Brand as entity · Incumbent capture

Getting recommended by an AI is a different job from getting quoted

This platform puts 87% of its AI appearances on a single URL and gets cited on an average of 2 pages a day, where our city-guide property gets 35. Same metric, same dashboard, opposite shape — because “recommend me a site” is answered by naming a brand, not by quoting a page.

16 months, ending 12 August 2026
Profiles anonymised throughout

87%

of AI appearances on one URL

2

pages cited a day, against 35 on a facts site

~20%

of Bing impressions from incumbent brand queries

20.4%

of clicks from outside Sri Lanka, 9+ countries

8.25K

Organic clicks from 286K impressions

Property

MatchMaker.lk · owned and operated

Type

Sri Lankan matrimony platform, user profiles

Incumbent

Newspaper classified sections, for generations

Window

16 months, ending 12 August 2026

Site size

930 pages ranking in Google

Privacy

Every individual profile anonymised on this page

Before the data

Why names are blanked out on this page

MatchMaker.lk is a matrimony platform. Its highest-traffic individual pages are real people's proposal profiles, and its single top query is somebody searching for a named private individual. Those pages are public on the site — but reproducing a person's name and marital status in an agency's marketing material is a different act from hosting their profile for the purpose they consented to.

So every profile URL and personal-name query below is anonymised. The numbers are untouched and the analysis loses nothing. We would rather explain this than have you notice the gaps and wonder.

The finding

Two pages cited a day, and that is the strategy working

Across four owned properties we now have the same metric from the same Bing dashboard — average cited pages per day — and the spread is enormous. The city guide gets 35. The pilgrim guide gets 11. This one gets 2.

The instinct is to read 2 as a failure. It is not. It is what happens when the questions your market asks have no factual answer to quote.

Nobody asks an assistant for the phone number on a particular person's proposal profile. They ask “what is a good matrimony site in Sri Lanka” — and there is no fact that answers that. The model has to name an organisation. So the citations collapse onto the front door: 87% of this site's AI appearances, and 77% of its organic clicks, land on the homepage.

Concentration is not a symptom of a thin site. It is the correct shape for a market whose queries want a recommendation rather than an answer — and it means the work is entity clarity, not extractable claims.

Average pages cited per day, across four owned properties

The same “Avg. Cited Pages” figure from each property's Bing AI Performance dashboard. The label after each name is what kind of question that property answers — which is the entire explanation for the spread.

CoimbatoreJunction · facts 35
VelankanniShrine · facts + places 11
MatchMaker.lk · recommendations 2

These are daily averages, so they are deliberately not totalled — a column of averages sums to nothing meaningful. Source: Bing Webmaster Tools AI Performance dashboards, 6-month windows, exported 12 August 2026. The fourth portfolio property, SEOTamil.com, reports per-page citations rather than a daily average and so is not directly comparable on this metric.

Two questions, two answer shapes, two different jobs

Both columns describe AI search. They need almost nothing in common from the page, which is why one AI-visibility strategy applied to a whole site tends to half-work.

A fact question

“What does a Red Taxi cost per km?”

One correct answer exists.

The AI quotes a page

It lifts the number and shows it. The source is credited underneath.

So you optimise for extractability

Real tables, question headings, the claim near the top. Citations spread across many deep pages — 35 a day on our city guide.

Click often unnecessary

A recommendation question

“Best matrimony site in Sri Lanka?”

No fact answers this. It needs a judgement.

The AI names a brand

It has to be confident enough about an organisation to put its name in a recommendation.

So you optimise for entity clarity

One unambiguous, well-described, well-cited organisation. Citations concentrate on the front door — 87% of appearances on one URL, 2 pages a day.

Click is the whole point
Both columns are measured on the same metric from the same Bing dashboard — “Avg. Cited Pages” — on two properties we own.

Evidence · AI search

5.61K appearances, 87% on one URL

Google Search Console generative AI features report for matchmaker.lk showing 5.61K impressions climbing steadily

Google · 5.61K AI feature appearances

Daily volume climbs from roughly 40 in mid-May 2026 to about 160 by early August — a 4x rise. Top page: the homepage at 4,895, then /muslim-matrimony/ at 204. The drop-off after first place is the concentration this study is about.

