How to get a software product listed on G2: eligibility, submission and verification, with a dated check you run yourself

You get a software product on to G2 by creating or claiming its product profile, completing the product information G2 asks for, and submitting it to G2's own review, which is a human moderation step you do not control. The part nobody can honestly promise you is how long that review takes. A time to a claim decision is a fact only if somebody timed it, wrote down where the clock started and stopped, and repeated it enough times to have a distribution. As of 29 September 2026 we have run zero timed G2 claim submissions, so we publish no median and no average, and we will not present a single submission as a median if we run one. What this page gives you instead is the full protocol, the fourteen fields to log, the denominator arithmetic written out, and the nearest real fraction we do hold, which comes from legal research software sold in India and is not a G2 number.

The answer, first

Four statements, each one checkable against the rest of this page.

The submission itself is short. The waiting is the unknown. Filling in a product profile is an afternoon. What happens after you press submit is a queue you cannot see, and every article that gives you a confident number for that queue is either reporting one person's experience as a rule or repeating another article.

Our count of timed submissions is zero, as of 29 September 2026. Not one, not a handful. So there is no median on this page. One observation is not a median, and two are not either. A median needs a set of results you can order, and the number of results in that set has to be printed next to it.

We do not restate G2's own rules as dated quotations. Help pages get edited without an announcement, and a paraphrase of one with no date next to it becomes a false claim quietly. The page to read is G2's own vendor help material on creating and claiming a product profile, and G2's community guidelines for reviews. Open them on the day you read this, record the URL, the date and the wording you relied on, and keep a copy.

A listing is not the same thing as being cited. Getting on to a directory is a distribution step, and whether that directory reaches an AI answer about your category is a separate question with its own measurement. Ours, for one sector, is below and it is not encouraging.

How it was measured

This is the protocol we would run, written so that you can run it without us and so that two people running it separately can compare results. It is not a description of something we have completed.

One. Freeze the definition of the event you are timing. Almost every disagreement about how long a directory takes is a disagreement about the clock. Write down, before you submit anything, which of the following four moments counts as the start and which counts as the stop.

• Start A: the moment you press submit on the profile or claim form.
• Start B: the moment you receive an automated acknowledgement, which can be hours later.
• Stop A: the moment a decision reaches you by email, whatever the decision is.
• Stop B: the moment the profile is publicly visible to a signed out visitor, which is a different moment from the decision.

We use Start A to Stop A, because both ends are timestamped in your own inbox and neither depends on a page refreshing. If you use a different pair, print which pair on the same line as the number.

Two. Log fourteen fields per submission. A row without these is an anecdote, not a data point.

• Product name and the exact profile URL, once it exists.
• Whether this was a new profile or a claim of a profile that already existed.
• The submission timestamp, in ISO form, with the time zone.
• The acknowledgement timestamp, or blank if none arrived.
• The decision timestamp.
• The decision itself, quoted in the words used.
• Every field the form required, listed, so a later reader knows what the form asked that month.
• Whether any verification step was requested, and what evidence it asked for.
• Whether a follow up message from you was needed before the decision, and when it was sent.
• Whether the account used was a work email address on the product's own domain.
• The category or categories requested.
• Any rejection reason, quoted.
• The date you read the official guidance, and its URL.
• A saved copy of the guidance page as it read that day.

Three. Decide the repetition count before you start, and print it. One submission gives you one duration. Five submissions across five products over three months gives you five durations, which you can order and describe as a median of five. That is still five. State it as a median of five, never as the time G2 takes.

Four. Do the denominator arithmetic on the page. Write it out so a reader can see the size of what you are claiming. If you submit 5 products and 4 receive a decision within your observation window, the reportable line is 4 of 5 decided by the cut off date, with the cut off date named, and the fifth row marked as still open rather than dropped. Dropping the open row is how a reported median gets shorter than reality.

Five. Record anything the platform tells you about its own process as a claim, not as a fact. This applies to time estimates in a help page, to a support agent's reply, and to anything an AI assistant tells you about a directory's rules. The reason is measured: in the aiknowsus.com audit of September 2026, Perplexity withdrew its own earlier statement when asked how many of the questions it had actually searched for, saying it could not honestly substantiate the claim that it had run a live search for each one, and in another batch that its claim to have searched all five was not adequately supported. Three separate batches produced that retraction. A system's account of its own work is evidence about what it said, not about what it did.

Six. Keep the blind rule if you also test visibility. If part of your reason for listing is to appear in AI answers, then the questions you use to check that must not contain your brand name. On 27 July 2026 two runs of the same 78 questions happened on the same day for clawlaw.in. The run whose wrapper named the brand came back ranking it first on almost every question. The blind run put the company second by breadth and absent altogether from the litigation due diligence questions it most wanted to win. The first run was discarded, because the only thing it had measured was our own prompt.

What the numbers were

Timed G2 claim submissions completed by us: zero, as of 29 September 2026. So: no median, no average, no range, no borrowed figure from another article. The cell is empty and it stays empty until the protocol above has been run and the rows published.

