Directory Submission Checklist

Who this checklist is for

This page is the operating manual for submitting a SaaS product, AI tool, or early startup to directories without turning the week into a pile of half-filled forms. It is written for founders who will do the first twenty listings themselves, for marketers who have to hand a contractor a kit that does not leak passwords, and for agencies that need a client-safe tracking sheet plus proof. It is not a ranked list of directories. The live table lives on the homepage at #directories. This page is the work that happens before, during, and after those rows.

You should read it if you already opened a giant spreadsheet of "AI directories 2026" and realized you have no square logo, three description lengths, or a dedicated mailbox. You should also read it if you bought a done-for-you pack once, got a CSV of URLs, and still cannot say which listings are live, dofollow, or even about your product. If you only need a Product Hunt-style weekly leaderboard, this is the wrong page. Launch there, then come back here for directory coverage.

SubmitLoop sells two one-time packs on the submission service page: Light at $49 for 60 guaranteed directories (one site), and Growth at $99 for 177+ directories (one site, proof or refund). This checklist is how you DIY the public catalog, how you prepare a kit so a pack does not stall, and how you decide the hour you should stop typing into forms. Light works out to about $0.82 per directory in the guaranteed pool. Growth works out to about $0.56. Proof on both packs is a run log plus a screenshot or trace per claimed success. Growth refunds if claimed successes lack that proof. Neither pack promises a Domain Rating number. Neither pack is a substitute for a missing logo.

Printable asset lists by niche live in the submission checklist generator. Use that tool to get a client-side checklist, then keep this article open as the procedure. The generator tells you what to gather. This page tells you what "gathered" actually means, what to refuse to store in a sheet, and when the remaining queue is cheaper as Light or Growth than as founder evenings.

What done means before you open a form

Most directory sprints fail because nobody defined done. "Submitted" is not done. A thank-you page is not done. A row in a sheet that says "sent" is how agencies get fired. Define four layers and log them separately.

Layer one is form accepted: you completed the fields the site asked for, solved or skipped captcha honestly, and the site did not bounce you to a paywall you refused. Layer two is identity confirmed: if the directory emailed a verify link, you clicked it from the mailbox you control. Layer three is live URL: a public page exists that a stranger can open without logging in as you, and it describes your product rather than a placeholder. Layer four is QA: you checked the live URL for the right name, the right site, the right category, the link type if you care about SEO, and whether the outbound URL still points at you.

You will often sit in layer one for days while an editor works a queue. That is normal. Do not refresh twenty times an hour. Do not email the directory "any update?" on day one. Log the date of form accepted, set a follow-up date (fourteen days is a sane default for moderated indexes), and move to the next row. Founders who treat pending as failure start duplicate submissions, which some sites treat as spam.

Ahrefs DR and Moz DA are prioritization shortcuts, not success criteria. Google does not use those scores. A live, relevant, indexed listing on a site your buyers actually browse beats a DR 70 blogroll that never mentions your category. Keep scores in the sheet as sort keys. Do not write "DR up 4 points" as a weekly goal for a directory campaign. Recrawl lag makes that number move on Ahrefs' calendar, not yours.

For a concrete case, a two-person analytics SaaS submitted forty-two directories in a weekend, marked every row submitted, and posted a "we're live on 42 directories" tweet. Six weeks later, eleven URLs 404'd, eight were still "pending review," nine had listed a competitor because the form used a similar name field, and four had rewritten the description into generic "AI-powered platform" copy that could have been anyone. The move that usually works is treat live URL plus a screenshot as the only number you report to a cofounder or client. If you skip that, you will celebrate labor, not listings.

How to use this page with the catalog and the generator

Work in this order, even if it feels slow on day one.

First, open the submission checklist generator, pick SaaS, AI, or startup, and export or copy the asset list. Second, build the kit described below until every required item exists as a real file or a real paragraph, not a "TODO in Notion." Third, open the public directory catalog, filter to Guaranteed if you want the easy-form pool, or to SaaS / AI / Launch if you are matching category, then sort DR if SEO is the point. Fourth, copy the rows you intend to touch this week into your tracking sheet. Fifth, submit in tiers. Sixth, QA live URLs on a calendar, not in the same sitting as the form. Seventh, decide whether the leftover guaranteed pool is still worth your hours or whether Light or Growth is the cheaper close.

Do not invert the order. Submitting first and gathering assets mid-form is how you abandon a site with a half-upload and a session cookie that expires. Filtering the catalog before the kit is how you discover at 11pm that every remaining high-DR row wants a 1200 character description you never wrote.

If you are an agency, add a client sign-off step between kit and queue. The founder should approve the short description, the long description, the category list, and the screenshot set. Silent contractors invent taglines. Those taglines ship to fifty directories and become the public story of the product. Fixing that later is worse than waiting a day for a Slack thumbs-up.

If you are a solo founder with a day job, cap the first sprint at fifteen directories, all from the guaranteed filter, all guest-possible if you can help it. Finish those to live-URL QA before you open login-wall rows. Completeness on a small set teaches you the kit gaps. A hundred-tab blast teaches you nothing except that captchas exist.

The kit is the product of directory week

The kit is a folder plus a copy doc plus a password-manager vault. The pack, if you buy one, is labor on top of the kit. The catalog is a queue. Teams that skip the kit treat directories like a typing contest. Teams that finish the kit can pause for a week, hand the folder to a contractor, or buy Light, and the work still has a source of truth.

Name the folder after the public product name, not the legal entity, not the internal codename. Put it in a place the people who submit can actually open. If your designer lives in Figma and your submitter lives in Google Drive, export files. Directories do not load Figma frames.

Minimum kit contents:

• Square logo, PNG, 512px or larger, simple background
• Optional transparent PNG of the same mark
• Horizontal lockup if you have one (many forms ignore it; keep it anyway)
• Favicon 32px and 180px apple-touch if you will also update social cards later
• Three screenshots: dashboard or core UI, empty state, and one "aha" view
• Optional 15-second silent screen recording (mp4 or webm, under 20MB if you can)
• OG image you already use on the marketing site (some "featured image" fields accept it)
• Copy doc with name, URL, tagline, short / medium / long descriptions, pricing posture, categories, not-for line, integrations list, countries, languages
• Support URL, privacy URL, terms URL, docs URL if they exist
• Founder or company email that can receive verification (not a no-reply)
• Social profile URLs you are willing to show publicly
• Password manager vault named for this product, empty of sheet-sync

That list is the generator's job in short form. The rest of this article is how each item fails in the wild.