Google Search ConsoleGenerative AI (beta)Data from 18 May 2026
Bing Webmaster Tools AI Performance report for matchmaker.lk showing 3.8K citations and 2 average cited pages

Bing · 3.8K citations, 2 pages a day

Citation sources: Microsoft Copilots and Partners. The “Avg. Cited Pages” tile reads 2 — the number to compare against 35 on our city guide and 11 on the pilgrim guide. Same report, same window length, three different market shapes.

Bing Webmaster ToolsAI Performance (beta)6 months

Note the second-place page: /muslim-matrimony/ at 204 appearances. Community segment pages are the one place where a recommendation query does get more specific — “matrimony site for Muslims in Sri Lanka” — and it is the only page other than the homepage with meaningful AI presence. That is a signal about where the next pages should go.

The second play

Taking demand that is still searching for a newspaper

In Sri Lanka the default place to find a marriage proposal was, for generations, the classifieds in the Sunday Observer or Silumina. A lot of demand is still phrased that way.

Query Whose brand Impressions Clicks CTR Avg pos
srilankan matrimony Rival platform 1,017 9 0.88% 4.85
sunday observer marriage proposals Newspaper classifieds 942 10 1.06% 6.47
sri lanka matrimony Rival platform 660 17 2.58% 5.29
lanka matrimony Rival platform 562 5 0.89% 5.37
sunday observer marriage proposal Newspaper classifieds 540 9 1.67% 6.96
lanka matrimony free marriage proposals Rival platform 530 40 7.55% 3.88
hitad marriage proposals Classifieds site 330 12 3.64% 5.27
mangala yojana Sinhala category term 299 58 19.40% 4.63
silumina marriage proposals Newspaper classifieds 240 3 1.25% 6.13

~5,120

Bing impressions from queries naming an incumbent — roughly 20% of the property's total

19.40%

CTR on “mangala yojana”, the Sinhala category term — the best-converting category query on the site

The person typing “sunday observer marriage proposals” wants marriage proposals. The newspaper is just the noun they learned for the category. Answering that intent honestly is legitimate competitive SEO — and it is worth saying plainly that the pages do not pretend to be the Sunday Observer, because that line matters.

The vernacular finding is the more useful one for anyone operating outside English. “Mangala yojana” converts at 19.40%, better than every English category term on the property, from position 4.63. A keyword tool set to English would never have surfaced it.

The SaaS translation is exact: your competitor's brand name is a keyword. “[Competitor] alternative” and “[legacy tool] replacement” pages are this same play, and in most categories they are the highest-intent traffic available.

The constraints

Three languages, and other people’s private lives

01

The incumbent is a newspaper, not a website

For generations the default place to find a marriage proposal in Sri Lanka was the Sunday Observer or Silumina classifieds. People still search for those sections by name. You cannot outrank a brand nobody is comparing you to — you have to be the result that appears when they look for it.

02

Three languages, three scripts, one market

Sinhala, Tamil and English, plus transliteration. “Mangala yojana” and “marriage proposals” are the same intent in different alphabets, and a site that only covers English gives away the majority of the market.

03

The pages are other people’s private lives

Every profile is a real person looking for a spouse. That constrains what can be indexed, what can be shown in a snippet and — as this page demonstrates — what can be reproduced in a case study afterwards.

What was actually built

Six decisions, and what each one bought

01

Rank for the incumbents by name

Pages that legitimately answer “Sunday Observer marriage proposals” and “hitad marriage proposals” — because the person typing that wants proposals, not a newspaper. Around 20% of Bing impressions now come from queries aimed at somebody else's brand.

02

Cover the category in every language it is asked in

“Mangala yojana” in Sinhala converts at 19.4%, better than any English category term on the property. Transliterated and vernacular terms are not a nice-to-have in this market, they are where the intent is.

03

Segment pages by the way people actually filter

Christian matrimony, Muslim matrimony, grooms in Jaffna. Community and location are how this decision genuinely gets made, so they get real pages rather than search parameters.

04

Make the homepage the entity

Because the query behind most of this demand is “recommend me a site”, the homepage has to be an unambiguous, well-described entity. It now carries 87% of the site's AI appearances and 77% of its organic clicks.

05

Build for the diaspora, not just the island

One in five clicks comes from outside Sri Lanka, across at least nine countries. Japan and Qatar convert above 5% — better than the home market — because a migrant worker planning a marriage has a narrow window and no patience for a vague page.

