Questions Before You Pay for Launchpad Tooling
This is the page to read with a quote already in front of you. Twenty-five questions, grouped so you can stop early when a group fails, a sheet for scoring the answers rather than the impression, and the four replies that mean the conversation is over. Nothing here depends on which class of tool you are buying.
- Format
- Twenty-five questions in five groups, plus a scoring sheet
- Use it when
- A seller has quoted you a price and you have not paid
- Applies to
- Every class on this shelf, whatever it is called in the pitch
- Hard stop
- Four answers end the evaluation regardless of the score
- Not covered
- Which seller to choose, which this desk will not tell you
Before paying for any launchpad tool, get answers to a fixed set of questions in five groups: what the tool actually does, who holds the keys, what the total cost is, what evidence exists for the claims, and what happens when something fails. Score the answers rather than the impression, and stop the evaluation entirely if any of four specific replies appears. The list below is that procedure, written to be used with a quote already in hand.
None of it depends on the class of tool. A bundler, a volume tool and a wallet manager all sit behind the same five groups, because the risks that actually cost buyers money are structural rather than technical. If a seller cannot handle these questions, the feature list is irrelevant.
How to use the list
Work the groups in order and stop when one fails badly. The order is deliberate: group B contains the questions whose wrong answers can cost you the entire balance rather than the purchase price, so it comes before anything about features or price. There is no benefit to a careful cost analysis of a tool that will not let you hold your own keys.
Send the questions as a written message, all at once, and keep the replies. This is not adversarial; it is ordinary procurement, and any seller who has run their product at scale can answer most of these from memory. The purpose of writing is that a claim in a message is checkable later, and a claim in a call is a memory of a claim.
Expect to spend an hour. That is short relative to the money involved, and considerably shorter than the time spent afterwards by teams who skipped it.
A. What it actually does
The purpose of this group is to find out which class you are buying, in your own words rather than the seller's. Vocabulary in this market is deliberately loose, and a product described as a launch suite may be four things or one thing with three names.
- A1. Which class is this, in one sentence? Bundler, sniper, volume tool, visibility tool or wallet manager. If the answer is more than one, price each part separately.
- A2. When the run finishes, what will I have that I did not have before? A supply position, executed trades, funded wallets or messages. An answer describing an outcome rather than an object is a warning.
- A3. Which venues does it support, by name? Not "all of them". Include what happens when a token migrates from a launchpad curve to a standard pool mid-campaign.
- A4. What exactly will appear on the chain? Transaction types, how many, from how many wallets, over what period. Anyone who has operated the tool can describe this precisely.
- A5. What can this tool not do? A seller who cannot name a single limit either does not understand the product or is not going to be honest about anything harder than this.
B. Custody and keys
This group decides whether the rest of the evaluation is worth doing. Every multi-wallet tool holds or generates keys, and the arrangement determines the maximum you can lose. Test rather than discuss wherever a test is possible.
- B1. Where are private keys generated, and where are they stored? A location and a mechanism, not the word secure.
- B2. Can I export every private key on demand? Then verify it on a throwaway wallet before funding anything real.
- B3. Can I import keys I generated myself? The safest arrangement, and the answer tells you how the product thinks about control.
- B4. What is the maximum I can lose if you are compromised? If the honest answer is everything in the wallets, it is better to hear it stated than to infer it later.
- B5. Are wallets ever reused across clients? Somebody else's history attached to wallets trading your token is a problem you inherit silently.
The export question is the one to actually perform rather than ask. Create one wallet, fund it with a trivial amount, export the key, import it into a wallet you already use, and move the funds yourself. A product that makes this a support conversation has told you where control sits.
C. Cost
Every tool on this shelf has a headline price and a real cost, and the gap between them is usually larger than the difference between two sellers. This group closes the gap before you pay rather than after.
- C1. Itemise everything I will pay. Service fee, network fees, venue fees, priority fees or tips, and rent. Which of those are inside the quote and which are extra.
- C2. Is the service fee charged on volume, on capital deployed, per transaction or on time? The four produce very different bills for identical work.
- C3. What capital is locked rather than spent? Rent in each account, balances sitting in wallets, and value converted into token supply.
