# Read the growth curve

> What the clicks and impressions lines in Search Console are telling you, for a new site and an established one, how to find the cause of a jump, and when a setup is worth copying.

Module 12 · Track 2 · SEO and AI search · Foundation · 10 min · promotedby.ai Academy, version 2026-11-05 · https://promotedby.ai/academy/read-the-growth-curve

## The short version

- Read clicks and impressions together. Their relationship says more than either line alone.
- On a new site impressions usually move first, and clicks follow; judge weeks, not days.
- Average position is averaged over every impression, so a flood of new low-ranked searches can make it worse while nothing on the site got worse.
- To explain a step, compare the period after it with the period before, sort the Pages and Queries tabs by the clicks difference, then check your change log, Google's announced updates and Search Console's data anomalies page.
- Steady slopes come from compounding; steps come from events. Healthy sites usually show both.
- When a setup works, copy the setup (page types, structure, internal links, promotion), and research each new niche on its own.

## After this module you can

- explain how Search Console calculates CTR and average position, and why average position can get worse while a site grows
- read the two lines together and tell what each common pattern means
- use Compare dates to find which pages and queries caused a step up or down
- tell compounding growth from one-off events, and decide what in a working setup is worth repeating

The Performance chart in Google Search Console has two main lines: clicks and impressions. Most people glance at clicks and move on. The relationship between the two lines tells you much more. It shows whether Google is putting your site in front of more searches, and whether people pick your result once they see it. Read together, the lines also help you work out whether a change came from something you did.

## How Search Console counts what you see

Four numbers sit above the chart. An *impression* is one time a link to your site appeared in Google's results. For most results it counts even if the person never scrolled down to it, as long as it was on the page they were looking at. Inside a carousel the item has to be scrolled into view. A *click* is a visit from a result to your site. Click-through rate (*CTR*) is clicks divided by impressions. *Average position* is where your result appeared, 1 being the top.

Two details matter for reading the chart.

First, the chart counts your site, not your pages. If three of your pages appear for one search, the chart records one impression, at the position of your highest result. The Pages table counts each page separately. That is why the chart and table totals often differ, and why CTR and average position usually look better in the chart than in the Pages table.

Second, average position is an average over impressions. Google's help says it "averages the position value for all impressions." Each time your site is shown, its topmost position that time goes into the average. That leads to a result that looks wrong at first.

> **Example: Average position got worse while the site grew**
>
> Say a page ranks at position 3 for one query and gets 1,000 impressions a week. Then you publish some new pages, and Google starts showing your site for many new long-tail searches (longer, more specific queries), mostly around position 35.
>
> | Measure | Week 1 | Week 2 |
> |---|---|---|
> | Original query, position 3 | 1,000 impressions, 80 clicks | 1,000 impressions, 80 clicks |
> | New searches, around position 35 | none | 3,000 impressions, 6 clicks |
> | Total impressions | 1,000 | 4,000 |
> | Total clicks | 80 | 86 |
> | CTR | 8.0% | 2.15% |
> | Average position | 3.0 | 27.0 |
>
> Week 2's average is (1,000 × 3 + 3,000 × 35) ÷ 4,000 = 27. Nothing that ranked before got worse. Clicks went up. Yet average position fell from 3 to 27 and CTR fell by almost three quarters.

So when impressions grow fast, a worse average position and a lower CTR are usually fine. To check whether existing rankings really slipped, look at a fixed set of queries: filter to your top queries from the earlier period and compare their positions, or check individual pages in the Pages tab. Google's own advice is to watch trends in impressions and clicks more than position alone.

## A new site: watch impressions first

Here is a new site's first 28 days.

*Figure: A new site, first 28 days (redrawn from Search Console). Search Console Performance chart for a new site over 28 days: 287 total clicks, 14.9K impressions, 1.9% average CTR, average position 12. Impressions climb steadily from zero to about 1,000 a day. Clicks follow the same direction but jump around, from 0 to about 24 a day, with a dip to 10 near the end.*

Over the four weeks it had 287 clicks, 14,900 impressions, a 1.9% click-through rate and an average position of 12. Impressions climbed steadily from zero. Clicks followed in the same direction but jumped around: 22 one day, 10 the next, 22 again.

