Home / Blog / AI Search Tracking: A Two-Lane SEO Case Study
Singapore skyline at dusk with two highway lanes of light trails, white and orange, for the two-lane SEO case study
October 28, 2026

AI Search Tracking: A Two-Lane SEO Case Study

By [email protected]

Disclosure: We use ZeroRank daily. ZeroRank gives extra credits for published case studies, and this is our entry. We wrote it as we would for a client, limits included.

Open Google today and many searches start with an AI-written answer above the links. Ask ChatGPT, Gemini or Perplexity for “best movers in Singapore” and you get three or four names, not ten blue links. If you are not one of those names, ranking well on Google only helps so much.

We run SEO and digital marketing for about 10 retained clients in Singapore and Quebec. This is what we saw when we started tracking AI answers, starting with our own site and then six clients, how we split our work to limit risk, and where things stand so far.

Why AI answers change the risk in SEO

Quick Answer: AI answers now take attention at the top of search results, and they name a handful of businesses. Google rankings, clicks and keyword reports cannot tell you whether an AI recommends you. A separate check is needed, and no single channel should carry all of a client’s growth.

SEO was always bumpy: algorithm updates, layout changes, a competitor with a bigger budget. AI answers add a second source of volatility. They pull from pages with a direct answer and from sites they already trust, such as review pages, directories and news articles. Nobody outside the AI companies fully controls which businesses get named, and the same question can give a different answer tomorrow.

That uncertainty is why we do not tell clients “switch to AI SEO.” We tell them to protect what works and add a second route.

The two-lane approach

Lane 1: Google, defended. Rankings, technical fixes, on-page work, backlinks and blogs, tracked with Search Console, GA4, DataForSEO and SE Ranking. This lane carries the traffic and leads clients already get. We do not drop it.

Lane 2: AI answers, attacked. We check which customer questions AI answers without naming the client, then fix the page that should answer it and, where it is honest to do so, get listed on the sites AI already cites.

Why both:

  • If Google changes, a client who is also named in AI answers is less exposed.
  • If AI answers prove unreliable or fade, lane 1 still pays the bills.
  • Much of the work helps both. A clear answer near the top of a page, FAQ markup and sensible internal links help Google and give an AI something to quote.

We measured ourselves first

Before showing client numbers, here is our own. Rezult.co is small in search, and the numbers say so. In the 28 days to 21 October: 18 clicks from Google, 4.9K impressions, average position 36.7. Website visits were 290, and ChatGPT was the only AI source sending traffic: 52 sessions in 90 days, almost level with Google organic.

ZeroRank tracks 19 agencies for us. In the last 30 days Rezult was named 4 times (sentiment 62). First Page Digital had 56, Impossible Marketing and OOm 51 each, MediaPlus Digital 48. The sites AI cites most for these questions are mediaplus.com.sg, clutch.co, marketingagency.sg and terris.sg. We are on none of those lists.

What this told us, and what we are doing about it in November:

  • Our selling pages do not rank. The web design and SEO Singapore pages show thousands of impressions at position 50 or worse and no clicks. We are rebuilding them with the buyer’s answer on top, real price ranges, local proof and FAQ markup.
  • We send ourselves to the wrong visitors. Some traffic comes from guides on topics we do not sell. Visitors, not buyers.
  • We cannot show one enquiry. GA4 has no key events set up, so we are adding tracking for WhatsApp, the contact form and phone.
  • We are absent from the lists AI reads. We will go after a Clutch profile, a DesignRush review ask and one list placement, free routes first.
  • We have claims to fix before we preach clarity. The site says “9 years” in one place and “since 2010” in another. We are correcting them.

We are doing to ourselves what we do for clients, in the same order, and we will publish the re-run either way.

What we found for clients

We use ZeroRank to ask AI platforms a client’s tracked questions and record which brands get named. Here is the last 30 days for six clients, with the 90-day figure in brackets. A mention is one AI answer naming the brand. The sets of questions differ per client, so compare within a row, not across rows.