06

Index the category, protect the people

Individual profiles are handled conservatively, and this case study anonymises every one of them. On a platform holding sensitive personal data, restraint is part of the product, not a limitation on the marketing.

Evidence · Search Console

8.25K clicks, and 54 queries at position #1

A 2.9% site-wide CTR at an average position of 10.4 hides two facts at once: a homepage that converts from the bottom of page one, and 54 outright #1 positions.

Google Search Console performance report for matchmaker.lk showing 8.25K clicks and 286K impressions

Search Console · 16-month performance

8.25K clicks, 286K impressions, 2.9% CTR, average position 10.4. Impressions turn upward again from around May 2026 after a long flat stretch.

Google Search ConsoleSearch type: Web16 months
Google Search Console query report for matchmaker.lk showing brand and category matrimony queries

Search Console · brand and category queries

“matchmaker.lk” converts at 60.33% from position 1.09, “matchmaking sri lanka” at 27.94% and “match maker” at 20.40%. The top row in the interface is a personal-name query, anonymised in our table below.

Google Search ConsoleQueries tab16 months

Where 1,000 the top 1,000 exportable queries actually sit

Recomputed from the raw query export. 54 queries average position #1 — the second-strongest #1 count in the portfolio — while 469 sit beyond page one. A single site-wide average position of 10.4 describes neither group. Search Console caps its export at 1,000 rows.

Average position #1 54 queries · 5.4% · under 1.5
Positions #2–3 66 queries · 6.6% · 1.5 to 3.0
Positions #4–10 411 queries · 41.1% · rest of page one
Beyond page one 469 queries · 46.9% · position 11+

120

Top 3 · 12.0% of the queries measured

531

Page one · 53.1% of the queries measured

Source: Google Search Console, matchmaker.lk, Search type: Web, 16 months ending 12 August 2026. Band thresholds are ours: #1 is an average below 1.5, since Search Console reports position as a decimal average rather than an integer rank.

Top queries by clicks · 16-month averages · names anonymised

Query Clicks Impressions CTR Avg position
[individual’s name] age anonymised 343 11,991 2.86% 8.04
free marriage proposals with phone number 184 2,181 8.44% 5.03
matrimony sri lanka 159 6,025 2.64% 6.44
matchmaker.lk 146 242 60.33% 1.09
free marriage proposals sri lanka 132 1,329 9.93% 5.9
marriage proposals 130 6,146 2.12% 10.85
sri lanka matrimony 121 6,880 1.76% 8.4
matchmaking sri lanka 114 408 27.94% 2.55
match maker 81 397 20.40% 1.61

The top query is a search for a named private individual whose proposal profile is on the site. It is anonymised here for the reasons set out above.

Top pages by clicks · Google, 16 months · profiles anonymised

Page Clicks Impressions CTR Avg position
/ 6,341 233,482 2.72% 10.08
/proposal/ 521 54,657 0.95% 18.22
/proposal/[profile]/ 436 15,828 2.75% 7.76
/[wedding-page]/ 72 5,592 1.29% 7.76
/about/ 64 51,658 0.12% 10.1
/proposal/[profile-2]/ 36 93 38.71% 3.49
/christian-matrimony/ 26 223 11.66% 9.27

The homepage takes 77% of all clicks from position 10.08. One anonymised profile page converts at 38.71% from position 3.49 on a small impression base — individual profiles convert extremely well when they surface, they just rarely surface.

Evidence · Bing

25.6K impressions at 4.66%, and 88% desktop

Bing Webmaster Tools search performance for matchmaker.lk showing 1.2K clicks and 25.6K impressions

Bing · 1.2K clicks, 25.6K impressions

Over 24 months at 4.66% CTR, well above the Google rate, and growing sharply from late 2025. The keyword list is dominated by rival-platform and newspaper-classifieds queries — the incumbent capture in the table above.

Bing Webmaster ToolsSearch Performance24 months

Bing top pages · 24 months

Page Impr CTR Pos
/ 16.5K 5.76% 5.58
/proposal/ 6.9K 2.67% 6.25
/about/ 70 10.00% 3.54
/grooms-in-jaffna/ 25 8.00% 4.96

The homepage again: 16.5K impressions and 948 clicks at 5.76%. Concentration on Bing too.

