From F&B to fintech, clinics to law firms, startups to enterprise. If your customers search on Google, we make sure they find you first, not your competitors.
One specialist team, focused only on the organic rankings that put you in front of ready-to-buy Singapore customers.
A clear, sequenced path from audit to rankings. You always know what we’re doing and why it matters for your leads.
Quick answer: A backlink database is an index built by crawling the web and recording every hyperlink found, then refreshing it on a schedule. Coverage and freshness differ by provider because crawl budgets and priorities differ. Separately, your own prospect database is a working sheet you maintain yourself.
Two different things share this name and conflating them causes real confusion. The first is the commercial link index that sits behind any tool showing you who links to whom: a crawled, stored, periodically refreshed picture of the web’s hyperlinks. The second is the working sheet you keep of sites you want to approach. This post explains how the first is built, why two providers looking at the same site report different numbers, and what a lost link means inside an index versus in reality. Then it covers building the second: the fields to keep, how to dedupe, and how to stop it going stale. It is not a comparison of which provider to buy, because the mechanism is the durable knowledge and the feature lists change every quarter. Understanding it properly is the difference between reading a report and being persuaded by one, and it belongs alongside the crawl and index thinking in our technical SEO work.
The mechanism is simple in outline and expensive in practice, and knowing it explains every discrepancy you will ever see.
A crawler starts from a seed set of known URLs and follows links outward. Every page it fetches yields a list of outbound hyperlinks. Each of those becomes a record: source URL, destination URL, anchor text, link attributes, first seen date. That record set, at scale, is the database.
Crawling is a budget allocation problem, and that is the key insight. The web is effectively unbounded and every provider has finite bandwidth, storage and processing. So each one decides, continuously, which URLs are worth fetching and how often. Those decisions are the product. Two providers with identical technology and different priorities will hold materially different pictures of the same web.
Then the raw records get processed into what you see. Duplicate URLs are canonicalised, redirect chains resolved, subdomains rolled up or not, and links classified. Authority scores are computed from the link graph the provider happens to hold. That last point matters: an authority metric is a calculation over one provider’s crawl, not a property of the site.
Finally, refresh. A link recorded in March may be gone by June, so crawlers revisit. Revisit frequency follows the same budget logic: frequently updated, well-linked pages get re-fetched often, a dormant association page in a small market may not be revisited for months. Your link on that page exists the whole time and appears in the index whenever the crawler next gets there.
If you have ever run one domain through two tools and got numbers that differed by half, this is why. None of it means one is broken.
Different crawl footprints. Provider A may have discovered a Singapore trade association’s member directory years ago and revisit it monthly. Provider B may never have crawled it. Neither is wrong about the web; they are right about different subsets of it.
Different counting rules. Whether subdomains roll into one referring domain, whether sitewide footer links count once or many times, whether redirected links are attributed to the source or the destination, whether links inside content rendered by JavaScript are captured at all. Each decision is defensible and each changes the total.
Different link classification. Some providers report only links they consider live and indexable; others include everything found. Some exclude nofollowed links from headline counts, some include them, some report both.
Different refresh cycles, which is the biggest cause of surprise. Your new link appears in one tool within days and another six weeks later. Nothing changed except crawl priority.
| Discrepancy you will see | Usual cause | What to do about it |
|---|---|---|
| One tool shows 40 domains, another 95 | Different crawl coverage plus different subdomain rules | Pick one as your baseline and stay with it |
| A new link is missing entirely | Not yet re-crawled, or the page is not indexed | Check indexation before assuming the tool is wrong |
| Authority score differs sharply | Metrics are computed over each provider’s own graph | Never compare a score across providers |
| A link shows as lost but is visibly there | Crawler failure, block, or a render dependency | Verify by eye before acting |
| Your console shows links no tool does | The search engine’s crawl is larger than any commercial one | Treat the console as authoritative for your own site |
The practical rule: pick one index as your reporting baseline and never mix sources in one report. Trend within a single index is meaningful. A number from one provider compared with a number from another is noise, and it is the most common way a link report gets misread.
For your own site, your search engine webmaster console outranks all of them. It reflects the crawl that actually feeds rankings. Commercial indexes are genuinely better for looking at sites you do not own, which is most of the work.