- C4. What comes back at the end, and when? Residual SOL, residual tokens, closed accounts and returned rent, with a stated timeframe.
- C5. What does a failed attempt cost? Failures still consume fees. A tool making many attempts spends money continuously whether or not anything works.
Do the arithmetic yourself once you have the answers. The protocol base fee is 5,000 lamports per signature and account rent minimums are fixed constants documented in the Solana documentation, so the only variables are the seller's fee and the venue's. Reduce every quote to a cost per unit of the thing you actually want, then compare.
D. Evidence
This group separates operators from resellers and resellers from people with a landing page. The standard is simple: anything claimed should be checkable by you, without the seller's help.
- D1. Give me two token addresses from runs you operated. Not screenshots. Addresses, which anyone can read.
- D2. What should I expect to see when I look at them? A specific description before you look is far more informative than an explanation afterwards.
- D3. Do I receive transaction signatures for my own run? A run you can audit is a service; a run reported as an image is a story.
- D4. Who operates this, and are you the operator or a reseller? Resellers are not disqualified, but a reseller cannot answer operational questions and should not pretend to.
- D5. Where did any figures in your pitch come from? Success rates and result claims should have a method behind them. Usually there is none, and asking establishes that quickly.
Verify at least one address yourself. Open a public explorer, find the window the seller described, and check whether what happened matches what they said would have happened. This is fifteen minutes of work and it is the single most informative thing in the whole list after the export test.
E. Failure and support
Everything above concerns the good case. This group concerns the day something goes wrong, which is when the difference between products actually shows.
- E1. What happens if a run stops halfway? What state am I left in, what is refunded, and can it be resumed.
- E2. How do I stop a run immediately? A control I can use myself, and how long it takes to take effect.
- E3. What happens if your service is unavailable during my run? Specifically what I can do without you, using keys I hold.
- E4. Who do I contact, and what is the realistic response time? Named channel, stated hours, and what happens outside them.
- E5. What is the refund policy, in writing? Including the case where the tool worked exactly as described and the outcome disappointed, which is not a defect.
Scoring the answers
Impressions are unreliable after an hour of enthusiastic messages, so convert the answers into a number. This sheet is an ordinal judgement rather than a measurement, and its only purpose is to stop a confident seller from outweighing a set of vague replies.
| Score | What earns it | Typical shape of the answer |
|---|---|---|
| 2 | Specific and independently checkable | Names, numbers, addresses, a demonstration you can run |
| 1 | Specific but only verifiable by trusting them | A clear policy stated in writing with nothing to check it against |
| 0 | Vague, deflected or unanswered | Reassurance, a change of subject, or a promise to explain on a call |
Twenty-five questions gives a maximum of fifty. Below thirty, the seller has not demonstrated they operate what they are selling and the sensible move is to walk. Between thirty and forty, proceed with a small first run and re-score afterwards using what you observed. Above forty, you have a seller worth dealing with, which is a different claim from having a tool worth buying.
Score group B separately as well. A total in the forties with a zero anywhere in group B is worse than a total in the low thirties with group B clean, because the group B failures are the ones that lose principal rather than value for money. If you are comparing several quotes at once, the ranking exercise resembles what a page on picking the best Solana volume bot has to work through, with the same caveat: the criteria are transparent, and the conclusion is still yours.
Replies that end it
Four answers that end the evaluation whatever the score
"Keys stay with us, it is more secure that way." It is not more secure for you. It means the maximum you can lose is everything you fund, decided by somebody else.
"Results are guaranteed." Nothing on a public chain guarantees inclusion, price or outcome. This claim is only made by people who know it is not true.
"We cannot share addresses, clients are confidential." Token addresses are public. Confidentiality is not the reason, and the real reason is the finding.
"Pay now, the slot is going." Time pressure is a sales instrument. Nothing about this market expires in twenty minutes, and any offer that does is not one worth taking.
What a good answer sounds like
It is worth knowing the positive pattern as well as the failure modes, because the difference is recognisable within two or three replies. Good answers are short, contain nouns and numbers, and frequently include a limitation the seller was not asked about.