Reuse the kit for Product Hunt, ScrollLaunch, ads, and sales one-pagers. If a directory asks for an extra field (founding year, employee count, "built with"), add it to the copy doc rather than arguing with the form. You only want to invent facts once.

Asset pack: logos that survive thumbnails

Directories crush logos. They crop to circles, they letterbox on navy backgrounds, they generate 64px thumbnails next to fifty other marks. A detailed illustration with tiny wordmarks becomes a smudge. Design for the worst thumbnail, not for your homepage hero.

Use a square canvas. 512 by 512 is the floor. 1024 by 1024 is better if the mark is simple. PNG-24 with a solid background usually beats a transparent file that gets placed on white, then on black, then on a random pastel card. If your brand is a wordmark only, make a square version with the first letter or a simple icon. Do not upload a 2000 by 400 banner into a square slot and hope.

Avoid light gray on white. Avoid thin strokes that vanish at 48px. Avoid screenshots of the logo sitting in a browser tab. Avoid putting the year or "AI" as a badge inside the mark unless that badge is the actual brand. Avoid uploading the app icon from iOS with rounded-rect already baked in if the directory will round it again (you get squircle-in-squircle).

Keep a "directory logo" export even if it differs slightly from the brand book. Brand books assume print and hero. Directories assume a table. If legal insists the official mark must include the full company name in 8pt type, that mark will fail. Make a directory-safe sibling and document that it is allowed for listings.

For a concrete case, a climate-tech team uploaded a detailed Earth illustration with a four-word slogan under it. On six directories the slogan became unreadable and the Earth looked like a green blob. They swapped to a bold monogram on a dark green square and approval rates on logo-required forms went up because reviewers could actually see a brand. The move that usually works is squint at the file at 32px in your file manager. If you cannot recognize it, the directory will not either.

If you have a light and dark version, pick one for the whole campaign so your featured-on strip later looks consistent. Switching mid-queue makes the homepage look like a scrapbook.

Do not "improve" the logo between directory number three and directory number forty unless you are willing to go back and replace files on the first three. Many sites never offer an edit path. Inconsistency is a real cost.

Asset pack: screenshots, empty states, and the aha view

Forms that ask for screenshots are not being fancy. Reviewers use them to decide if you are a real product or a landing page with a waitlist and a fake dashboard. A marketing site screenshot of a gradient hero is the fastest way to look unfinished.

Capture three classes of image.

Dashboard or core UI: the screen a paying user lives in. Crop browser chrome if it includes your localhost URL, an ngrok host, or a staging badge. Hide real customer names. Hide revenue numbers if they are not yours to publish. Hide the internal debug bar.

Empty state: the first-run screen with a clear call to action. Directories love empty states because they explain the product without a data dump. If your empty state is a blank table that says "No data," fix the product, then screenshot. "No data" in a directory thumbnail says the product does not work.

Aha view: the moment the product proves itself. For an AI writer, that might be a finished draft with a visible prompt. For a SaaS analytics tool, a chart that a buyer would recognize. For a developer tool, a terminal or IDE panel with a successful run, not an error stack unless the error is the product.

Technical notes that waste hours: WebP is not universally accepted. Keep PNG or JPEG copies. Some forms cap at 2MB. Some want 1280px on the long edge. Some want 16:9. Export a "directory screenshot" set at 1600px wide JPEG quality 80 and a PNG backup. Do not upload 4K PNG files that timeout on cheap hosts.

Avoid light-gray UI on white that looks blank when the directory compresses. Increase contrast for the export even if production UI is airy. This is a thumbnail, not a Dribbble shot.

If the product is an API with no UI, screenshot the docs getting-started page, a code sample with syntax highlighting, and a dashboard for API keys if you have one. If you truly have nothing visual, record the 15-second clip of a request and response. Do not submit three identical crops of the homepage.

Mobile screenshots: only if the product is actually mobile. A responsive marketing page is not a mobile app. App directories will bounce you.

For a concrete case, an AI meeting-notes startup submitted screenshots of a Figma prototype. Two editorial directories rejected them for "product not live." They recaptured production with a real workspace name (their own company) and the same forms accepted the listing. The lesson is that reviewers can tell. If you skip production captures, you are submitting fiction.

Asset pack: short video, GIFs, and when to skip them

Video is optional on most classic directories and required on a few modern indexes. A 10-20 second silent loop of the aha view is enough. Do not produce a two-minute brand film. Directories will not watch it. Host the file locally in the kit. Some forms upload. Some want a YouTube or Loom URL. Keep both: a file under 20MB and an unlisted YouTube link that does not autoplay ads over your demo.

Do not watermark "DRAFT." Do not include loud music. Do not include a voiceover that names a different product from a pivot. Do not use stock footage of smiling people in open offices. That footage is how you look like a template.

GIFs: useful when the form has no video field but has an extra image slot. Keep them under a few megabytes. A six-frame GIF of a toggle is better than a 12MB animation that fails to upload.

Skip video entirely if recording will delay the kit by more than half a day. Sixty listings with no video beat two listings with a cinematic trailer. You can add video later on sites that allow edits. You cannot recover a week you spent color-grading.

Copy versions: name, URL, tagline

Product name: the string on the homepage H1 or the nav, not the legal LLC name, not the GitHub org, not the domain without vowels. If you rebranded last month, pick the new name and use it everywhere. Dual names ("Acme, formerly PixelTools") belong in the long description once, not in the title field. Title fields get truncated.

URL: the marketing site that a directory visitor should open. Not the app subdomain if the app is behind login and looks empty. Not a UTM-less production URL if you plan to track (more on UTMs below). Not a Google Doc. Not a Notion public page unless that is truly the product. If the site is still behind basic auth, stop the campaign. Every vendor and every DIY session will stall, and you will blame directories for a gate you installed.