Here is how to read it.

1. On a new site, impressions usually move first. The site starts appearing for more searches, mostly below the top results, and clicks follow as some of those positions improve. A rising impressions line is the early sign that pages are being found and matched to searches, weeks before clicks look like much.
2. Day-to-day noise in clicks means little at this volume. When a whole day is 10 to 25 clicks, a swing of 12 is normal. Switch the chart to weekly granularity (Search Console weeks run Sunday to Saturday) and judge week against week.
3. An average position of 12 means many impressions come from just below the first ten results. Treat it as an average, not a place on a page. Those searches are within reach, and this is where [improving pages that already rank](https://promotedby.ai/academy/pages-from-search-console) pays off fastest.

On a new site the Queries tab is worth reading in full every week. It shows what Google thinks the site is about, which is not always what you intended.

## An established site: flat, then a step

Here is a different site over three months.

*Figure: An established site, three months (redrawn from Search Console). Search Console Performance chart for an established site over three months: 7.82K clicks, 388K impressions, 2% average CTR, average position 9.9. Clicks stay flat around 75 to 90 a day for about eleven weeks, then step up to about 190 a day in the final week, with impressions rising from about 4K to about 9K a day alongside.*

It held at about 75 clicks a day for eleven weeks, then rose to about 190 a day within a week, with impressions rising alongside. Over the quarter that came to 7,820 clicks and 388,000 impressions, a 2% CTR and an average position of 9.9.

A step like that has a cause, and you can usually find it in about 20 minutes:

1. In the Performance report, click the date filter and open the **Compare** tab. Set the week after the step against the week before it (or the two weeks after against the two weeks before, if you have them).
2. The table now shows both periods and a **Difference** column. On the **Queries** tab, sort by the clicks difference to see which searches gained most.
3. Do the same on the **Pages** tab. Does one page explain most of the gain, a group of pages, or the whole site?
4. Check the **Countries**, **Devices** and **Search appearance** tabs too. Google's traffic guide suggests clicking through all tabs, because a change sometimes sits in one country or one result type.
5. Match the dates. Look at your own change log and the annotations on the chart, then at Google's list of ranking updates on the [Search Status Dashboard](https://status.search.google.com/products/rGHU1u87FJnkP6W2GwMi/history), then at Search Console's [data anomalies page](https://support.google.com/webmasters/answer/6211453), which lists logging errors that can create jumps that never happened.

When you compare ranges longer than a couple of weeks, use weekly granularity so you aren't comparing a Monday with a Saturday. Comparisons don't show annotations, so note your dates before you switch to comparison mode. And when you find the cause, annotate it: right-click the chart on the date and add a note of up to 120 characters.

> **Watch out: Check for a data problem before you celebrate or panic**
>
> Google's data anomalies page records a logging error that affected impressions in the Performance report from May 13, 2025 until April 27, 2026. Only impressions and the metrics built on them (CTR and average position) were affected, not clicks. If you compare a recent period with one inside that window, part of any impressions change may be the fix to the logging, not your site.

## What the two lines are telling you

Each pattern below has more than one possible cause, so treat the "what it usually means" column as a first guess to confirm with the Compare steps above.

| Impressions | Clicks | What it usually means | What to do |
|---|---|---|---|
| Up | Up | More searches show you, or you moved higher, or more people search for your topic | Find the pages and queries behind it; do more of what caused it |
| Up | Flat | You appear for more searches, but low on the page | Improve the pages getting the new impressions ([find them here](https://promotedby.ai/academy/pages-from-search-console)) |
| Flat | Up | More people click what was already shown: better titles, snippets or rich results, or a move near the top | Note what changed and apply it to similar pages |
| Flat | Down | A small slip near the top, or a result that looks less useful than its neighbors | Check positions of top queries; compare your title and snippet with the results around them |
| Down sharply | Down sharply | An algorithmic update, a technical problem, a security or spam issue, or a reporting glitch | Check data anomalies, the Page indexing and Manual actions reports, and the ranking updates list |
| Down gradually | Down gradually | Seasonality, changing interest, or slow technical decay | Set the date to the last 16 months; check the queries in Google Trends |
| Flat | Flat, while you keep publishing | New pages aren't reaching new searches | Go back to [finding the pages to build](https://promotedby.ai/academy/pages-from-search-console) |