Bing by device · the inverse of Google

Device Impr Clicks CTR
Desktop 22,527 1,010 4.48%
Mobile 3,068 183 5.96%

88% of Bing impressions are desktop while Google runs 78% mobile — the third property in this portfolio to show that inversion. It is a property of the engines, not of the market.

Who it reaches

One in five clicks comes from outside the country

And the countries that convert best are not the one with the most clicks.

By country · Google

Country Clicks CTR Pos
Sri Lanka 6,568 3.02% 8.13
Australia 235 1.61% 31.81
United Arab Emirates 160 4.58% 7.81
United Kingdom 157 2.01% 13.21
United States 125 1.13% 20.44
Canada 120 3.71% 12.8
Japan 73 5.14% 6.81
Qatar 64 5.04% 7.6

Japan converts at 5.14% and Qatar at 5.04%, both above Sri Lanka's 3.02%. Migrant workers planning a marriage have a narrow window home and no patience for a vague page.

By device · Google

Device Clicks Share CTR
Mobile 6,417 77.8% 3.61%
Desktop 1,735 21.0% 1.63%
Tablet 98 1.2% 4.49%

Mobile converts at 3.61% against desktop's 1.63%, and desktop averages position 15.85 against mobile's 7.17. The site is materially better positioned on phones.

1,682

clicks from outside Sri Lanka — 20.4% of the total, across at least 9 countries

Every market with significant migration has this pattern, and it is routinely missed because a country report sorted by clicks buries it under the home market. If you sell anywhere with a diaspora — matrimony, remittance, property, education, legal — the second-through-tenth rows of that report are a segment, not noise.

Coverage, stated honestly

What is proven here, and what is not

Surface Status Finding Evidence
Google organic Documented 8,250 clicks from 286,275 impressions at 2.9%. 54 of the top 1,000 exportable queries average position #1 — the second-strongest #1 count in the portfolio. Search Console exports + screenshot
Google AI features Documented 5,610 appearances, of which 4,895 — 87% — land on the homepage. Daily volume roughly quadrupled across the reporting window. Search Console Generative AI report
Bing organic Documented 1,193 clicks from 25,595 impressions at 4.66%, including roughly 20% of impressions from queries aimed at incumbent newspaper and rival-platform brands. Bing Webmaster Tools exports
Bing Copilot Documented 3,800 citations at an average of just 2 cited pages a day — the most concentrated AI footprint of the four properties, and deliberately so. Bing AI Performance dashboard
Bing Copilot grounding queries Not audited The per-query citation-share export we used on the Velankanni study was not captured for this property, so no citation-share figure is published here. Export not captured
ChatGPT, Gemini, Perplexity Not audited No per-answer citation export exists for these surfaces, so they are not measured or claimed. No export available

Why a matrimony platform is in a SaaS SEO portfolio

Because your product has both kinds of query, and one strategy will not do

Every SaaS site contains both shapes at once, and treating them the same is the most common AI-visibility mistake we see.

Your docs, glossary and pricing explainers are fact pages. They win by being quotable — real tables, question headings, the claim near the top. Their citations spread across many deep URLs, and their CTR will fall as they succeed. That is the Coimbatore pattern.

Your homepage, category and comparison pages are entity pages. They win by being an organisation a model is confident enough to name when somebody asks for a recommendation. Their citations concentrate, and the click is the entire point. That is this property.

So we now audit them separately, with different targets. A fact page is judged on citation count and extractability. An entity page is judged on whether the assistant names you when asked to recommend something in your category — and, unlike a ranking, you can just go and ask it.

The second transfer is the competitive one. Ranking for “sunday observer marriage proposals” is structurally identical to ranking for “[competitor] alternative” — demand phrased as somebody else's brand, because that brand taught the market the category. In most SaaS verticals those pages are the highest-intent traffic on the site, and they are usually built last.

Who did this work

An owned property, published with names removed

Alston Antony, operator of MatchMaker.lk

Alston Antony

Owner & SEO lead

MatchMaker.lk is owned and operated by Alston Antony. He has worked in search since 2010, holds an MSc in Computer Software Engineering (Distinction) from the University of Greenwich, and is a professional member of BCS, The Chartered Institute for IT. His day role is Senior Digital Marketing Manager at Brainstorm Force, whose software powers 7M+ sites.

Owning the property is what makes the exports publishable. Handling other people's personal data is what makes parts of them unpublishable, and both are true at once.