Tagline: about twelve words. Buyer plus job plus output. "AI-powered platform for modern teams" is not a tagline. It is a slot machine of leftover adjectives. "Turns customer calls into CRM notes for B2B sales teams" is a tagline. Count the words. If you cannot say it out loud without breathing, it is a paragraph, not a tagline.

Write two taglines if you sell into two buyers (for example developers and marketing). Put the primary in the default field. Use the secondary only on directories whose audience is obviously the second buyer. Do not A/B test taglines across fifty sites in the same week. You will not get clean data. You will get a messy public record.

Forbidden tagline moves: stuffing the company city, stuffing "Y Combinator W24" unless the form has a field for it, stuffing keyword lists, stuffing a URL, stuffing emoji. Some sites strip emoji. Some display them as tofu. None of that helps reviewers.

Copy versions: short, medium, and long

You need three lengths because forms are chaotic. Some cap at 140 characters. Some want 300-600. Some want 1000-2000. Some give you a TinyMCE box and a "be thorough" hint with no counter. If you only write one paragraph, you will spend the sprint resizing it badly, cutting mid-sentence, or repeating the tagline until it sounds like spam.

Short (about 300-400 characters, also keep a 140-character sibling): name the buyer, the painful job, and the output. One sentence can do it. Two is the maximum. No origin story. No "we are passionate." No feature laundry list.

Medium (about 600-900 characters): short plus how it works in one pass, plus who it is not for in one clause, plus pricing posture in one clause ("free to start," "usage-based," "seat licenses from X," "waitlist"). This is the version most SaaS directories actually want.

Long (about 1200-2000 characters): medium plus a paragraph a human can skim on integrations, security posture if you sell to companies that ask (SOC 2 in progress is a fact, "bank-grade" is not), and a concrete workflow. End with a clear next step: start free, book a demo, or join the waitlist. Match the site. Do not put "book a demo" on a directory of free indie tools if you have no self-serve.

Write these in a doc with character counts visible. Paste into a tester that counts characters, not words. Directory limits are usually characters. Word processors lie.

Customize the first sentence per audience, not the whole essay. AI directories want to know the model posture (wrapper, fine-tune, own model, or "we call several providers"). Developer directories want the language and the install. HR SaaS directories want the buyer title. Changing the first sentence is enough. Rewriting the whole long version fifty times is how copy drifts and contradicts pricing.

For a concrete case, an indie AI image tool used the same 1800-character essay on a general startup directory and on a "new SaaS this week" list. The general directory published it. The SaaS list rejected it as "not SaaS" because the essay never said there was an account, a plan, or a dashboard. They added one sentence about workspaces and plans to the medium version and resubmitted where allowed. The move that usually works is keep a one-line "what kind of product is this" at the top of medium and long: SaaS with seats, AI tool with credits, marketplace, API, or downloadable app. If you skip the type line, reviewers guess, and they guess wrong.

Do not let ChatGPT generate a new essay per directory from scratch. The model will invent features, change pricing, and add "trusted by 10,000 teams" you cannot defend. Use the approved doc. A light first-sentence swap is the only generation you should allow, and a human should read it.

Copy versions: pricing, plans, integrations, and the not-for line

Pricing fields are where honest products look unfinished and dishonest products look finished. If you are waitlist-only, say waitlist. If you have a free tier, say what is limited. If you have no public price, say "contact for pricing" only on directories that serve enterprise buyers. On indie lists, no public price reads as vapor.

Do not paste your entire pricing page HTML. Do not paste a table that breaks in a plain textarea. Write three lines: free includes X, paid starts at Y and adds Z, enterprise is custom. If prices are in EUR and USD, pick the currency of the directory's audience or state both in short form.

Integrations: a comma-separated list of the ones you actually ship. "Integrates with everything via Zapier" is allowed if Zapier is live. Listing Slack, Notion, and Salesforce because they are on the roadmap is how you get angry directory traffic. Keep a "shipped integrations" list in the kit, dated.

The not-for line saves you from bad-fit traffic and from category mismatches. "Not a general chatbot. Not a replacement for your IDE. Not for consumer photo fun." One of those sentences belongs in the long description. Reviewers on niche lists use it to accept you. Buyers use it to leave, which is also a win.

Support and legal URLs: many forms ask. If you do not have a /privacy page, add a simple one before the sprint. A missing privacy URL is a hard stop on a surprising number of European-leaning indexes. A missing support email is a hard stop on older PHP directories that still think they are running a web ring.

Founding year, team size, HQ city: only if true. Do not pick 2026 because this is the year on the calendar if you incorporated in 2024. Do not pick "1-10" if you are a solo founder and the next option is "just me." Accuracy here is easy and some directories display it.

Category mapping for SaaS, AI, and startups

Categories are not SEO tags you stuff. They are how a human moderator files you, and how a visitor browses. Pick one primary. Pick a secondary only if the form allows it and both are true. Three categories is the maximum you should ever want. Five is a spam signal.

Primary for most readers of this page is one of: SaaS, AI / machine learning, developer tools, marketing, productivity, or "startup" if the directory is a launch index with no verticals. If you are an AI feature inside a SaaS workflow, do not default to AI on a general SaaS directory. File under the job (marketing, support, analytics) and mention AI in the description. File under AI on AI-only indexes. Dual-primary is how you look unfocused.

Never pick "other" if a closer bucket exists. Never pick the paid featured category by accident. Never pick a geography category for a country you do not serve. Never pick "crypto" because your landing page has a wallet joke.

Build a small map in the copy doc:

• Canonical primary
• Canonical secondary
• AI-directory primary (often "AI productivity" or similar)
• Launch-directory primary (often "SaaS" or "productivity")
• What you will not select (the tempting wrong buckets)

When a form's taxonomy is alien (they have "bots," "agents," "copilots," "LLM wrappers" as separate boxes), read the definitions if they exist. If they do not, pick the box that matches how a buyer would browse, not how your pitch deck is titled. An agent product that is used as a copilot in a human workflow may belong in copilots on that site. You can explain nuance in the long description. You cannot explain nuance in a dropdown.

Language and country filters on the catalog matter more than founders think. A US English AI copilot listed on a Turkish general web directory rarely converts and often fails category checks. A climate-tech team once burned a day on EU local chambers that required a VAT number they did not have. Filter language and country first, then DR. If you skip it, your "100 submissions" include 40 forms you could never complete.

