Quick answer
A traffic spike is a sudden rise in visits, events, errors or conversions well above the normal level for that day and hour. It is defined against a baseline, and it does not say whether the rise is good: it can be real attention, bots or a tracking error.
Formula: Spike multiple = visitors in the hour ÷ average of the same hour over the past 7 days
What is a traffic spike?
A traffic spike is a sudden rise in visits, events, errors or conversions well above what is normal for that day and hour. "Normal" is the important word: a spike is defined against a baseline, not against a fixed number. A site that usually gets 80 visitors an hour has a spike at 400; a site that usually gets 50,000 does not.
A spike tells you something happened, not whether it is good. It can be a post that took off, a mention in the press, a newsletter, or a crawler hammering your pages. The work is to sort it into one of those quickly. The rest of this page is a short method for doing that.
Where spikes come from
| Source | Typical sign | Example |
|---|---|---|
| Campaign or newsletter | Traffic from a known source, one landing page, starts at a known time | Email sent at 09:00 |
| Social or press mention | One referrer, one article, many new visitors | A post shared by a large account |
| Search ranking change | Gradual climb, organic search, many pages | A page reaching page one |
| Bots and scrapers | One country or network, very short or zero engagement, odd page mix | A scraper reading every product page |
| Attack | Abrupt, enormous, often one URL; the site slows or fails | A flood of requests meant to overload |
| Measurement artifact | Numbers rise but real behavior does not | Duplicated tracking code, an uptime monitor |
One analytics glossary lists these among the causes of a spike: viral social posts, newsletters, paid campaigns, product announcements, rising rankings, press coverage, bots and scrapers, uptime monitors and duplicated tracking (see Sources). The two measurement artifacts are the cheapest to check and the easiest to forget.
Four checks to tell real growth from noise
- Human or machine? Look at whether the extra traffic is recognized as bots. Real people spread across pages and times; bots arrive in blocks.
- How concentrated is it? Count the share held by the top source, top landing page and top country. A real mention concentrates on one source and one page but spreads across countries and devices. Bot traffic concentrates on everything at once.
- Does quality move with volume? Compare pages per session, average duration and conversion rate in the spike against the baseline. Real attention usually lowers conversion a little (new audiences) but not to zero.
- Is it your own tracking? Check for a recent release, a duplicated tag or a monitor pinging the site.
The rule that ties these together: a spike is growth only if people behave like people. If volume rises and duration and conversions do not move at all, it is not attention. If volume rises, duration holds and conversions rise in proportion, it is growth.
A worked example: two spikes of the same size
Sample data: a site's visitors in the same hour over the past seven days are 80, 85, 78, 90, 82, 88 and 81. The sum is 584, so the baseline is 584 ÷ 7 = 83.4 visitors. Today the same hour has 1,250 visitors, which is 1,250 ÷ 83.4 = 15.0 times the baseline. Any rule that fires at three times the baseline (3 × 83.4 = 250) would catch it. The volume is identical in the two cases below, and the meaning is opposite.
| Normal hour | Spike A: viral post | Spike B: scraper | |
|---|---|---|---|
| Visitors in the hour | 83 | 1,250 | 1,250 |
| Sessions in the hour | 90 | 1,300 | 1,300 |
| Top source share | 38% | 71% (one social site) | 97% (direct) |
| Top landing page share | 22% | 68% (one article) | 99% (one URL) |
| Top country share | 45% | 42% | 94% |
| Pages per session | 2.6 | 1.4 | 1.0 |
| Average duration | 96 s | 38 s | 0 s |
| Conversion rate | 2.0% | 0.6% | 0.0% |
| Conversions in the hour | about 2 | about 8 | 0 |
Spike A: 1,300 sessions at 0.6% gives 7.8, about 8 conversions. At the usual 2.0% it would have been 26. The audience is real but not ready to buy, so the useful work is on the article: add a next step for visitors who arrived cold. Spike B: everything concentrates, nobody engages, nobody converts. Treat it as noise: exclude it from analysis and, if it burdens the server, block it there.