How to verify every number on this page

  1. Google totals come from Search Console for matchmaker.lk, Search type: Web, the 16-month window ending 12 August 2026. Site totals use the Devices export, which is complete.
  2. Pages is a real total; queries is a cap. The pages export returned 930 rows, under Search Console's 1,000-row ceiling. The query export hit the cap at 1,000, so position bands are shares of that sample.
  3. Personal names are removed, numbers are not. Anonymisation replaces the query string and the URL slug only. Clicks, impressions, CTR and position are exactly as exported.
  4. The incumbent total is approximate and marked as such. Roughly 5,120 Bing impressions come from queries naming a newspaper section or rival platform. Classifying a query as “aimed at an incumbent” is a judgement, so the figure carries a tilde.
  5. The cross-property comparison uses the “Avg. Cited Pages” tile from each property's Bing AI Performance dashboard over a 6-month window. Bing states these reports are a sample of overall activity.
  6. No citation-share figure appears here. The per-query export we used on the Velankanni study was not captured for this property, so that surface is marked not audited.
  7. The site is live. Open matchmaker.lk and ask an assistant to recommend a Sri Lankan matrimony site.

Case study FAQ

The questions this raises

What does it mean to be “recommended” rather than “quoted” by an AI?

They are two different jobs. A fact query — “what does a Red Taxi cost per km” — gets answered by quoting a page, so the winning asset is an extractable claim. A recommendation query — “best matrimony site in Sri Lanka” — gets answered by naming a brand, so the winning asset is an unambiguous entity the model is confident enough to recommend. On this property, 87% of AI appearances land on the homepage and Copilot cites an average of just 2 pages, versus 35 on our city-guide property. Same platform, same metric, opposite shape — because the questions are different.

Why does the site rank for a newspaper’s brand name?

Because the person searching “sunday observer marriage proposals” wants marriage proposals, not a newspaper. In Sri Lanka the newspaper classified sections were the category default for generations, so a large share of demand is still phrased as the incumbent's brand. Around 20% of this property's Bing impressions come from queries naming a newspaper section or a rival platform. Answering the underlying intent honestly is legitimate; pretending to be the Sunday Observer would not be, and the pages do not.

Why are the names and profile URLs blanked out on this page?

Because matchmaker.lk is a matrimony platform, so its highest-traffic pages are individual people's proposal profiles and its single top query is a search for a named private individual. Those pages are public, but reproducing a real person's name and marital status in an agency's marketing material is a different act from hosting their profile. Every profile URL and personal-name query on this page is anonymised. Nothing analytical is lost.

Is a 2.9% CTR at position 10.4 good?

For this mix, yes — and the reason is instructive. The homepage alone takes 6,341 clicks from position 10.08, because a branded and category-level query set converts even from the bottom of page one. Meanwhile the site holds 54 queries at average position #1, mostly brand and long-tail category terms. The single site-wide average hides both facts, which is why every case study on this site shows the distribution instead.

What is the diaspora finding?

One in five clicks — 1,682 of 8,250 — comes from outside Sri Lanka, across at least nine countries: Australia, the UAE, the UK, the US, Canada, India, Japan, Qatar and Italy. Japan converts at 5.14% and Qatar at 5.04%, both above Sri Lanka's 3.02%. Any market with significant migration has this pattern, and it is routinely missed because a country report sorted by clicks buries it under the home market.

How does this transfer to a SaaS company?

Two ways, directly. First, separate your fact pages from your entity page: your docs win citations by being quotable, your homepage and category pages win by being an unmistakable entity a model will name in a recommendation. They need different work. Second, your competitor's brand name is a keyword — the “[competitor] alternative” and “[legacy tool] replacement” pages are the same play as ranking for a newspaper's classifieds section, and in most SaaS categories they are the highest-intent traffic available.

Ask the assistant about yourself

Does the AI name you when asked to recommend one?

It takes thirty seconds to find out, and most teams have never checked. Send Alston your product URL and category and you will get a read on whether you are an entity the models will recommend — and which of your pages should be chasing citations instead.

Free SaaS SEO growth blueprint

Opportunity map, keyword gaps, technical priorities and a 90-day plan.

Founder-led, never junior-led

You work directly with Alston Antony from audit to reporting.

Reported against revenue

Signups, trials, MRR and AI share-of-voice, not impressions.

An honest yes or no

We only take on SaaS companies we believe we can get results for.