For review-style sites (the ones that look like G2 cousins), category is closer to a buyer intent taxonomy. "Best X software" pages are built from those categories. A mismatch here is not just a filing error. It is a page you will never appear on. Spend the extra minute.

If the form requires a category ID that only appears after you pick a parent, do that parent first and wait for the child list. Do not paste a number from a previous site. IDs are not portable. This checklist will not give you site-specific selectors. Semantic intent is the portable part: you are looking for the control whose label is category, marketplace, or industry.

Tracking sheet: columns that matter

The sheet is an operations object, not a scrapbook. One tab per product. One row per directory domain (not per URL variant). If you submit to example.com/submit and later find example.com/tools/submit, that is still one row unless they are truly different properties.

Suggested columns:

• Directory name
• Domain
• Catalog slug or notes from SubmitLoop if you use the public table
• Submit URL you actually used
• Tier (1, 2, 3, skip)
• Pricing on the directory (free, freemium, paid fee amount)
• Link type if known (dofollow, nofollow, unknown)
• DR / DA as of the day you queued (optional, snapshot them)
• Language / country
• Account required (yes/no)
• Captcha (none, checkbox, image, other)
• Email verify (yes/no)
• Status (see the vocabulary below)
• Date queued
• Date submitted
• Date verified email
• Date live
• Live listing URL
• Outbound URL on the listing (yours, with or without UTM)
• UTM used
• Featured-on eligible (yes/no, after QA)
• Owner (you, contractor, SubmitLoop pack)
• Evidence link (folder of screenshots, not a password)
• Last QA date
• Notes (editorial delay, category chosen, "already listed")

That is enough. If you add twenty more columns, nobody will fill them. Color scales and sparkline widgets are how tracking sheets die.

Use data validation for status so people cannot type "done-ish." Freeze the header. Filter views per owner. Do not merge cells. Do not put two directories in one row.

Export a CSV backup monthly. Sheets get wrecked by a sort that did not include all columns. You want a dated dump in the kit folder.

If you use the SubmitLoop catalog as the source of the queue, copy DR, pricing, and guaranteed tags at queue time. Those numbers change. Your sheet is a snapshot of why you chose the row, not a live replica of the catalog. When you return next quarter, re-check the catalog rather than trusting last quarter's DR cell as physics.

Tracking sheet: status vocabulary

Pick a closed set. Suggested:

• queued: in this week's list, not opened
• blocked_kit: opened, missing an asset or copy length
• in_progress: form open in a browser
• submitted_form: layer one done
• pending_verify: waiting on email
• pending_review: verified or no verify, editor has not published
• live: layer three done, URL in the cell
• already_listed: product was there, you claimed or skipped
• paid_only: paywall, you will not pay (or you will, on a different budget line)
• unreachable: site down, parked, or hard bot wall you will not fight this week
• rejected: editor said no, capture the reason in notes
• skipped_irrelevant: you opened it and the audience is wrong
• duplicate_row: merged into another domain
• pack_claimed: SubmitLoop or another vendor owns the attempt

Do not use "submitted" as a grab bag. Do not use green fill without a status. Do not let a contractor invent "probably live."

already_listed is a success for coverage purposes if the listing is yours and correct. It is a failure if it is a scavenger listing with the wrong URL. QA it like a live row.

paid_only is not a moral judgment. It is a budget flag. Move those rows to a paid placements tab with fee, whether a dofollow is claimed, and whether you actually wired money. Mixing "we might pay this someday" into the free queue is how invoices surprise you.

unreachable should be retried once on another network or after a day, then parked. Three hours on a Cloudflare loop is not grit. It is a bad hourly rate.

Tracking sheet: no passwords, ever

This is the rule that saves you from a Slack leak, a contractor offboarding mess, and a client security questionnaire.

The sheet may contain: the login email or username as a reminder of which identity you used, a password-manager item name (for example "dirs/acme/betalist"), and whether Google OAuth was used. The sheet may not contain: passwords, TOTP seeds, backup codes, session tokens, API keys, IMAP passwords, or screenshots of a password manager.

Put secrets in a vault: 1Password, Bitwarden, or whatever your company already uses. One vault or one tag per product. If a contractor needs access, share the vault or the items, not a tab called "logins." When the contractor leaves, revoke the vault. You cannot revoke a spreadsheet version history.

If a done-for-you vendor emails you a CSV of passwords, rotate anything that reused a personal password, import the rest into the vault, and delete the CSV from email after import. Ask them not to send passwords in plaintext next time. SubmitLoop's proof model is a run log plus screenshot or trace of the form, not a credential dump as the core deliverable. If you need the account because the directory created one, store it in the vault and record only the item name in the sheet.

Google OAuth and magic links: prefer them when the directory allows, because there is no extra password to mishandle. Record "OAuth-Google" in the account-required column. Still use a dedicated mailbox so verification mail is not mixed with your primary inbox chaos.

Never put the vault master password in the sheet "just for the team." Never put IMAP credentials in the sheet so a friend can "help verify." Never screenshot a filled password field for evidence. Evidence is the thank-you page, the live listing, or the verify email subject, not the secret.

For a concrete case, a freelance SEO sent a client a Google Sheet that included passwords in column H. The client's security lead found it in a routine shared-drive audit, killed the engagement, and rotated a dozen accounts. The listings were fine. The process was not. The move that usually works is a sheet without secrets plus a vault with a naming scheme the sheet can point at. If you skip it, you will eventually paste a password into a ticket.

Tiers: how to order the week

Tiers stop you from random-walking the catalog. They are not a moral ranking of directories. They are a labor ranking.

Tier 1: high relevance, high enough DR if SEO matters, dofollow when you can get it, audience overlap with your buyers, form you can complete with the kit. Finish these first even if they are slower (login wall, email verify, a few extra fields). Ten great Tier 1 lives beat forty Tier 3 blogrolls.

Tier 2: the guaranteed / easy-form pool. This is the set SubmitLoop Light ($49 / 60) and Growth ($99 / 177+) are built to grind with proof. Guest forms, sane field counts, fewer mid-form $59 upsells. DIY these if your hourly rate is low and you like forms. Buy them if your hourly rate is not low.

