A creative SEO strategy is one that produces an asset competitors cannot get by rewriting the same article in different words: an original data point, a free tool, an archive of real cases, an internal figure nobody publishes. Everything else is content production on topics someone else has already covered.
The distinction is not philosophical. Ahrefs analysed around 14 billion pages in its Content Explorer index and found that 96.55% get no organic traffic from Google at all (study published December 2023, global sample). The 3.45% that do get traffic are not made of better-written pages: they are made of pages that answer a question with something the others do not have.
What has changed: positions 11-20 are worth almost nothing today
Before listing the tactics, you need to understand why average content has stopped working, and it is worth doing with first-party numbers rather than borrowed benchmarks.
We exported from Google Search Console the 1,000 queries with the most impressions for an Italian outdoor equipment e-commerce site we work on, from 1 May to 20 August 2026, and grouped them by average position.
| Average position | Queries | Impressions | Clicks | CTR |
|---|---|---|---|---|
| 1-3 | 256 | 13,507 | 1,373 | 10.16% |
| 4-10 | 368 | 22,277 | 261 | 1.17% |
| 11-20 | 260 | 9,203 | 30 | 0.32% |
| 21-50 | 108 | 857 | 3 | 0.35% |
The jump between the first band and the second is almost ninefold. Between the first and the third it is thirty-two times. Of those 1,000 queries, 734 produced impressions and zero clicks: 16,237 impressions that did not turn into a single visitor.
The same pattern holds on our own site. Visilay's SEO service page, over the same period, collected 8,780 impressions and six clicks from an average position of 19.1. The title tag has little to do with it. For people searching, position 19 simply does not exist.
On top of this came the AI-generated summary. The Pew Research Center followed 900 US adults across 68,879 Google searches made in March 2025: when the results page contained an AI summary, users clicked a result in 8% of visits, against 15% on pages without a summary. The source cited inside the summary was clicked in 1% of cases. It is a US sample, but the mechanics are the same in the UK, where AI Overviews have appeared on searches since August 2024.
Ahrefs measured the same thing on 300,000 keywords, comparing March 2024 and March 2025: the CTR of position one falls by 34.5% when an AI Overview appears. Here too: desktop, United States.
The maths is simple. Content that replicates what is already on page one reaches, at best, positions 11-20. And positions 11-20 are worth a 0.32% CTR. The seven tactics below exist to avoid exactly that end point. If you are missing the level above, the one about priorities, start with SEO strategy and come back here afterwards.
Seven creative SEO strategies and what they really produce
1. Publish a data point nobody else has
It is the tactic with the best ratio of effort to result, and very few companies use it because it means looking at your own data instead of reading other people's.
Ahrefs documented the most cited case: a statistics page built specifically to attract links collected 36 editorial backlinks from 32 unique domains, with 515 outreach emails and a conversion rate of 5.71%. The page then reached position one for the keyword it had targeted.
The table above is our example: we have not seen any other agency publish the CTR curve by position band for an e-commerce site in 2026. Anyone who wants to quote that number has to quote us. You do not need a huge dataset, you need a dataset that only exists at your company: the logs from your management system, three years of quote requests, real delivery times, reasons for returns.
Working rule: the data must be published in HTML, inside a real table. If you put it inside an image, neither Google nor ChatGPT reads it, and the piece loses the one thing it existed for. The same goes for schema markup: marking up the data helps it get picked up.
2. Build a free tool instead of yet another article
A calculator, a configurator, a checker. Screaming Frog grew this way: the free crawler, up to 500 URLs, has brought in more links than any content about crawling. Buffer did the same with its salary calculator, powered by its own public salary data.
The workable version for a small or mid-sized business costs little: a quote calculator, a size configurator, a consumption simulator. These are pages that answer a commercial question with an interaction rather than a paragraph, and for that reason they hold up even when AI summarises the informational part.
3. Open up the numbers others keep closed
Transparency works as an SEO tactic for a mechanical reason: a number nobody publishes is news, and news gets linked. Buffer has published its revenue and salaries for years. Hardly any company in most sectors does the same, which is exactly why it works.
The realistic versions: a price list with real prices instead of "contact us", average response times, the return rate, how many jobs out of a hundred you turn down and why. Each of these is also an E-E-A-T signal, meaning experience demonstrated rather than claimed.
4. Do digital PR on a local data point, not a global trend
Digital PR always gets stuck at the same point: it pitches newsrooms a topic everyone is already covering. A local data point turns this around, because a local or trade publication has no way of getting hold of it by itself.
The starting material is public and free. The ONS article Artificial intelligence in UK businesses: 2023 to 2026, published in July 2026, reports that self-reported AI use among UK businesses with 10 or more employees rose from around 12% to around 35% since late 2023: roughly two in three still use none. Adoption ranges from almost three-fifths in information and communication to 13% in construction, and lack of expertise is one of the most cited barriers.
On these numbers you can build a piece a trade publication will happily run, as long as you cross them with something of your own: data from your region, your sector, your clients in aggregate form. The link comes from the second piece, not the first. We have written about how to judge the links you earn in the guides to backlinks and to link building.
5. Turn completed work into pages, one by one
In industrial B2B it is the tactic that pays best and the one that always gets postponed. Every installation, every order, every plant is a page with original photos, measurements, timings, constraints overcome. This is content no competitor can write and no LLM can generate, because it happened to you.