Client Own mentions, 30 days (90 days) Top tracked brand, 30 days (90 days) Own sentiment
Aspect Mover (movers, SG) 35 (47) Move Move Movers 139 (208) 71
EasyClean (cleaning, SG) 17 (21) Sureclean 203 (254) 69
Sea Friends (yacht charters, SG) 15 (15) White Sails 213 (290) 70
Maximal (e-bikes, SG) 27 (29) MOBOT 104 (135) 75
Metadesign Interior (office fit-out, SG) 21 (41) Conexus Studio 39 67
CleanAxisCo (cleaning, SG) 0 (0) Helpling 104 none

Three things we noticed:

  • Most clients are far behind the leader. Aspect Mover has 35 mentions to the top competitor’s 139, even though AI rates it fine when it finds it (average position 2.5, sentiment 71).
  • Being mentioned is not the same as being recommended. Ace Academic is named often (32 mentions) but averages position 7.2 with sentiment 52, the lowest of the group, so there is room to improve how AI describes the brand.
  • Some are at zero. CleanAxisCo, a newer site, has no mentions in 90 days. That is a clear baseline to build from.

In our experience, the gap is bigger than clients expect, and smaller than the AI hype suggests: it is fixable one question at a time, but it is not fixed in a week.

What a ZeroRank report looks like

This is the live dashboard for Aspect Mover, a Singapore moving company, covering the last 30 days.

ZeroRank AI visibility report for Aspect Mover: visibility by brand, sentiment and most cited sources

How to read it:

  • Visibility is the share of AI chats that mention a brand. Move Move Movers appears in 36%, CYC Movers 24%, Cube Movers 18%, and Aspect Mover 10% (rank 2.7, sentiment 74).
  • Sentiment for Aspect Mover: 81.8% positive, 18.2% neutral, 0% negative across 33 answers. AI speaks well of the brand when it finds it.
  • Top sources show what AI reads. storhub.com.sg leads at 15.1%, then aspectmover.sg itself at 12.4%, movings.sg at 11.8%, sunnymovers.sg and fixmove.sg. These are the sites worth being listed on or matching.
  • Recent chats show the real answers. In several, AI names CYC Movers, Cube Movers and Move Move Movers and does not mention Aspect Mover at all.

The tool is at zerorank.ai, where you can run a free check on your own domain.

What ZeroRank recommends, and what we do with it

ZeroRank groups its advice into Technical, On-Page and Off-Page. Our rules for each:

Technical. A readiness score for AI crawlers (robots rules, schema, llms.txt). Scores ran from 11 of 15 for Aspect Mover to 14 of 17 for Ace Academic, Phono and Yoga Lounge Sadhana. We fix real robots and schema problems first. ZeroRank itself says Google does not use llms.txt and AI crawlers rarely request it, so we treat it as optional.

On-Page. The most useful part. It lists saved AI answers where a competitor is named and the client is not. Examples from our October read: CleanAxisCo missing in 15 of 15 answers for “affordable house cleaning service in Singapore”; Metadesign Interior missing in 10 of 11 for “best office interior design firms in Singapore”; Maximal missing in 5 of 5 for “best foldable electric bike for Singapore commuters.” Usually the fix is an answer-first paragraph, FAQ markup and links on a page that already exists, not a new page.

Off-Page. Cards for community threads, profile pages and articles AI already cites, such as a Tripadvisor profile for Sea Friends. We approach real, relevant sites for inclusion. We skip “mention the brand” cards for Reddit, Quora and Facebook. That is undisclosed promotion, it gets accounts banned, and it damages trust. If an owner wants to answer a real question in a community, they do it themselves and say who they are.