This single distinction prevents most of the panic that link reports cause.
In a database, lost means the crawler did not find the link the last time it looked at that page. That is all it means. It is an observation about a crawl, not a statement about the web.
Five things produce that observation, and only two are real losses. The page was edited and the link genuinely removed. The page was deleted or the site closed. Those are real. Then: the crawler was blocked or timed out on that visit. The link now depends on client-side rendering the crawler did not execute. Or the URL changed and the link moved with it, so the old record went stale while the link itself is fine.
So verify by eye before you do anything. Open the page. Use the browser’s find function on your domain name. Check whether the page itself is still indexed, because a link on a deindexed page passes nothing regardless of what any tool reports. Two minutes of checking prevents an unnecessary and slightly embarrassing email to a publisher.
“First seen” is equally soft. It means first crawled, not first published. A link first seen in August may have gone live in May. This matters when you are attributing results to a campaign window, and it is why we date link findings by verification rather than by the tool’s timestamp.
Attrition is normal and should be budgeted for. Sites redesign, editors prune, businesses close, domains change hands. A profile that loses a small share of its domains annually is behaving ordinarily. The number to watch is not whether you lose links but whether you lose them faster than you earn them, and whether the ones you lose are the ones that mattered. On large catalogue sites the churn is usually compounded by internal URL changes rather than external removals, which is a recurring theme in our ecommerce SEO work.
Most agencies read the lost-link column as a to-do list. We have found that reading it as a data-quality column first is more productive: work out which entries are crawl artefacts, which are dead pages that were already contributing nothing, and which are genuine removals, and the list that remains is usually short enough to act on the same week. The same internal-first logic sits behind our ecommerce case study, where 200+ redirect chains from two prior platform migrations were silently leaking PageRank.
Now the second meaning. This is a working sheet, it is the most valuable asset in a link programme, and almost nobody builds it properly.
Start with the identity fields. Domain, in one canonical form: lowercase, no protocol, no www, no trailing slash. This is your deduplication key and getting it right on day one saves hours later. Then the organisation name, and the type: association, register, trade publication, national media, supplier, platform partner, client, event, community, other.
Then the qualification fields, which are the point of the whole exercise. Relevance to your sector, scored one to five by your own judgement. Whether the site links outward at all, which you check by looking. Whether you have an existing relationship, and of what kind. The specific reason this domain is on the list, written as a sentence, because a reason you cannot write down is not a reason. And the asset or angle you would approach with.
Then the contact fields. Named person where you have one, role, email, source of that email, and the date you found it. Contact data decays faster than anything else in the sheet.
Then the activity log, which is what turns a list into a database. Date of first approach, channel, what you sent, response, response date, outcome, and date of next planned touch. This is the section that prevents the single worst outreach error, which is approaching the same editor twice with the same thing because two people on your side were working the same list.
Finally the result fields. Live URL if published, date verified, anchor text used, placement type, and a status flag: prospect, contacted, replied, published, declined, do not approach. That last value needs a reason attached, and it is worth respecting permanently.
| Field group | Fields | Why it earns its column |
|---|---|---|
| Identity | canonical domain, organisation, type | Dedupe key and segmentation |
| Qualification | relevance score, links out, relationship, reason, angle | Stops the list becoming a mailing list |
| Contact | person, role, email, source, date found | Decays fastest, needs a date stamp |
| Activity | approach dates, channel, what sent, response, next touch | Prevents duplicate and badly timed contact |
| Result | live URL, verified date, anchor, placement type, status | Your permanent, portable record |
Keep it in something two people can edit and one person owns. A shared spreadsheet is entirely sufficient for the first few hundred rows, and for a lean team it is the format we recommend rather than a dedicated platform, which is the same reasoning we apply to tooling in small business SEO engagements. The tooling is not the hard part; the discipline of filling in the reason column is.
A prospect list becomes unusable through duplication long before it becomes too large.
Canonicalise the domain on entry, not later. Lowercase, strip the protocol, strip www, strip trailing slash and any path. Do it with a formula so it cannot be forgotten. Every duplicate you will ever have starts as a formatting variant.
Group by owner, not only by domain. One publisher may run four titles, and one association may have a main site, an events subdomain and a separate members portal. Approaching all four separately reads as spam to the single human who receives them. Add an owner or group column and sort by it before any outreach round.