The nearest real fraction we hold, with its sector named. Across eighteen blind commercial questions on 18 August 2026, ChatGPT made clawlaw.in the top source on exactly one of the eighteen, and named the company's own comparison page in its answer as still being the vendor's own editorial page. That is 1 of 18. The sector is legal research software sold in India. It is not a G2 figure, it is not a figure for any other sector, and it must not be moved on to a page about one.

Three further dated results that bear on whether a directory listing reaches an answer.

• 0 of 6. Across six questions in the clawlaw.in programme on 17 September 2026, no third party review or directory source made it into any answer at all. Claude's run. One well known review site did appear in the raw results and was discarded, because the list it offered was of American products and so was not an answer to an India question.
• Official pages and vendors' own pages dominated the citations. Across the whole aiknowsus.com capture of September 2026, 24 batches and 72 conversations, the domains cited most often were the assistants' own documentation and the vendors' own websites, with a single well known review site far down the list. Perplexity.
• 161 recordings of not being cited. Across those same 24 batches and 72 conversations on our own domain, the phrase recording that we were not cited appears 161 times in the engines' own self audits of their answers.

Read together, those say something useful about why you would list at all. A directory profile is worth having because buyers look at directories and because it is another public page carrying your facts. In the one sector we measured, in the months we measured, it was not the thing that decided the AI answer.

What this cannot tell you

• It cannot tell you how long your claim will take. Nobody outside the moderation queue can, and a number from 2024 describes a team that has since changed size.
• It cannot tell you whether you will be approved. Approval depends on evidence about your product that we cannot see from here.
• It cannot be compared with somebody else's timing claim unless they published their clock definition, and almost nobody does.
• Our 1 of 18 and 0 of 6 are legal technology in India. They are the nearest real fractions we hold and they are not evidence about software categories in other markets.
• No part of this can promise you a position in an AI answer. We do not sell that and we would not believe anybody who did.
• A directory profile is not under your control. The category names, the comparison set and the display can change without telling you, which is a reason to keep the same facts on a page you own.

Sources and change log

Sources. The timing protocol and the fourteen fields are ours. The dated results come from two recorded bodies of work: the clawlaw.in programme from July 2026, in Tier_1/GEO_BASELINE_RESULTS_2026-07-27.md, Tier_1/GEO_GAP_ANALYSIS_2026-08-18.md and Tier_1/claude_response_17_09_audit.md, and the aiknowsus.com audit of September 2026 across 24 batches and 72 conversations. The counts were produced by a script over those files rather than from memory. No figure on this page describes G2, because we have not measured G2.

Change log. 29 September 2026, first publication, stating zero timed submissions. This page will be updated when the first timed submission is logged, at which point it will carry the count of submissions, the clock definition, the individual durations and the open rows, in that order. It will be updated again if any figure above is corrected, and when the capture files behind the counts are published, which will move several rows from uncheckable to checkable.

Common questions

Why will you not just tell me the usual approval time?

Because we have not timed one, and a number we did not measure would be indistinguishable on the page from one we did. The cost of publishing it is not embarrassment, it is that the figure gets copied into other pages and becomes a fact with no origin. Run five submissions, log the fields above, and you will own a figure that is true for your account.

Is a free listing enough, or do we need to pay?

Read G2's own vendor pages on the day you decide, record what the free profile includes and what a paid tier adds, and write the date and URL next to each line. We deliberately do not restate any platform's prices or tiers here. On 17 September 2026 two clawlaw.in comparison pages stated competitors' prices with no link, no date and no source. Claude checked the figures independently, found them correct, and still treated the pages as advocacy and used official sources instead. Correct and undated was not good enough.

Does a G2 profile help an AI assistant find us?

It gives an assistant one more public page that states your facts, which is worth something. In the one sector where we measured it, it was not decisive: across six questions on 17 September 2026, no third party review or directory source reached any answer at all in Claude's run. Treat the listing as a distribution step and measure your visibility separately, blind.

What is the single most common reason a claim stalls?

We have no measured answer, so we will not rank reasons. What we can tell you is what to log so that your own answer arrives: the required fields as the form presented them, whether a verification step was requested, and whether the account used a work email on the product's own domain. Those three columns explain most stalled rows in any submission log we have kept.

Our product is new. Are we eligible at all?

Eligibility is set by the platform and stated on its own pages, so read those rather than any summary of them. What you can prepare in advance is the evidence pack: a live product page, a working sign up or demo route, a public price or a stated pricing model, a support contact, and a description of the category in the words buyers use. Those are needed for almost every directory, and they are needed for your own site regardless.

Should we ask an assistant to summarise the listing rules for us?

Use it to find the pages, not to state the rules. Record whatever it tells you as a claim with the date attached, then check it against the official page. In our own September 2026 audit the assistant withdrew its claim to have searched in three separate batches. A rule with no URL and no read date is not a rule you can rely on.

What to do first

Open G2's vendor help pages today, save a copy, and write the date and URL at the top of a new sheet. Create the sheet with the fourteen columns above and one row per product you intend to submit. Fix your clock definition as Start A to Stop A before you submit anything. Then submit one product and let the log fill itself, so that in three months you hold a small, honest set of durations instead of a sentence you copied.