Auction Domain Research Filtering Playbook

This is the auction-domain filtering guide behind my SparkNamer auction shortlist.

It explains how I move from raw auction inventory to a smaller list of names that are worth reviewing manually.

If you want the daily filtered picks instead of only the framework, subscribe from the sticky box and I will send the shortlist, buyer thesis, risk notes, and watch / skip / consider verdicts.

The Core Idea

Auction domains look simple from the outside:

Find cheap names.
Bid on the good ones.
Sell them later.

In practice, the hard part is not finding cheap domains.

The hard part is avoiding names that are cheap for a reason.

Our current auction-domain research workflow is built around one principle:

Metrics create the shortlist.
Wayback and trademark checks remove obvious risk.
Brand taste decides whether the name is worth buying.
Max bid protects the downside.

The goal is not to buy more domains.

The goal is to buy fewer names that are easier to explain, easier to price, and easier to imagine in the hands of a real buyer.

Explore auction-domain ideas with SparkNamer

1. Input Sources

The research starts with daily auction and droplist files.

The current input stack:

GoDaddy auction export
Namecheap auction export
GoDaddy droplist 24h
GoDaddy droplist 72h
GoDaddy droplist ending today
GoDaddy droplist top1000

Each file has a different role.

The raw inventory is usually too large to review manually from scratch.

The first job is not to pick winners.

The first job is to remove obvious noise quickly.

2. Machine Filter

The machine filter creates the first research queue.

It keeps domains with basic auction/value signals:

Low price, usually $1-$30
Low bids, ideally 0-1 bids
.com preferred
Older age if available
Clear commercial keywords
Clear business use case
Valuation gap if available
Referring domains/backlinks if available
DotDB or exact-TLD signal if available

Useful commercial terms include:

tool
book
legal
health
data
cloud
lead
job
claim
game
finance
scheduler
booking
pay
med
hire
design
wine
travel
lab
app
agent
code

The script also removes or flags obvious bad patterns:

Typos
Trademark traps
Adult/casino/pharma terms
Awkward grammar
Forced AI/SEO names
Cheap keyword mashups
Names that only look good because of metrics

The machine filter is not allowed to make the final buying decision.

It is only a ranking and triage step.

3. Why Metrics Are Not Enough

Auction exports can make weak names look attractive.

A domain can have:

• old age
• a low current bid
• backlinks
• referring domains
• a GoDaddy estimated value
• a valuation gap
• commercial keywords

and still be a bad buy.

The name may sound ugly.

The words may not fit together.

The buyer pool may be too small.

The Wayback history may be polluted.

There may be trademark noise in the same category.

This is why we treat metrics as a funnel, not a verdict.

4. TLD Pattern

The default priority is still .com.

That does not mean other TLDs are impossible. It means they need a stronger reason to survive.

The current TLD rule:

.com is the default priority.
Alt TLDs are only worth checking if the SLD is very strong.
A weak TLD can still work if the SLD is excellent.
A good SLD can make a bad TLD better.
Domain hacks can work if the full name reads naturally.
Alt TLDs get a strong liquidity discount.

Examples:

discount.link can work because the full string reads as a natural commercial phrase.
.dev is a weaker TLD, but it is becoming more accepted.
.ai and .io only work when the name itself is clean, strong, and commercially obvious.

The important lesson:

A good SLD can improve a weak TLD.
A weak TLD cannot rescue a weak SLD.

5. Brand Taste Filter

After the machine filter, every candidate gets a human taste check.

This became more important after reviewing feedback from experienced domain investors. A name can pass a spreadsheet screen and still fail the human screen.

The brand-taste questions:

Does this sound like a real brand?
Is it easy to say?
Is it easy to spell?
Is it easy to remember?
Does it feel premium, functional, or cheap?
Would a founder actually name a product this?
Does the wording feel natural?
Does it have weird double letters?
Does it have awkward grammar?
Is it just a forced AI/SEO keyword combo?

Strict rules:

If Brand Taste is weak, do not mark it Investable.
If Investor Gut Check is No, do not buy, even if the metrics look good.
If Phonetic / Spelling Feel is bad, skip unless the category is extremely strong.

This filter removes names that are technically interesting but emotionally flat.

That includes many names with:

• forced AI wording
• forced SEO wording
• random tech suffixes
• unclear abbreviations
• cheap lead-gen phrasing
• awkward grammar
• low-trust medical/legal/finance wording

6. Business Fit

Every domain must answer a buyer question.

The key questions:

Who would buy this?
What would they use it for?
Is the buyer pool broad enough?
Is there real commercial value?
Is this a business, product, marketplace, lead-gen, SaaS, or content brand?
Is it too local or one-buyer-only?

Good signs:

Clear SaaS use case
Clear lead-gen use case
Clear vertical category
Clear product/workflow name
Multiple possible buyer groups
Commercial intent
Easy explanation

Weak signs:

Only one obvious buyer
Local-only use case
No clear monetization
Cute but no buyer
Metrics good but name not usable
Too much explanation required

The best candidates usually have both:

name quality
buyer clarity

One without the other is not enough.

7. Automated Prebid Check

Once a candidate survives the early filters, it goes through a prebid script.

The script checks:

Auction page reachability
Best-effort live/static price parse
Wayback closest snapshot
Trademark exact hits
Trademark close hits
Auction URL
Manual flags

This script is useful, but it is not magic.

GoDaddy often blocks static automation. That means the final live page still needs to be opened manually before bidding.