Tier 3: long tail, unknown link type, weak relevance, or high friction you might still want for a brand mention. Do these last, or never. Many founders should never do Tier 3 themselves.

Skip tier: paywall you will not buy, wrong country, dead site, "submit your URL" boxes that are just comment spam, and anything that demands a backlink to the directory from your homepage as a condition of a nofollow listing. Reciprocal homepage links are a different product. Do not mix them into a launch-week directory sprint unless you have a policy for partner pages.

Paid placements are a parallel track, not a fourth tier inside the free queue. Budget them with fees visible (the catalog keeps paid rows public for that reason). A $200 listing on a relevant site can be correct. A $29 listing on a spam network is not "cheap SEO." It is a line item you will forget to disclose to a client.

Sort within a tier by friction: guest before login, no captcha before captcha, no email verify before verify, short forms before wizards. You want early wins so the kit gets tested while you still have energy to fix screenshots.

Do not sort only by DR. A DR 80 general web directory that files you under "computers" next to shareware from 2004 is a weak Tier 1. A DR 45 AI index that your buyers subscribe to is a strong Tier 1.

For a concrete case, a payroll SaaS spent two days on general "startup directories" with DR 60+ and got listings that never ranked for payroll terms and never referred a signup. The following week they did eight HR-specific lists with lower DR and picked up three trial starts. The move that usually works is relevance first, then DR, then friction. If you skip relevance, you will own a trophy case of logos nobody in your ICP clicks.

Captcha without turning the night into a side quest

Captcha is a cost, not a puzzle worthy of your identity. Checkbox reCAPTCHA, image grids, hCaptcha, Cloudflare managed challenges, and "prove you are human" waits all show up. Your job is to complete them when the listing is worth it, and to skip the row when the challenge is a wall.

Do not outsource captcha solving to random browser extensions you just installed. Do not paste challenge screenshots into public Discords. Do not write site-specific click scripts. If you are a SubmitLoop worker in our own pipeline, captcha belongs in the captcha phase of the engine and in the human loop. If you are a founder DIY, you are the human loop. Sit up, solve the checkbox, continue.

If a directory is nothing but Cloudflare loops on every retry, mark unreachable or skip for the week. Try a normal browser profile that you use for real sites, not a fresh automation profile, if you are DIY in a headed browser. Headless automation is how you look like a bot. Our own agent rules refuse headless Playwright for this reason. You should refuse it too.

Invisible captchas that trip on submit are why you should not spray-fill with a password manager and a hope. Fill, pause, submit once, wait. Double-submit creates duplicate listings or duplicate "pending" tickets.

Score captcha in the sheet so next quarter you know which rows cost three extra minutes. That score is how you decide the DIY versus pack cutoff. A queue of sixty easy forms with no captcha is a different product from a queue of sixty image grids.

If you buy Growth, you are paying partly to not sit in image grids. That is legitimate. It is not lazy. It is pricing your attention.

Email verification without wrecking your primary inbox

Many directories still send a confirm link. Some send a login password you must rotate into the vault. Some send both. Treat the mailbox as part of the kit.

Use a dedicated address you control, on a domain you control if possible (dirs@yourproduct.com or a plus-address if the sites honor plus-addresses; many do not). A personal Gmail works if you filter aggressively. A no-reply or a group inbox nobody watches does not work. Founders miss verify links and then swear the directory is broken.

IMAP access is how a pack or an assistant confirms links without sitting in Gmail all day. If you will never grant IMAP, you must click links yourself within a few hours. Log pending_verify and check twice a day. Links expire.

Do not use a throwaway inbox you cannot recover. In six months you will need to edit the listing or confirm a re-verification and the throwaway will be gone. Do not use a shared password on that mailbox. Do not forward all directory mail to a Slack channel that is public to the company. Verification URLs are capability links.

Some sites require the same email as the public "contact" on the listing. Some will display it. If you do not want a personal address harvested, this is another reason for a dedicated mailbox.

For a concrete case, a founder used plus-addressing (name+betalist@gmail.com) and lost three listings because those directories stripped or rejected the plus. They switched to a real alias on their domain and the verify mails arrived. The move that usually works is test the mailbox on the first two verify-required sites before you queue thirty. If you skip the test, you will batch-fail.

When the verify mail is a password reset style "here is your password," change it in the vault immediately. Do not leave the emailed password as the live one. Do not put the emailed password in the sheet.

If a directory requires phone SMS verify, decide in advance whether your product is worth a phone number on that site. Many are not. Skip rather than using a personal mobile on a random PHP board.

Guest forms versus login walls

If the listing form is visible without an account, use it. Guest submit is faster and creates fewer secrets. If the site is a login wall with no visible form, you need an account, which means the vault, the mailbox, and more time.

Do not create accounts "just in case" on every site before you know the form is the listing form. Some logins are for posting comments, not for adding products. Read the page. Look for submit, add listing, add product, get listed. If those CTAs lead to a pricing page with no free tier, you are in paid_only territory.

OAuth with Google is fine if you use a Google account you intend to keep. Do not OAuth with a personal account you will lose when you leave a job. Do not OAuth with a client's personal account if you are an agency. Use the client's workspace identity.

Multi-step wizards: complete the step, re-read the page, continue. Do not assume step two is the same as the last directory. Do not hardcode a sequence in your head from a blog post dated 2019.

If you hit a plan picker, choose free when a free plan exists and is actually a listing, not a 14-day featured trial that converts to $59. If the only way forward is a paid plan, stop and log paid_only with the price. That decision belongs to budget, not to momentum.

UTM conventions that survive editors

You want to know which directories send traffic. You also want the listing to remain correct if an editor strips query strings, wraps your URL in a redirect, or replaces it with the naked domain they found on your homepage.

Recommended pattern:

`https://yourproduct.com/?utm_source=DIRECTORYDOMAIN&utm_medium=directory&utm_campaign=launch2026`

Keep utm_source as the directory's registrable domain, lowercase, no spaces (producthunt is an exception if you also launch there; still be consistent). Keep utm_medium as directory so GA4 and your warehouse can group. Keep utm_campaign as a quarter or a launch name you will not be embarrassed by in two years. Optional utm_content for the pack versus DIY if you care.

