Free Traffic to Website: What Spikes and What Compounds

Yes. Free traffic for a website is real, and it costs time instead of money. The useful question isn't whether a source is free. It's whether the hours you put in pay out once or keep paying.

"Usually" matters in that table. Shape belongs to a specific page on a specific site, not to a platform. A directory listing that people keep finding can trickle for years, and a search page can lose its ranking and fall off. So treat the table as a hypothesis and measure your own curves with the readings further down. Nothing in this post gives a decay rate for any platform, because we don't know of a documented one that would hold for your site.

A spike comes from a distribution event. Someone else's feed, thread or listing page pushes attention toward you once, and the attention moves on when newer items arrive. Nothing you built sits where people keep looking, so the traffic stops when the push stops.

Compounding traffic comes from a stock. A page in a search index, a subscriber in your list and a link on one of your pages are all still there next month, waiting for the next person who searches, opens an email or reads that page. Each new piece adds to the stock instead of replacing last week's effort.

That gives a simple way to read any week in your analytics:

visits this week = new spikes this week + everything older that is still earning

A site built on spikes has a small second term, so its traffic resets whenever the owner stops posting. A site built on compounders grows its second term, and a quiet week still has traffic.

Spikes aren't worthless. The best ones leave a compounder behind: an email signup, a link from a page people keep reading, or a question you hadn't thought of that becomes a page people search for. The rest of this post keeps asking the same thing of every spike. What did it leave?

Judge a spike source by its residue, not its first-day number.

The last two rows are spikes by construction, since nothing arrives unless the exchange or the bot is running. What each free bot offer hands over, and what it costs you instead of money, is covered in can you get bot traffic for free. The point here is narrower: neither can compound, and both make your decay curves harder to read.

The first four rows can be worth the time when you capture the residue. Put an email signup on the page the spike lands on. Answer the questions in the thread on your own site, where they can keep earning. Link the page the directory sends to from pages that already get traffic.

Before any reading, exclude bought, bot and exchange sessions, or your curves describe the vendor instead of your audience. How to spot bought traffic in analytics covers the patterns to look for.

GA4 shows every channel. Search Console shows one of them, and it's usually the compounding one, so read it separately.

1. Open Performance, then Search results. The default view covers "the past three months"; widen it for older pages.
2. Set the time granularity to Weekly. The report offers "Hourly, Daily, Weekly, or Monthly" to spot trends (Performance report).
3. Add a Page filter for the URL you're judging and switch on Clicks and Impressions.
4. Open the Queries tab for the same page and note how many queries earned impressions.

A compounding page shows three things over time: weekly impressions that hold or climb, clicks that follow them, and a query count that grows as the page gets shown for phrasings you never targeted. A page whose impressions arrived in one burst and fell away is spiking in search too, which happens when a topic trends briefly.

Two limits. Search Console counts clicks from Google Search results, including AI Overviews and AI Mode, and nothing from other sources, so a page doing well through email or AI assistants can look quiet here. And a visit that loads your URL without a search adds nothing to these numbers, which is why a GA4 organic label on its own proves little, as organic traffic generator: what's generated, what's earned explains.

An example site publishes two pieces in the same week. Every number below is invented to show the method. None of it is SparkCliks data or a study result.

• Piece A, the spike. A free checklist announced in a forum thread and a social post, with replies to comments over the next few days. Time spent: 5 hours.
• Piece B, the compounder. A how-to page at https://www.sparkcliks.com/example-how-to/ that answers a question people search for, with links to it added from three older pages that already get traffic. Time spent: 6 hours.

Weekly sessions by landing page, from Reading 1, bought and bot traffic excluded:

The two numbers from Reading 1:

What it says. Judged after one week, A looks more than fifty times better per hour. Judged after 13 weeks, the running totals have just crossed and A is still slightly ahead per hour. The difference is the run rate: A has settled at 10 sessions a week and B is at 170.

If both curves simply held their week-13 rate for another 13 weeks, which you would have to confirm rather than assume, A would add 130 sessions and B would add 2,210. Per hour of work that's about 296 for A against about 602 for B at week 26. The volume crossover came in week 13, and the per-hour crossover would come later still. That's the trap in judging free traffic early: a quick check always favors the spike.

The cohort check. Reading 2 on the same site shows 780 new users from Organic Social and Referral in week 1, with 23 returning in week 2 and 5 in week 4. The 40 email signups the checklist collected are the part of A that compounds, and Reading 3 would show them coming back through the Email channel.

With a few hours a week, spend most of them where the work stays in place, and spend spike time only where the spike leaves something behind.

Run a quick check before any free traffic task:

1. Will this still send visits in 90 days if I do nothing more?
2. If not, does it capture something that will, such as a signup, a link or a new page idea?
3. Can I measure its tail share in GA4, with bought and bot sessions excluded?
4. Am I judging it at week 13 or later, not week 1?

A task that fails the first two questions is a spike with no residue. Do it only when the timing matters more than the tail.

Bought traffic of every kind is a spike. It arrives while the order runs and stops when it ends, and no tail follows, because the visitors weren't looking for you. That applies to what we sell.

• Website Traffic has a free NANO tier: 6,000 page views and 2,000 unique visitors a month, visits of up to 30 seconds. Our delivery policy says the traffic is "delivered using Data Centre IPs only." It's useful for checking that analytics, filters and reports behave. It won't produce returning customers or a tail, and our own FAQ says of the visits: "It does not make purchases." Exclude it before running any reading above.
• Realistic Traffic is the same kind of delivery with scrolling and clicks on the page, has no free tier, and is a spike for the same reason.
• SERP Clicks has no free tier. Its entry plan, LITE, is $47 a month for 120 clicks, or 60 geo-targeted clicks at 2 credits each. Paid people search your keyword and click your listing, so those clicks appear in Search Console, and they stop when the plan stops. Filter the keywords you order out of any compounding reading.

None of this buys a ranking. Google denies using click data as a direct ranking factor, while Microsoft says Bing "considers how users interact with search results" (How Bing delivers search results). A free bot that types searches into Google is sending what Google's spam policies call machine-generated traffic, "the practice of sending automated queries to Google" (spam policies). The public record on click data is in click data as a ranking signal.