Manual live checks:

Current price
Current bid count
Time left
Auction fees
Renewal cost
Whether the listing is still active
Whether the listing changed from bid to buy-now or closed

The script helps reduce manual work.

It does not remove final responsibility.

8. Wayback Audit

We do not trust a simple "Wayback exists" signal.

The workflow fetches real archived HTML snapshots and classifies the visible history.

Wayback classifications:

GREEN_NO_OBVIOUS_SPAM
YELLOW_PARKED_OR_SALES
YELLOW_THIN
NO_SNAPSHOT
FETCH_FAILED
RED_RISK

Skip immediately if the audit finds:

Adult history
Casino history
Pharma spam
Malware or hacked content
Payday loan spam
Foreign spam

Accept with caution if the history is:

Mostly parked
For-sale page
Default hosting page
Thin placeholder
No snapshot

This step matters because some names look good until the archive shows what they were used for.

One of the strongest lessons from the workflow:

Good name + bad history = usually skip.

9. Trademark Risk

Trademark checks are not legal advice, but they are useful risk screens.

The current classification:

Low: no exact hits, generic phrase
Medium-Low: close generic hits, not same category
Medium: close marks in related category
High: exact or very close active mark
Critical: exact active mark in same commercial category

Rules:

Exact active trademark: usually skip.
Close trademark in same category: lower bid or skip.
Generic phrase with no exact mark: okay for small bet.
Medical, legal, finance, and insurance names always get extra caution.

Close marks do not always kill a name.

But they lower confidence.

They also lower max bid.

10. Brand / Investor Decision Layer

To avoid buying "metric-only" names, the final table includes these human judgment fields:

Brand Taste
Phonetic / Spelling Feel
Premium Feel
Investor Gut Check
Decision Tier

Suggested values:

These fields are intentionally subjective.

Domain investing has a taste layer. The workflow should admit that instead of pretending every decision is purely numerical.

11. Decision Tier Rules

The final recommendation uses four tiers.

Investable

Good taste, clean use case, buyer pool exists, and no major Wayback or trademark issue.

This does not mean "bid aggressively."

It means the domain is worth considering within a disciplined max bid.

Cheap Experiment

Usable name, but one or more limitations:

• smaller buyer pool
• weaker premium feel
• close trademark noise
• parked/thin history
• category is useful but not urgent
• name is good enough only at a low price

These names can be bought, but only very cheap.

Data-Interesting Not Buyable

Metrics look interesting, but the name itself is not convincing.

Common reasons:

• awkward phrasing
• low brand taste
• ugly spelling
• too much explanation required
• weak buyer pool
• feels like a spreadsheet artifact

This tier is important because it prevents the machine filter from overfitting.

Skip

Reject the name.

Common reasons:

• bad Wayback history
• trademark risk
• ugly grammar
• typo
• spammy wording
• no buyer thesis
• legal/medical/finance risk without enough upside

12. Max Bid Logic

Max bid is where the research becomes practical.

Current max-bid logic:

Investable:
Usually $8-$30 depending on quality.

Cheap Experiment:
Usually $3-$12.

Alt TLD:
Usually $1-$10 unless the SLD is exceptionally strong.

Trademark / Wayback caution:
Lower bid or skip.

Legal / medical / finance:
Lower bid unless very clean and generic.

Awkward name:
$0 or skip.

Most auction mistakes happen when a buyer turns a small-bet name into an emotional chase.

The max bid should be set before bidding.

13. Final Table Schema

The final research table uses this schema:

Research Date
Domain
Source
Auction Type
Current Price
Bids
Time Left
Domain Type
Business Fit
Commercial Use Case
Buyer Pool Signal
DotDB Signal
Wayback Signal
Similar Brands / Companies
Trademark Risk
Brand Taste
Phonetic / Spelling Feel
Premium Feel
Investor Gut Check
Main Risk
Research Confidence
Suggested Action
Max Bid
Decision Tier
Final Decision
Public Note
Last Checked

The purpose of the table is not just to list domains.

The purpose is to make the decision easy to audit later.

14. Final Selection Rule

Do not force a full list.

If the day is weak, return fewer names.

The current rule:

Do not force 20 domains.
If the day is weak, return only 5-12 names.
If nothing is strong, say "no strong buy today."
Quality over quantity.
A shorter list with better taste is better than a long list with weak names.

This is one of the most important improvements to the workflow.

The final list should feel buyable, not just complete.

15. Manual Checks Before Bidding

Even after automation, some checks remain manual.

Before bidding, confirm:

Live auction price
Current bid count
Time left
Auction fees
Renewal cost
Exact Wayback snapshot if the domain is high priority
Exact trademark search
Whether similar brands already exist
Whether the name still feels good after seeing it on the auction page

The last item sounds soft, but it matters.

Sometimes a name looks better in a spreadsheet than it does when you stare at it alone.

16. What We Learned

The workflow improved over time because the first versions were too metric-driven.

The biggest insights:

Cheap is not enough.
Old is not enough.
Backlinks are not enough.
GoDaddy estimated value is not enough.
Commercial keywords are not enough.
Alt TLDs need stronger SLDs.
Good SLD can make a weak TLD better.
Bad Wayback history can kill a good-looking name.
Trademark noise should lower max bid.
Brand taste deserves its own columns.
Investor gut check should be explicit.
Do not force the list to 20 names.

The best version of this workflow combines machine speed with human taste.

The machine removes noise.

The scripts reduce risk.

The final decision still needs taste, buyer logic, and bid discipline.

Use SparkNamer to build and review your own domain shortlist