SaaS · UK · Direct engagement
B2B SaaS:
from invisible to answer-engine cited.
Topical authority plus answer-engine optimisation (AEO): structured data, answer-first content, entity cleanup. Organic up 127% in five months, with citations in AI Overviews for money queries.
A UK B2B SaaS platform grew organic traffic +127% in 5 months. A direct engagement. We built authority on the topic from a standing start and added the AI-search work: structured data, consistent naming and pages that answer first. Now Google AI Overviews cite the product on commercial searches, as well as ranking it in the normal results.
A UK B2B SaaS company could not be found in normal search, while its buyers had started asking AI tools about its category.
- Built a map of the topics buyers ask about, and pages that answer first
- Added structured data across the site and made the product's name consistent
- Checked that AI crawlers could reach the site, and built pages around questions
Result: +127% organic traffic in five months, with citations in AI Overviews for buying searches
Organic traffic,
before and after.
+127%
Organic traffic · 5 months
- AI search
- Cited in Google's AI Overviews for buying searches
Search visibility,
before and after.
| Before | After | |
|---|---|---|
| Organic traffic | Flat, before the work began | +127% growth (5 months) |
| AI-search visibility | Invisible to answer engines | Cited in AI Overviews for buying searches |
| How machines describe the product | A different product story in different places | One consistent story, with structured data across the site |
- Organic trafficBeforeFlat, before the work beganAfter+127% growth (5 months)
- AI-search visibilityBeforeInvisible to answer enginesAfterCited in AI Overviews for buying searches
- How machines describe the productBeforeA different product story in different placesAfterOne consistent story, with structured data across the site
Invisible where
buyers actually look.
A UK B2B SaaS company came to us directly. It had a sales pipeline problem that looked like a marketing one. Nearly every good lead came from paid ads or outbound sales. Organic search, the channel that keeps growing without a monthly bill, brought in almost nothing.
The deeper issue was being seen where buyers had moved to. People researching the category asked ChatGPT and read Google AI Overviews. The product appeared in neither, nor in the normal results below them. The site had thin content and no clear structure of topics. It also named the product in so many ways that answer engines could not say for sure what it was.
The brief we agreed was specific. Make organic search a real source of sales leads, and make the brand something answer engines cite. Judge it by traffic and sales pipeline, not by rankings for their own sake.
Build authority first,
then earn citations.
A topic map from a standing start
We grouped the questions buyers ask into topics the site could really own, in order of commercial value. Every piece we published traced back to this map. No stray blog posts, and no content for its own sake.
Pages that answer first
Each page opens by answering its own question, in words an engine can quote as they stand. The detail follows underneath. The same layout serves a person skimming and an AI deciding what to cite.
Schema and consistent naming
Structured data across the site, and one story about what the product is, told the same way everywhere machines look. When machines are unsure what a product is, it quietly drops out of AI search. This unglamorous work fixes that.
Answer-engine (AEO) set-up
We checked that the engines that matter can crawl the site, used question headings and wrote passages an engine can quote as they stand. This is the method from our AEO and GEO services (answer-engine and generative-engine optimisation). It was built into the SEO work from the start, not added later.
Reports tied to sales
A weekly review in Prism linked rankings to visits to sign-ups. AI citations were tracked beside normal rankings. So progress was judged on the numbers the board cares about.
The decisionsThree choices, and what we turned down.Open to read
AI search is new, so most of these decisions had no proven method to follow. These are the choices that mattered.
Chase high-volume keywords, or build authority on a topic first?
Authority first. Depth on a tight set of topics before breadth across a wider one.
Going straight for the highest-volume terms. A keyword tool suggests that, and most plans do it. But a site with no authority in the category competes with sites that have it. The effort produces pages that sit on page four for ever. Depth in a narrower area can be won. Breadth in a crowded one cannot.
Aim for normal rankings or for AI citations?
Both, from one structure. Pages that answer first, schema and clear naming help normal rankings and AI answers at the same time.
Treating AEO as a separate job for later. Many teams do, and it doubles the work. Adding answer-first structure to content written another way means rewriting the content. Done together, it is one job. Done apart, it is two.
Publish faster with lighter pieces, or slower with deeper ones?
Slower and deeper. Fewer pages, each complete on its subject.
A busier publishing calendar. Volume is easier to sell and easier to report on, because there is always something new. But it produces the thin, me-too content that AI answers now absorb and answer directly. So the traffic it earns is the traffic most at risk.
What we delivered,
item by item.
The AI citation log,
how we report AI mentions.
Before we rewrite anything, we agree with you a set of questions to track. Each week we ask AI tools the same questions and record which page each one cites. The questions and pages are hidden here.
| Question | AI tool | Cited page | This week |
|---|---|---|---|
| AI Overviews | Cited | ||
| AI Overviews | Cited | ||
| AI Overviews | Cited | ||
| AI Overviews | Not yet | ||
| AI Overviews | Cited |
The hard partsWhat this work really took.Open to read
Making the product's name consistent took more work than it sounds. It was also less visible than anything else we did. Machines decide what an organisation is by checking that it is described the same way everywhere they read. That means the site, the structured data, outside profiles, and how the company describes itself in different places. Where those disagree, machines stay unsure, and the website itself shows no sign of it. Finding and fixing it is slow, and it looks dull in a monthly report.
The delay before results show was the harder problem to manage. A B2B SaaS sales cycle is long. Content that ranks in month three may not lead to a closed deal until month eight or later. In that first stretch, an online-shop style revenue report makes the work look like it is failing. In fact it is working exactly as planned. Holding a plan through that time means agreeing up front what will show, and when.
Nobody can control AI citations the way they can work on a ranking. You can make content easy to quote, keep naming consistent and answer questions fully. But the choice of what to cite belongs to a system nobody outside it can see into. The honest way to say it is this: the work greatly raises the chance of being cited. It cannot deliver citations on a schedule.
When this approach
fits, and when it doesn't.
Many agencies oversell AI-search work right now. Here is where this approach fits.
This approach fits if
- You sell something people think hard about, and several people sign off on it. It is not an impulse buy.
- Buyers already search for your category in words they know. You are competing for known terms, not inventing one.
- You can wait two quarters before judging the work on sales pipeline rather than on rankings.
- The business has real expertise that content can draw on.
It probably doesn't fit if
- You need leads this month. Nothing here is fast, and paid ads are the right answer to an urgent gap in leads.
- Your product is so new that nobody searches for the category yet. Then you need to create demand first. Search follows demand; it does not create it.
- Nobody in the business can spare time to check content for accuracy. Content this deep needs that review. Without it, you get the thin content this work set out to avoid.
The services behind this result
Every search project we run works this way: classic SEO with AI search built in.
Practitioner notes
on this work.
Want a result like this
on your own numbers?
Send the brief and an itemised, priced proposal comes back. Or start with the free website review: we measure one page of your website and a person reviews the report.
Reply within four business hours (typical) · No retainer required · NDA standard
Last updated