Two rows deserve a closer look. The "flat impressions, falling clicks" pattern is one Google calls out in its guide to traffic drops: impressions holding steady while clicks fall can mean your title and snippet aren't the best they could be, "or perhaps other sites had a more appealing rich result." A small drop in position, say from 2 to 4, also produces it. Google recommends against radical changes to a page that was performing well, since small moves can reverse without you doing anything.

For gradual declines, the 16-month view is the first thing to check. Google's guide recommends it to "make sure it's not a drop that happens every year due to a festivity or a trend." A gift shop that falls every January is not broken.

## Compounding or effort?

A smooth, steady slope is compounding: pages getting older, collecting links and earning trust, and the site appearing for more searches over time. Steps are events: something you published, something that linked to you or sent you people, or something Google changed. Most healthy sites show both, a rising base with steps on top.

The two call for different responses. Compounding rewards patience and consistency, so the risk is interrupting it with big, untested changes to pages that already work. Steps reward investigation, because a step you can explain is one you might repeat. If the line is flat and you are still publishing, the new pages are not reaching new searches. Go back to [finding the pages to build](https://promotedby.ai/academy/pages-from-search-console) before writing more of the same.

Promotion shows up in this chart too, indirectly. When a placement reaches the right people, some of them search for the product by name afterwards. Search Console's **Branded** and **Non-branded** filter (under the query filter, for sites with enough impressions) separates those searches. If you log your placements by date, you can see whether branded impressions rose in the weeks after. A link from a page nobody visits brings nobody to search for you.

## When a setup works, copy the setup

Daniel (@seobydan), an SEO who posts on X, described a niche site only as being in "consumer electronics (niched)": one narrow corner of a big market. Asked whether the approach would carry over to other niches, Daniel replied:

> "Different niches, but very similar setup in general. It's rather scalable."

What is worth copying from a site that works is the setup: the types of pages, how the site is structured, how pages link to each other, and how new pages get promoted. The niche itself is not what you copy. Before you repeat a setup somewhere new, research that niche on its own terms: who searches, what ranks now, and whether you can do better than that. A setup that worked because nobody else had covered a corner well will not work in a corner that is already covered well.

Write the setup down as you find it. A short list like "comparison page per competitor, one how-to per task, each linked from a hub page, promoted in two communities" is something you or an agent can repeat. "We did good SEO" is not.

> **For AI agents: Watching the growth curve**
>
> - Weekly: request daily `clicks`, `impressions`, `ctr`, `position` with `dimensions: ["date"]` for the last 16 months via `searchAnalytics.query` (`dataState: "final"`). Flag a step when the latest 7-day average of clicks differs from the prior 28-day average by more than you'd expect from the series' usual weekly variation.
> - To explain a step: two requests with `dimensions: ["page"]` and `["query"]`, one for 7 days after and one for 7 days before; rank rows by clicks difference; report the top 10 of each.
> - Before naming a cause, check the operator's change log and annotations, the Search Status Dashboard ranking updates, and the Search Console data anomalies page for the same dates. If none match, say the cause is unknown.
> - Report position changes for a fixed set of queries, never the site-wide average alone.
> - For products promoted through promotedby.ai, list accepted placements by date and report branded impressions before and after, if the property has the branded filter.