What a spike does to your site and your data
A spike is not only a reporting event. A sudden surge from a popular link can slow a small site or take it offline; this is sometimes called the Slashdot effect, and it behaves much like an attack although it is accidental. If pages get slow, visitors leave, so a successful spike can reduce its own conversions.
It affects the data as well. At very high rates, collection services protect themselves. In TapCub, when a minute carries more than 1,000 events and more than ten times the same hour's average over the previous seven days, collection keeps one in every ten events for about ten minutes. The reports in those minutes show fewer events than arrived. A marked-up chart helps later readers: note the cause of a known spike with a date annotation so the next analyst does not chase it.
Common mistakes
- Celebrating before checking. A large number with zero duration is not an audience.
- Judging against yesterday. Compare against the same hour or weekday over several weeks. A Monday is not a Sunday.
- Using a fixed number as the alarm. A threshold that fits a small site fires all day on a large one. Use a multiple of the baseline.
- Alerting on small sites. At 5 visitors an hour, 15 is three times normal and means nothing. Add a floor.
- Ignoring errors during a spike. Errors and slow pages are part of the story; a spike with a rising error rate costs you visitors.
- Blocking before understanding. Excluding a crawler from your reports only stops it being counted; it is still reading your site. Refusing it takes a rule on the server.
- Leaving a one-off spike in the baseline. Rules that learn from recent history, such as an anomaly band built from the same weekday over several weeks, get wider after a spike. Annotate it so you can discount it.
For the opposite problem, see what to do when traffic suddenly drops. For spam that shows up as referrals rather than visits, see referral spam. The method of finding unusual values from a baseline is anomaly detection, and bot explains what the machines in the table are.
Handling a spike in TapCub
Take the 1,250-visitor hour. This is how it plays out.
- Get told. In Configuration › Settings › Notifications, TapCub has a built-in traffic spike alert among its operational alerts, sent to the channels you set up: email, WeCom, DingTalk, Feishu or a webhook. By default it compares the last hour's visitors with the average for the same hour over the previous seven days, and fires above three times that and at least 50 visitors. The message names the top sources and entry pages. For your own metrics, use All-platform analytics › Operations › Monitors: pick
pv,uv,sessions,bounce_rate,avg_duration,error_rateorerrors, then a condition (fixed threshold, period change, forecast band or anomaly). A dry run reports how many times the last 30 days would have fired. By default a repeat notification waits about 60 minutes, and three hits in a row pause the rule. An alert only notifies; it takes no action for you. - Ask: human or bot? Open Web analytics › Bots. New projects use "detect and report separately", so recognized bots go only to that report and do not inflate human visitors. With only the page script you see bots that run JavaScript; crawlers that do not need server collection or a log import.
- Ask: concentrated? Open Web analytics › Sources and Web analytics › Pages, and use Web analytics › Realtime while it is happening. The AI Assistant under All-platform analytics › Operations › AI Assistant can break a change down by channel, source domain, country, device and landing page, so you see which part contributed.
- Ask: real? Compare average duration and conversions in Web analytics › Goals against the baseline, as in the table.
- Leave a note. Add a campaign or incident annotation under Configuration › Annotations.
TapCub measures and notifies. It does not scale your servers, block traffic or change your site.
Sources
- Swetrix, Traffic spike, definition and typical causes, checked 2026-11-17.
- Wikipedia, Slashdot effect, sudden traffic surges that overload small sites, checked 2026-11-17.
- TapCub documentation, Monitors, annotations & reports; Bots, AI crawlers & AI traffic; Troubleshooting, checked 2026-11-17.
Was this explanation clear?
We use your answers to fix and expand this page.
Couldn’t send. Please try again in a minute.
Thanks for the feedback. We’ll use it to improve this page.You’ve already sent feedback on this page. Thanks.

Data Analyst
Explores traffic sources, conversion funnels, and retention, using clear metric definitions and concrete examples to explain what the numbers mean.