Watch for the same contact across different domains. A freelance editor may appear under three publications. Your activity log needs to be readable by person as well as by domain, or you will contact them three times in a fortnight.
Merge conservatively and keep the history. When two rows are the same organisation, keep the fuller qualification notes and concatenate the activity logs rather than discarding one. Losing the record that somebody declined in March is how you get a firmer no in September.
Mark exclusions explicitly instead of deleting. Sites you have decided not to approach, for relevance, quality or brand reasons, should stay in the sheet with a status of do not approach and a reason. Deleting them means someone rediscovers and re-adds them next quarter. In sectors where the institutional layer is dense and interlinked, this owner-level grouping is most of the maintenance work, which is what makes the listings pass in our real estate SEO engagements slower than it looks.
A prospect database has a half-life. Plan maintenance or it quietly becomes a list of dead addresses.
Re-verify contacts every six months, and sooner in high-turnover sectors. Editors move, association secretariats rotate, generic inboxes get replaced. An email that bounces costs you the opportunity silently.
Re-verify your live links quarterly, by eye, on a sample. Take twenty of your most valuable placements, open them, confirm the link and the indexation, and update the verified date. Automated monitoring is useful for alerting but a human check on the ones that matter catches render-dependent and moved links that tools mishandle.
Add rows continuously from three cheap sources. Competitor referring domain lists tell you which local sites in your sector link out at all. Your own unlinked mentions are pre-qualified by definition. And your own paperwork keeps producing new eligibility as you add memberships, certifications and platform partnerships.
Review status flags monthly and set next-touch dates deliberately. The most common failure in a link programme is not a bad list, it is a good list nobody followed up on. A prospect who did not reply is not a no; a prospect contacted four times in six weeks is a closed door.
Keep one dated export a quarter. A snapshot you can diff is how you answer the question of what actually changed, and it is what makes the database portable if the person maintaining it leaves. Because the useful pool of linking domains in Singapore is small enough to enumerate, a well-maintained sheet here approaches genuine completeness in a way it never would in a larger market, and it becomes the reference document for every sector we work in across our industry SEO engagements.
Field notes: Data quality problems are usually duplication and stale records rather than size, and our medical case study shows the same thing in listings. The GP clinic in Toa Payoh had no directory listings of any kind, and four conflicting old listings with outdated phone numbers were still live. Before adding anything new, those four were resolved, and consistent NAP data was then submitted to 40+ Singapore healthcare and general business directories. Over 6 months, alongside technical and content work, Domain Authority moved from 8 to 19. A clean, deduplicated record of what already exists is the starting point for any database, which is why verification by eye sits ahead of outreach in our own process.
Learn the mechanism once and you stop being surprised by link data for good. A backlink index is one provider’s crawl of the web, refreshed on a budget, counted under house rules and scored against its own graph, which is why two tools disagree and why neither is lying. Pick one, use it for trend, use your own console as the authority on your own site, and verify anything important by opening the page. Then put your energy into the database that is genuinely yours: a prospect sheet with a canonical domain key, a written reason per row, an activity log and dated verification. In a market with a countable pool of useful linking domains, that sheet can get close to complete, and complete is worth more here than any tool subscription. It is also the asset that survives a change of agency or staff, which is exactly why it should live in your systems rather than a supplier’s. If you want a read on whether an authority gap is what is actually holding you back before you build any of it, that is what our SEO audit and consulting work is for, and the broader programme it sits in is set out on our SEO services page.
We recommend building a lightweight prospect database before paying for a full crawler subscription, because in our experience most small Singapore sites need a few hundred qualified rows, not the millions a commercial index covers. Our team keeps these as a plain spreadsheet with contact, relevance score, and last-contacted date, and that has outperformed several paid tools we have trialled for clients on a tight budget.
In its commercial sense it is an index built by crawling the web and recording every hyperlink found, with the source page, destination, anchor text, attributes and dates, then refreshed on a schedule. In its practical sense for a business, it is your own maintained sheet of sites you want to earn links from. The two share a name and nothing else, and most confusion about link data comes from mixing them up.