Before you write up any change in the chart, run through this:

- [ ] Granularity set to weekly for anything longer than two weeks
- [ ] Compared with the period before, and with the same period last year if you have 16 months of data
- [ ] Sorted Pages and Queries by the clicks difference
- [ ] Checked your change log, the ranking updates list and the data anomalies page for the same dates
- [ ] Checked position for a fixed set of queries, not only the site-wide average
- [ ] Added an annotation on the date once you know the cause

## Try it

> **Try it: Read your own curve (20 minutes)**
>
> Take 20 minutes with your own Search Console property. Open the Performance report for the last 3 months and switch on all four metrics. Set the granularity to weekly and find the pattern in the table above that best matches your two lines. Then open the date filter, choose **Compare**, and compare the last 28 days with the previous period. Sort the Pages tab by the clicks difference and write down the three pages that changed most.
>
> You're done when you can write two sentences: which pattern your site shows, and which pages or queries account for most of the latest change. Add an annotation on the chart for the most likely cause, or for "cause unknown" with the date, so next week's review starts from there.

## Check yourself

1. Over a month, impressions doubled and average position went from 8 to 15, while clicks rose slightly. What is the most likely reading?
   - a) The site's rankings collapsed
   - b) The site started appearing for many new searches at low positions, which pulls the average down
   - c) Google is counting impressions wrongly
   - Answer: b) The site started appearing for many new searches at low positions, which pulls the average down. Average position is averaged across all impressions. New low-ranked searches add many high position numbers without anything that ranked before getting worse.

2. Impressions are flat, but clicks dropped 25% over two weeks. Where do you look first?
   - a) The positions of your top queries and what your results look like next to the others on the page
   - b) The number of pages in your sitemap
   - c) Your server logs for crawl errors
   - Answer: a) The positions of your top queries and what your results look like next to the others on the page. Google's traffic drop guide says flat impressions with falling clicks can come from a small slip near the top, or from a title and snippet that look less useful than other results.

3. Clicks jumped from about 75 to about 190 a day within a week. What is the first step to find the cause?
   - a) Wait three months to see whether it holds
   - b) Compare the week after with the week before, then sort the Pages and Queries tabs by the clicks difference
   - c) Look at the average position for the whole quarter
   - Answer: b) Compare the week after with the week before, then sort the Pages and Queries tabs by the clicks difference. The Difference column shows which pages and queries account for the change. Then you match those to your change log and to Google's announced updates.

4. A new site shows 22 clicks one day, 10 the next and 22 the day after. What should you conclude?
   - a) Something broke on the low day
   - b) Nothing yet; at this volume daily numbers are noisy, so compare whole weeks
   - c) Google is testing the site and will stop soon
   - Answer: b) Nothing yet; at this volume daily numbers are noisy, so compare whole weeks. With a few dozen clicks a day, random variation is large. Weekly granularity smooths it out.

## Sources

- [Search Console Help, What are impressions, position, and clicks?](https://support.google.com/webmasters/answer/7042828)
- [Search Console Help, Performance report (Search results), overview and basic setup](https://support.google.com/webmasters/answer/7576553)
- [Search Console Help, Performance report, about the data (aggregation by property vs page)](https://support.google.com/webmasters/answer/17011364)
- [Search Console Help, Performance report, advanced filtering and comparison](https://support.google.com/webmasters/answer/17011165)
- [Search Console Help, Performance report, dimensions and data groupings](https://support.google.com/webmasters/answer/17011259)
- [Search Console Help, Performance report, common tasks and use cases](https://support.google.com/webmasters/answer/17010961)
- [Search Console Help, Data anomalies in Search Console](https://support.google.com/webmasters/answer/6211453)
- [Search Console Help, Search Console annotations](https://support.google.com/webmasters/answer/16530728)
- [Google Search Central, Debugging drops in Google Search traffic](https://developers.google.com/search/docs/monitor-debug/debugging-search-traffic-drops)
- [Google Search Status Dashboard, ranking updates history](https://status.search.google.com/products/rGHU1u87FJnkP6W2GwMi/history)
- [Daniel (@seobydan) on X, on scaling a niche-site setup](https://x.com/seobydan)