A good answer to the venue question names venues and says what happens at migration. A good answer on fees itemises and admits which parts vary with network conditions. A good answer on failure describes the state you are left in without softening it. A good answer to "what can it not do" arrives immediately, because somebody who runs the tool has watched it not do things.
The public equivalent of a good answer is documentation you can read before contacting anyone. When a Solana volume bot pro console publishes its venue coverage, its fee model and what a run puts on the chain, it has answered most of group A and much of group C in advance, and you can use that as the standard a private quote has to meet. Published answers are also answers somebody can be held to, which is why fewer sellers publish them than could.
Getting it in writing
Written answers matter for two reasons that have nothing to do with distrust. The first is accuracy: a seller writing an answer checks it, while a seller saying it improvises. The second is that a written claim is a reference point when something diverges from it, and divergence is common even with honest sellers.
Ask for the whole set as a message rather than a call. If a seller insists on voice, write your own summary afterwards, send it, and ask them to confirm it is correct. Almost everyone confirms, and the confirmation does the same job. Refusal to confirm a summary of their own statements is a strong signal in itself.
Keep the replies somewhere you can find them after the run. The most useful moment for this record is three weeks later, when a fee appears that nobody mentioned and the question is what was actually agreed.
Deciding not to buy
The list is designed to produce a decision, and one of the available decisions is that nothing here matches the problem. That outcome is more common than the market would prefer, and it is a legitimate result of a careful hour rather than a failure of the exercise.
- Re-read your own problem sentence. The one you wrote before contacting anybody. If the tool does not address it directly, the score does not matter.
- Check whether the timing still applies. Several classes only act at a moment that may already have passed, and no seller will volunteer that.
- Compare the total cost against the budget you actually have. Not the one that appears if the launch goes well.
- Ask what happens if you do nothing this week. Frequently the answer is very little, which is useful information about how urgent the purchase really is.
- If you proceed, start small. A minimal first run with keys you hold produces more reliable information than any amount of further questioning.
Whatever the decision, the record of the exercise is worth keeping. The next time somebody offers a launch tool, the same twenty-five questions apply, the scoring works the same way, and the hour becomes twenty minutes. The class notes elsewhere on this shelf explain what each answer should look like for a particular kind of tool, and the comparison method sets out how those classes were defined in the first place.
Questions buyers ask about this class
What is the single most important question?
Whether every private key can be exported on demand, tested on a throwaway wallet before you fund anything real. It is the only question whose wrong answer can cost you the whole balance rather than the purchase price, and it takes about a minute to verify rather than to discuss.
Should I expect sellers to answer all of this?
A serious one will answer most of it quickly, because the answers are things they already know. You are not asking for trade secrets. Venue coverage, fee itemisation, key handling and what happens when a run fails are operational facts, and reluctance to state them is itself a finding.
What if a seller refuses to put anything in writing?
Treat that as the answer. Written replies cost nothing when they are true, and every reason offered for avoiding them is a reason that would not apply to somebody who intended to deliver what they described. A voice call with no record is a conversation you cannot refer back to.
Do these questions apply to free tools too?
Yes, and in some ways more. A free tool still holds keys, still spends your SOL on fees, and still decides what happens to residual balances. The absence of a subscription removes one cost and none of the risks, and it also removes the leverage a paying customer has when something goes wrong.
How do I check a claim about past work?
Ask for token addresses rather than screenshots, then read the relevant window yourself in a public explorer. Screenshots prove nothing at all and take seconds to fabricate. An address is checkable by anyone, which is exactly why a seller who has done the work will give you one and a seller who has not will explain why they cannot.
Is a higher price a signal of quality here?
No. Price in this market reflects sales effort more than capability, and the most expensive quote is frequently the one with the least verifiable answers. Score the answers, not the invoice, and remember that a cheap tool with real key export is safer than an expensive one without it.
What if I have already paid and now have doubts?
Test the export path immediately on a wallet with a small balance, before any larger run. If keys come out, most of the downside is contained and the rest is a question of value for money. If they do not, treat the funded balance as the amount at risk and size everything from that point accordingly.
Written by The Tool Shelf Desk. Figures on this page are either protocol constants or arithmetic labelled as illustration, and no page here reports a measurement the desk did not make. How a class gets defined, and what this desk refuses to do, is set out in the comparison method.