Where to invest: the sources AI already trusts

Numbers show the gap. The more useful part is the map of where AI looks for answers. For each client, ZeroRank lists the domains it cites most, and they fall into a few types:

  • Listicles and local directories. For Aspect Mover, AI cites kakilist.com, misslobang.com, morebetter.sg and storhub.com.sg alongside competitor sites. CleanAxisCo and EasyClean see misslobang.com and superclean.sg. Getting included in a list that AI already reads is worth more than another post on our own site.
  • Review and travel platforms. Sea Friends sees tripadvisor.com.sg, klook.com and seeksophie.com. A complete Tripadvisor profile with real reviews came up as the top off-page task.
  • Authority sources. Cleaning questions cite mom.gov.sg and nea.gov.sg. Linking to those in our own answers, accurately, helps the page read as reliable.
  • Competitor sites. Often the loudest source is the competitor’s own page. That tells us which page type to match with a better, more direct answer.
  • Communities. Reddit, Quora and Facebook threads appear for many brands. We read them to learn what people actually ask, then decide with the owner whether an honest, disclosed answer is worth giving. We do not post for them.

The point is to put effort where AI already looks, ranked by how often it is cited, instead of guessing. Some of these are free to pursue (a profile, a correction, an honest answer), some are paid placements we would disclose, and some are not worth the time.

What has changed, and what comes next

What changed in how we work:

  • Every monthly report now has an AI-search section next to the Google one. Each recommendation links back to the saved AI answer behind it, so the client can check it.
  • Priorities are ranked by evidence. “Missing in 15 of 15 answers” is easier to rank than a general feeling that we should do more.
  • Client conversations moved from “are clicks down because SEO stopped working?” to “which question did we close this month?”

What has not changed yet:

  • AI visibility: too early to call. The page changes from October’s read are only starting to ship, and ZeroRank shows 0 growth so far. Our plan: ship the page changes and the first profile and listicle approaches, leave the question set alone for a month, re-run it, then adjust. We will share the result either way.
  • Reading cause and effect. AI answers vary from run to run, so we compare the same question set month on month and look at the pattern, not a single result.
  • Client retention. We believe a clearer report helps clients stay, and we will share real numbers once we have tracked them.

Good to know before you start

  • It plugs into our own tools. ZeroRank is an external platform, but we reach it through its API, an MCP and a small command-line tool we built. Rankings, sources and chats flow straight into the monthly report with no copy and paste.
  • Credits are easy to manage. Each AI answer uses credits, so tracking is limited by a monthly allowance (ours is 1,800 across all clients). We would like more, but it works: we track 10 to 15 prompts per brand each week, plan the month, and recharge when we need extra runs.
  • Support is fast. In our experience the team replies within a few hours to a day, and when we ask for a change or raise an issue it gets done. It feels close to custom-made for how an agency works.
  • Read counts as direction. “Missing in 8 of 11 answers” tells you where to act. Track the same questions month after month to see the trend.
  • You can start with no tool at all. Write down the 10 questions your customers ask, put each into ChatGPT, Gemini and Perplexity, and note whether your business is named and who is. It takes an hour, costs nothing, and shows you whether this matters for you. A tracker earns its place once you manage several brands or want monthly trend lines.

Frequently Asked Questions

Does AI search replace Google SEO?

No. Google still sends the traffic for most of our clients. AI answers are an additional route, which is why we run both lanes instead of switching.

How can I check if AI names my business?

Ask the questions your customers would ask in ChatGPT, Gemini and Perplexity, and note who is named. Repeat monthly. Answers change, so look at the pattern over several questions, not one result.

What actually improves your chance of being named?

In our work, a direct answer near the top of the page, FAQ markup, clear internal links, and listings on the review and article sites AI already cites. We have not seen evidence that llms.txt or tricks help.

Is it worth posting in Reddit or Facebook threads to get mentioned?

Only if the business owner answers honestly and says who they are. Paid or hidden brand mentions risk bans and reputation damage.

Conclusion

AI answers add risk to SEO, but the response does not need to be dramatic: keep the Google work that pays, add a second lane for AI answers, and measure both honestly. If you want to see where your business stands, start with the free hour of manual checks above. If you would like us to run the two lanes for you, talk to Rezult.


Reviewed and fact-checked by Bobby Dennie, Rezult, [email protected]. More client work in our case studies.

Need content made for you?

We write keyword-targeted articles that rank — while you focus on running your business.

Get Content Help
Contact us