The Hour Nobody Bids On: Reading the UK's Mid-Afternoon Traffic Trough
Photo by Photo by James Genchi on Unsplash on Unsplash
There is a moment, usually around 15:10, when a real-time dashboard can make a perfectly healthy UK shop look like it has been switched off at the wall. Sessions soften, add-to-carts go quiet, and someone in the team asks whether the site is down. It rarely is. What you are looking at is the mid-afternoon trough: a recurring dip that shows up in plenty of small-business and e-commerce accounts, and one that is easy to misread.
The temptation is to treat the dip as a verdict: "our audience doesn't browse at 3pm." That may be true, partly. But before you start pausing campaigns and moving budgets, it is worth working out whether you are seeing behaviour, a reporting quirk, or a bit of both. The difference matters, because the fix is different each time.
First, make sure the clock is telling the truth
A surprising number of "dips" are timestamps having a bad day. Start with the boring checks. Is the analytics property set to the right UK time zone, and has it survived the switch between GMT and BST without smearing an hour of traffic into the wrong slot? Are you looking at a report built on event time or processing time? Consent banners and delayed consent signals can push recorded activity later than the visit itself, which flattens the early part of an hour and fattens the next one.
Then look at session behaviour. A default 30-minute session timeout can make a lunchtime browse and an early-afternoon return look like one long visit, or split a single visit across two hours if someone leaves a tab open. Bot filtering, caching and server-side tagging can all nudge the shape too. None of this means your data is broken. It means the hour-by-hour view is a model, not a CCTV recording.
A quick sanity check: compare the same hour across four or five ordinary weekdays, not one. If the dip moves around, it is probably noise, a campaign pause, or a bank holiday you forgot about. If it sits in the same place like a coffee ring, you have something worth investigating.
Split the hour before you blame the hour
Total sessions by hour is a blunt instrument. A dip in overall traffic might be entirely driven by paid search pausing, while organic and email hold steady. Or it might be a mobile-only effect: desktop sessions flat, mobile down, because people are on the school run rather than scrolling. Pull the hour-of-day report and add a secondary dimension or two: channel, device, new versus returning, and landing page type.
More useful still, look at quality signals alongside volume. If sessions drop at 15:00 but conversion rate and revenue per session hold up, the people who are there may be the ones who matter. If sessions are flat but conversion falls off a cliff, you do not have a traffic problem, you have an intent or experience problem. That is a different article, and a different fix.
For shops, it is also worth separating browsing from buying. A mid-afternoon lull in product views can sit alongside a steady trickle of checkout completions from people who filled a basket earlier. Session counts alone will not tell you that.
The school run, the sugar crash and the 15:30 cut-off
Some of the dip is genuinely human. In the UK, the window between roughly 15:00 and 16:00 collides with school pickups, the tail end of office meetings, and the point in the day when consumer attention moves from "browsing" to "getting out the door." For B2B audiences, it is often a meeting-heavy slot. For consumer shops, it can be a mobile-first lull followed by an evening rebound once tea is on and the sofa has been reached.
There is also a practical retail effect: next-day delivery cut-offs. If your cutoff is 3pm or 4pm, some of the urgency has already gone by mid-afternoon. People who missed it may simply wait until the evening rather than browse now. That is not disinterest. It is a timetable.
None of this is universal. A click-and-collect shop, a trade supplier, or a business selling to night-shift workers will have a different curve. The point is to build your own picture rather than importing a generic "best time to post" chart from a blog written for a different market.
Scheduling around the trough, not into it
Once you have a repeatable dip, the response is usually small adjustments rather than grand strategy.
On paid search and social, dayparting is the obvious lever, but use it gently. Bid adjustments are safer than hard exclusions, because a quiet hour can also be a cheap hour: fewer competitors bidding, lower cost per click, and a visitor who is genuinely looking rather than half-watching a video. Test a modest reduction in spend between 14:30 and 16:30 before you switch the hour off entirely. Keep brand campaigns separate; someone searching for your name at 15:15 is not the same as someone browsing a generic category term.
Email is often the easiest win. If your sends cluster at 15:00 because that is when someone in the office gets round to it, try moving them earlier, before the lunchtime scroll ends, or later, into the early evening. Test one change at a time and give it at least three or four sends before drawing conclusions. Open rates by hour are notoriously noisy.
For content and social publishing, the dip is a reasonable argument for scheduling around it rather than posting into it. A lunchtime publish and an early-evening push will often do more than a 15:00 post that lands while your audience is in the car park. For staffing, the trough is a good slot for tasks that need quiet: updating product copy, reviewing search terms, or tidying the 404 log. It is a poor slot for assuming the phones will ring.
A one-week test you can actually run
Keep it small and keep it honest. On Monday, record a baseline: sessions, conversion rate, cost per acquisition and revenue for each hour from 13:00 to 18:00, split by your main channels. From Tuesday to Thursday, move a modest slice of budget, say 20%, out of the 14:30–16:30 window and into either 07:00–09:00 or 19:00–21:00. Change nothing else: same creative, same landing pages, same bidding strategy.
On Friday, compare the test days with the baseline and with the same weekdays from the previous fortnight. Look at cost per acquisition and revenue, not just clicks. If the shifted budget performs better, repeat the test for another week before making it permanent. If it performs worse, you have learned that the quiet hour was quietly efficient. Either result is useful.
| What to compare | Why it matters |
|---|---|
| Sessions by hour | Shows the shape, not the cause |
| Conversion rate by hour | Separates volume from quality |
| Cost per acquisition by hour | Reveals whether cheap traffic is actually cheap |
| Revenue per session | Stops you optimising for clicks alone |
What not to conclude
Do not declare that "nobody shops at 3pm" on the strength of one week. Do not pause brand activity because a generic report says afternoons are slow. And do not ignore the possibility that your dip is partly your own doing: a campaign budget that runs out at 14:00, a feed that stops refreshing, or a chat widget that only appears during office hours can all manufacture a trough that looks like audience behaviour.
The mid-afternoon slump is best treated as a prompt, not a verdict. Check the clock, split the hour, then test a small change. If the dip is real, you will save some budget and stop shouting into an empty room. If it is not, you will have learned something more valuable: that your quietest hour was doing more work than it got credit for.