Put the UTM URL in the sheet column for outbound. Put the naked URL in the kit as a fallback. If a form says "no tracking parameters" or the preview strips them, use the naked URL. A live listing with a clean URL beats a rejected listing because you argued about query strings.

Do not use a different campaign name per directory. You will never analyze 80 campaigns. Do not use utm_source=submitloop on DIY rows. That will mix pack traffic with your hands. Do not use a link shortener that you do not control. When the shortener dies, the listing points at nothing.

Some directories require the URL to match the homepage exactly, including https and www. Check your canonical. If the site is www and you submit the apex, you may fail a uniqueness check or split analytics. Pick the canonical and use it everywhere, UTMs attached after the path.

QR codes, Bitly, and "smart links" do not belong in directory URL fields. They belong in print. Editors sometimes visit the URL. If they land on a redirect chain through three trackers, you look like an affiliate spam site.

For a concrete case, a growth lead used unique utm_campaign values per directory and then could not build a simple report. They collapsed to one campaign and source-per-domain, and the channel group became readable. The move that usually works is boring UTMs. If you skip boring, you will screenshot GA4 and still not know if directories matter.

After listings go live, spot-check that the outbound link still has the UTM. Editors rewrite. If they stripped it, log "stripped" and do not fight unless the site offers an edit. Fighting an editor over a query string is a bad use of a founder.

Featured-on strip: when the logo earns a place on your site

A featured-on row on the marketing site is social proof, not a backlink farm display. Add a directory logo only when all of the following are true: the listing is live, it is about your product, the live URL is public, you would be comfortable with a customer clicking it, and the directory is not a known spam network.

Do not add logos from submitted_form rows. Do not add logos from pending_review. Do not add logos from sites that listed you with the wrong name. Do not add 40 logos at 12px each. Pick a short row (six to twelve) of recognizable or relevant names. Update it when QA says yes, not when the worker hits submit.

Link the logo to your listing URL, not to the directory homepage. The homepage does not prove you are listed. Use rel attributes that match your policy. If you do not want to pass extra equity to a nofollow directory, you still can show the logo for humans. Humans are the point of the strip.

Keep logo files for the strip separate from your product mark. You are displaying their brand. Follow their brand guidelines if they publish any. Do not recolor a directory logo to match your pastel theme.

For a concrete case, an AI infra company added twenty directory marks the morning after a pack delivery. Half the URLs were still thank-you pages. A prospect clicked two, hit 404s, and asked in a demo whether the launch was fake. They cut the strip to five live, QA'd listings and the objection disappeared. The move that usually works is a small honest strip. If you skip QA, the strip becomes a liability.

Remove logos when a listing dies, when a directory turns into a parked domain, or when you are no longer willing to stand behind the site. Quarterly hygiene includes the strip. It is not set-and-forget decoration.

If you buy a SubmitLoop pack, wait for live URLs in the report, not for "form accepted" counts, before you redesign the homepage. Light often finishes forms in about a week after a complete profile. Live pages lag. Plan the design change for when QA passes.

When to stop DIY and buy Light at $49 for 60

DIY is a teacher. It shows you which assets are missing, which copy length you lack, and whether your site is even crawlable. DIY is a bad factory. The guaranteed easy-form pool is repetitive on purpose. That repetition is what a pack is for.

Stop DIY and buy SubmitLoop Light ($49, 60 guaranteed directories, one site) when most of the following are true:

• The kit exists (logo, screenshots, three copy lengths, categories, mailbox)
• You have already done Tier 1 yourself or decided Tier 1 is a short list
• The remaining work is mostly guaranteed / easy forms
• Your fully loaded hourly rate times expected hours exceeds $49 by a lot
• You want a run log plus screenshot or trace, not a memory of tabs
• You can live without a numeric DR guarantee (we refuse those)

Time math: if a careful directory takes 12 minutes including captcha and email verify, 60 directories are 12 hours. At $150/hour that is $1,800 of founder time. Light at $49 is not a close contest. At $30/hour in a market where your time is actually cheap, DIY can still be rational for learning, not for the sixtieth identical guest form.

Buy Light rather than Growth when you want the cheap close on the easy pool and you will keep high-touch or paid rows in-house. Buy Light when this is the first product and you are not sure directories will matter; $49 is an experiment. Buy Light when an agency needs proof for a small client without committing to 177 forms.

Do not buy Light if the kit is still a pile of TODOs. The pack will pause on your missing PNG. Do not buy Light if you expected 60 homepage dofollow links from DR 90 newspapers. That is not the pool. Do not buy Light if you wanted a tailored niche list of 12 editorial sites only. That is a different service shape (some competitors sell tailoring; we sell a guaranteed easy-form pool plus a public catalog you can DIY for the rest).

If you are mid-sprint and exhausted, it is allowed to stop at 15 DIY rows, clean the sheet, and hand the leftover guaranteed queue to Light. Partial DIY plus a pack is a normal hybrid. Double-paying two DFY vendors for the same week is the thing to avoid, unless you have compared lists and accepted overlap as already_listed.

When to buy Growth at $99 for 177+

Growth is $99 for 177+ directories on one site, with proof, and a refund if we claim successes we cannot evidence with a log plus screenshot or trace. It is the right SKU when the easy pool is the job and you want it mostly finished, not sampled.

Buy Growth when:

• Light's 60 would still leave you with a long guaranteed queue you will not personally finish
• You need coverage for a client report or a board slide that is about volume-with-proof, not about twelve logos
• $99 is cheaper than another week of contractor hours at even modest rates
• You understand editorial rejects and pending_review are not refunds
• You understand we still do not guarantee Ahrefs DR

177 forms are 177 forms. Delivery takes longer than Light. Incomplete logos, placeholder copy, or basic auth on the marketing site delay you, not us. Live public URLs lag. Featured-on still waits on QA.

Growth is the wrong SKU if you only wanted the ten highest-DR SaaS lists in your category. Use the catalog, filter, and DIY or pick those as Tier 1. Growth is the wrong SKU if you wanted paid partner launches bundled into the pack. Paid rows stay public; you buy those yourself. Growth is the wrong SKU if you cannot QA a large report. A 177-row evidence pack you never open is theater, the same theater as a 100-row spreadsheet from a random freelancer.

