assume that the demand they stop serving has disappeared.
Pair this with setting the right service-level target, can work-life balance coexist with demand, the true cost of attrition, Consumer Duty for planners, and the cost-per-contact calculator.