Macropix, a Milan-based manufacturer of custom LED screens, built its rankings partly this way. For the Italian keyword "monitor pubblicitario" (advertising monitor) it went from position 88 to position 2 between 2020 and 2025. In July 2026, on the ledwall cluster, the domain's share of voice was 25.55% against Amazon's 18.08% (source: Semrush, Google Italy). The full case, with the table of nine keywords and their positions, is in the article on SEO for manufacturing; other projects are among the work we have published.
The detail almost everyone gets wrong is the image file name. Uploading IMG_4471.jpg throws away the only original content the company really owns.
6. Answer the questions tools show as zero volume
Keyword research tools estimate volume on monthly averages and round down. A question searched thirty times a year shows up as zero, so nobody covers it. These are exactly the questions a customer asks on the phone before signing.
The place to find them is not a tool: it is the sales team's inbox, the call log, the support tickets. A page answering a zero-volume question brings ten visits a month and closes contracts, because the people searching for it are at the end of their journey. What counts is the average contract value, not the volume: a keyword with 90 monthly searches can be worth more than one with 9,000, and this is also why search intent weighs more than volume when choosing what to write.
7. Write to be cited, not only to rank
Semrush, studying over 500 digital marketing topics across traditional search and AI platforms, found two things that change how you set up a piece of content. First: the average value of a visitor arriving from AI search is 4.4 times that of a visitor from organic search, measured on conversion rates. Second: around 90% of the pages cited by ChatGPT Search rank beyond position 21 on Google for the corresponding queries.
The second number is the interesting one. Being cited and ranking are two different games, and the first does not require beating whoever is on page one. It requires a format a machine can extract: a blunt answer in the first two lines of the section, a table with the unit of measurement stated, the source and date next to the figure, proper names instead of categories.
It is worth saying how much this weighs today: Ahrefs measured across 60,000 sites that total traffic from LLMs stays under 1% of the total, against 41.35% from Google Search. It is a small channel with high conversion, not a replacement. How to work on it is covered in the guide to generative engine optimisation and the one on how to appear in ChatGPT answers.
What they cost and what they return
The seven tactics are not alternatives to one another, but their costs and timescales are very different. This is the grid we use to assess them before proposing them.
| Tactic | Initial effort | Time to first result | What it produces | Metric |
|---|---|---|---|---|
| Original data | Medium | 1-3 months | Backlinks and citations | Referring domains |
| Free tool | High | 3-6 months | Backlinks and leads | Referring domains, sign-ups |
| Public internal numbers | Low | 1-2 months | Mentions and trust | Mentions, contact rate |
| Digital PR on local data | Medium | 2-4 months | Editorial backlinks | Referring domains |
| Project pages | Low and recurring | 3-9 months | Commercial rankings | Clicks and enquiries |
| Zero-volume questions | Low | 1-2 months | Qualified enquiries | Conversions per page |
| Citable format | Low | Immediate | Citations in LLMs | Brand mentions |
There is only one selection criterion: which of these seven things can you do that whoever is ahead of you in the SERP cannot. A manufacturer with twenty years of installations starts from the fifth. A software company with aggregated usage data starts from the first. A company with neither starts from the sixth, because it has customer questions regardless.
Three mistakes that make a creative tactic useless
- Publishing the data inside an image. A table inside a JPG is not read by crawlers or by models. It is the quickest way to do the work and throw it away.
- Not stating market and period. A number without "US sample, March 2025" next to it will not be picked up by anyone with a minimum of editorial control, and will be challenged in the first comment.
- Stopping at publication. The 36 citations of the Ahrefs page came after 515 emails. Original content does not spread by itself: someone has to know it exists.
A fourth, less visible mistake: measuring a creative tactic with the metrics of ordinary content. A project page that brings 40 visits a month and two quote requests should be judged on the requests, not the visits. We have written about which numbers to look at in the guide to SEO KPIs.
The SEO strategy decides what to work on and in what order: goals, priorities, site architecture. Creative SEO strategies are the tactics that build an advantage competitors cannot copy, such as an original data point or a tool. The first is the level of choices, the second is the level of execution.
They are most useful to smaller players, because they are the only way to overtake domains with more authority without spending as much as they do. Of the seven tactics, three (public internal numbers, zero-volume questions, citable format) have a low initial cost and do not need a team.
One to three months for an original data point that attracts citations, three to nine for project pages that need to gain positions on commercial keywords. Timescales depend on the authority of the starting domain: on a new site, double them.
It is worth investing in content that brings something of your own. According to the Pew Research Center, in the United States an AI summary reduces clicks from 15% to 8% of visits, and the effect hits generic informational pages hardest. Content with your own data, project pages and answers to specific commercial questions is still what generates enquiries.
With the metric that matches what it produces: referring domains for original data and digital PR, enquiries for project pages and zero-volume questions, brand mentions for the citable format. Visits alone do not tell you whether the tactic has worked.
Something the industry rarely says: the creative part of these tactics is not in the idea, it is in deciding which figure to bring out. Everyone has the idea "let's publish a study". Choosing to publish your return rate, or the CTR curve by position band, or how many jobs you turn down and why, means accepting that the number becomes public and open to challenge. That is where almost all creative work stalls, not at the writing stage. If you want to find out which of these numbers you already have without knowing it, let's talk.