Cost per directory in the queue is about $0.56 before you count already_listed and rejects. That number wins only if the 177 are real forms in the easy pool. That is why we tag guaranteed. Ask (or read the catalog) what carries the tag before you pay, the same way you would ask any vendor for a target list.

Agencies: Growth's refund path is about missing proof, not about a client being sad that Google did not move. Set that expectation in the SOW. If the client defines success as live URL only, your QA process (below) is still required after we deliver form-level evidence. Form accepted is not the same as indexed.

For a concrete case, a seed-stage AI startup's contractor quoted 20 hours to grind leftover directories after launch week. At $75/hour that was $1,500. They bought Growth at $99, kept Tier 1 and two paid placements in-house, and spent the contractor time on QA and the featured-on strip instead of on captchas. The lesson is to pay for repetition and keep judgment in-house. If you skip that split, you either burn money on founder captchas or you outsource judgment to a pack that cannot know your category as well as you do.

QA of live URLs: the part everyone skips

QA is a calendar block, not a vibe. Book it at day 14 and day 45 after a sprint, plus a pass whenever a pack report lands.

For each row in live or already_listed:

• Open the live URL in a logged-out window
• Confirm the product name
• Confirm the description is yours (or an acceptable editor rewrite)
• Confirm the outbound link goes to you (canonical domain)
• Note if UTMs survived
• Note link type if you can determine it (view source or a link checker you trust); if unknown, leave unknown
• Screenshot the listing into the evidence folder
• Mark featured-on eligible only if you would show it to a customer
• If 404, mark it and decide whether to resubmit or drop
• If it is a competitor, treat it as an incident: claim flow if any, or a note to legal/brand, not a silent sheet change

Do not QA from memory. Do not QA only the first five rows. Sampling is allowed on a 177-row pack if you time-box, but sample across DR bands and across dates, not just the top of the sheet.

Check that the listing is indexable: no obvious noindex on the page if you were promised a public listing, not blocked by robots on that URL (you cannot control the whole site). If the site requires login to see your listing, it is not a public listing. Demote it from featured-on and from any "backlink" story you were telling yourself.

Outbound redirects: if they wrap your URL, log it. Wrappers are common. They can still refer traffic. They may change how you think about "dofollow from the directory domain." Do not panic. Record.

For a concrete case, a B2B SaaS found eight live listings pointing at an old Webflow domain they had redirected. The redirects worked, but two directories had uniqueness rules that later flagged the new domain as a duplicate when they tried to update. They updated where edit existed and left the rest. The move that usually works is QA the outbound host, not just "it loads." If you skip the host check, you will split brand signals across domains you thought you had killed.

When a pack delivers evidence of form success but no live URL yet, do not mark live. Mark submitted_form or pending_review. Set the follow-up date. QA is for URLs a stranger can open.

Quarterly hygiene

Directories rot. Sites die, get parked, turn paid-only, change category taxonomies, or lose the listing in a redesign. Your marketing site changes too: new name, new canonical, new pricing, new screenshots that make old thumbnails look like a different product.

Once a quarter:

• Re-open the catalog and refresh DR / pricing / guaranteed tags on rows you still care about
• Re-QA live URLs (dead, wrong product, wrong outbound)
• Update descriptions on sites that have an edit path if pricing or positioning changed
• Replace screenshots if the UI shipped a visible redesign
• Rotate any passwords the vault still holds if a contractor left
• Confirm the dedicated mailbox still receives mail
• Trim the featured-on strip
• Archive skip and unreachable rows rather than staring at them
• Decide if leftover guaranteed volume is a DIY weekend or a Light / Growth buy
• Re-run the checklist generator if the product type changed (you added an API, you became mobile, you left waitlist)

Do not run a full new 100-site blast every quarter as a ritual. Hygiene is maintenance of what you have, plus a small queue of new high-relevance sites. Blast volume is for launch windows and for pack purchases you will actually QA.

If you rebranded, hygiene is a project, not a checklist item. Make a new kit. New logo, new copy, new screenshots. Then walk live rows for edit links. Where you cannot edit, decide whether a new submission is allowed or whether you live with the old name until the listing dies.

If you shut down a SKU but the company continues, take down or edit listings that describe the dead SKU. Strangers will still arrive. Support will still get mail.

Document the quarterly pass in the sheet with a date in last QA. If nothing changed, still date it. An undated sheet is how you convince yourself you "just checked" in January when it is August.

A one-week DIY sprint that respects a day job

Day 0 (two hours): generator, kit files, copy doc, vault, mailbox test, sheet columns. Do not submit yet. If the kit is not done, the week has not started.

Day 1 (ninety minutes): catalog filter, pick Tier 1 plus a few guaranteed rows, cap at fifteen total for the week if you are solo. Copy them into the sheet.

Day 2 (ninety minutes): submit the first five. Expect to discover a missing privacy page or a screenshot with localhost. Fix the kit immediately.

Day 3 (ninety minutes): submit the next five. Process verify mail from day 2.

Day 4 (ninety minutes): last five, plus any pending_verify. Stop at fifteen even if you feel heroic.

Day 5 (forty-five minutes): evidence folders, statuses, follow-up dates. No new domains.

Day 14: first live URL QA.

If day 2 reveals that every remaining row is a login wall with image captcha and you hate it, stop. Buy Light. The sprint did its job: it proved the kit and the first listings. The factory is optional.

Agencies can compress this into a billed "kit week" and a billed "submit week" with client approval between them. Do not combine them into an unscoped "we'll submit you everywhere" line item. Everywhere is how you eat unpaid captcha hours.

Agencies, contractors, and clients

Write the rules into the SOW:

• Kit supplied by client, or billed as a deliverable
• Sheet without passwords
• Vault sharing method
• Definition of done (form evidence vs live URL)
• Which tiers are in scope
• Whether paid directories are extra
• Whether a SubmitLoop pack will be used, and who owns the SubmitLoop account
• Featured-on changes need client approval
• Quarterly hygiene is a separate retainer task, not a surprise

Do not dump a 200-row sheet on a client Slack and call it a report. Send live URLs, a sample of screenshots, and a count by status. Offer the sheet as an appendix.