Because each crawled a different subset of the web, applies different counting rules for subdomains, sitewide links and redirects, classifies links differently, and refreshes on its own schedule. All four differences are legitimate design decisions rather than errors. Choose one index as your reporting baseline, track trend inside it, and never compare a raw number or an authority score across two providers.
Coverage varies by region, sector and moment, and any answer stated as a fact will be out of date shortly. For your own site, treat your search engine webmaster console as the authority, because it reflects the crawl that actually informs rankings. For competitor and prospect research, run one domain you know well through the free tiers of two or three indexes and see which finds the local sources you recognise.
Not necessarily. It means the crawler did not see it on its last visit, which can also happen if the crawler was blocked or timed out, if the link now depends on client-side rendering, or if the URL changed. Open the page and search for your domain before doing anything. A meaningful share of reported losses in our own checks turn out to still be live.
Continuously, but unevenly, because revisit frequency follows crawl priority. Busy, well-linked pages are re-fetched often; a quiet association page in a small market may wait weeks or months. That is why a new link can appear in one tool within days and another much later, and why a link you gained yesterday being absent today is not evidence of anything.
Five groups: identity, meaning a canonical domain, the organisation and its type; qualification, meaning a relevance score, whether it links out, any existing relationship, and a written reason; contact, meaning a named person with the date you found the address; an activity log of every approach and response; and results, meaning live URL, verified date, anchor and status. The written reason column is the one that keeps it a prospect list rather than a mailing list.
Canonicalise the domain on entry with a formula, so lowercase, no protocol, no www, no trailing slash and no path. Then add an owner or group column, because one publisher can run several titles and one association can have several properties, and approaching them separately reads as spam to the person receiving it. Track contacts by person as well as by domain, and merge by concatenating history rather than deleting rows.
A shared spreadsheet is genuinely adequate for the first several hundred rows and has the advantage of being portable and owned by you. Dedicated outreach tools earn their place when several people are sending, because sequencing and reply tracking become the bottleneck. The deciding factor is rarely the software: it is whether someone reliably fills in the reason and next-touch columns.
Re-verify contacts every six months, sample-verify your live links by eye each quarter, and add rows continuously from three sources: competitor referring domains, your own unlinked mentions, and your own new memberships and certifications. Review status flags monthly and set deliberate next-touch dates. Keep one dated export a quarter so you can diff what changed and so the asset survives staff changes.
Yes, more so here than in a large market, because the pool of genuinely useful local linking domains is small enough to enumerate by hand. Work outward from trade associations, chambers and bilateral councils, statutory registers and licensing bodies, sector trade titles, the national business press, and the platform partner directories you are eligible for. An afternoon of browsing plus your own paperwork will produce a better-qualified list than an export.
If you are staring at two link reports that disagree, or you want a second opinion on a prospect list before anyone starts sending, we will happily read it. We can also tell you whether the authority gap is real in the first place, which decides whether the database is worth building now or later. Start a conversation through our contact page.
Natalie leads SEO strategy at Singapore SEO Agency, helping local and regional businesses build organic search programmes that drive qualified leads. She specialises in technical SEO and content-led authority building for Singapore SMEs.
Get a free SEO audit for your Singapore website — we'll show you exactly where you stand, what's holding you back, and what it would take to rank on page 1.
Get Your Free SEO Audit →
Domain authority is a useful filter for weak sites but a poor sales promise. Learn why scores move on their own and what to ask before you buy links or SEO.
Paying for an SEO keyword research tool? Test its Singapore data on 20 keywords you know, see why KD differs by vendor, and pick one tool you can stick with.
Whats bounce rate, and how is it different from exits, engagement time or dwell time? Learn what each one really answers and what to ask your SEO agency.
Google Analytics website traffic is a consented sample, not a head count. See where visits come from, why tools disagree and which number to trust each month.
Google analytics consulting services for Singapore SMEs: what consultants do, indicative SGD costs, deliverables to demand and red flags. Buy the right scope.
Bounce rate in GA4 is not the old Universal Analytics number. Learn what it measures now, why old benchmarks fail and how to judge each page by its job.
Fast, no obligation. We reply within 24 hrs.
Singapore’s specialist SEO agency for SMEs. We rank your business on Google — and only Google. No distractions, just results.
© 2026 Singapore SEO Agency. All rights reserved.