If you use a contractor for DIY, they get the kit, the sheet, and vault items for the directories they will create, not your entire company vault. They do not get to invent copy. They do not get to pay for paid_only rows on your card without a written yes.

If the client insists on seeing passwords, refuse and explain. Offer a screenshare of a successful login instead. Clients who insist anyway are telling you they will leak the sheet. Price that risk or walk.

Common failure modes (and the fix)

Missing square logo: export one before any form. Generator first.

One description length: write three. Character-count them.

Homepage URL behind auth: remove the gate or stop the campaign.

Personal email as public contact: dedicated mailbox.

Passwords in the sheet: vault, then delete the column from version history if you can.

All tabs, no tiers: cap the week, Tier 1 first.

Paid upsell mid-form: log paid_only, close the tab.

Duplicate submissions because pending felt like failure: wait for the follow-up date.

Featured-on before live: wait for QA.

Buying two DFY packs the same week: dedupe lists on domain first.

Expecting DR to move tomorrow: read recrawl reality, refuse numeric guarantees.

Headless bot blasting: stop. Use a headed browser. Be a person.

Site-specific selector folklore from a 2019 blog: ignore it. Fill by labels and intent.

No evidence folder: screenshot as you go, or you will not remember which thank-you page was which.

Contractor used their own identity: listings become unrecoverable. Always client or product identity.

Wrong category on review sites: map categories in the kit, not at 1am.

UTM uniqueness theater: one campaign, source per domain.

Ignoring language and country: filter first.

Treating already_listed as a waste: QA it; it may be the win.

Never scheduling quarterly hygiene: listings rot, strip lies, mailbox dies.

What SubmitLoop will and will not do for you

The public catalog lets you filter DR, DA, traffic, dofollow versus nofollow, pricing, language, country, and guaranteed. Paid rows stay visible. Surplus free rows may sit behind email. Browsing is free.

The submission service is manual form work with software for queues and evidence. Light is $49 / 60 guaranteed. Growth is $99 / 177+ with a refund if claimed successes lack proof. A person fills forms. We do not promise editors will publish before lunch. We do not promise Google will rank you. We do not promise Ahrefs will add four DR. We do not bundle random paid placements into the pack so the count looks bigger.

You still own the kit, the vault, the mailbox, the featured-on strip, and QA of live URLs. If you want us to submit, send a complete profile. If you want to DIY, this checklist is the path. If you want both, DIY Tier 1 and paid rows, then buy Light or Growth for the repetitive guaranteed pool.

Competitors exist: LaunchDirectories, BacklinkBot, SubmitSaaS, TinyLaunch, SaaSCity, ScrollLaunch, and others. Compare entry price, what the count means, proof, refunds, and whether the catalog is usable without a card. Discount any page (including ours) that ranks itself first without showing checkout numbers. The submission service page is where our SKUs live. This checklist is how you stay honest if you never pay us.

A second pass on assets: file hygiene that prevents upload loops

Name files like a grown-up: `acme-logo-1024.png`, `acme-shot-dashboard.jpg`, `acme-shot-empty.jpg`, `acme-shot-aha.jpg`, `acme-demo-15s.mp4`. Do not leave `Screen Shot 2026-08-02 at 4.22.01 PM.png` in the kit. Some upload widgets choke on spaces and commas. Some evidence zips become unreadable to the next contractor.

Keep an `exports/` or `kit/final/` folder that only contains the files you are willing to upload. Keep `kit/raw/` for PSD, Figma PNG dumps, and 4K captures. Submitters should never browse raw. They will pick the wrong one.

Check EXIF if you care about leaking a username or GPS from a phone screenshot. Most product screenshots from a desktop do not have GPS. Phone photos of a laptop screen do. Do not use phone photos of a laptop screen.

If legal requires a trademark footnote on the logo, that footnote does not belong in the 64px thumbnail. Keep legal on the website footer.

When a form asks for both a logo and a "brand image," do not upload the same file twice unless you have to. Use the square mark and a wider OG image. Repeated identical images look like a bot kit even when you are a person.

If your logo is purely type, test it on a dark card and a light card. Directories will not ask which background you prefer.

A second pass on copy: claims you should not ship

Do not claim user counts you cannot support. Do not claim "used by NASA" because someone at a contractor once opened a trial. Do not claim SOC 2 Type II if you are in Type I or in a spreadsheet of controls. Do not claim "no training on your data" if your subprocessors' policies do not match that sentence. Directory copy is public and durable. It will be screenshotted in a sales cycle you do not control.

Do not use superlatives that invite a challenge: "#1 AI tool," "best CRM," "fastest API in the world." Moderators on better indexes treat them as spam. Write specific: "indexes tickets in under X on Y plan" only if true.

Do not paste investor names unless those investors are public on your site. A directory is not the place to leak a round.

Do not contradict the pricing page. If the site says $29/month and the directory essay says $19, you will support-ticket yourself.

Keep a changelog at the top of the copy doc: date, what changed, who approved. When a contractor asks which paragraph is current, the changelog is the answer.

A second pass on categories: marketplaces, agencies, and open source

If you are a marketplace with two-sided language, pick the side the directory's audience is. A "tools for founders" list wants the product they can sign up for today, not a manifesto about both sides.

If you are an agency, many software directories will reject you. That is correct. Do not force an agency into a SaaS taxonomy. Use agency lists, or skip. Packs aimed at software products will not magically make you a SaaS.

If you are open source with a hosted paid tier, say both. File as SaaS on SaaS lists if the hosted product is real. File as open source on OSS directories with the repo URL in the field that asks for it. Do not submit only the GitHub URL to a SaaS directory that wanted a product homepage.

If you are a browser extension, have screenshots of the extension in a browser, not just a marketing site. If you are a Chrome extension, some general startup directories still accept you; some want the store URL as the primary URL. Put the store URL in notes and follow the form label.

Tracking sheet: sharing, versions, and evidence paths

Share the sheet with "commenter" by default for clients, "editor" for people who submit. Turn on version history. Prohibit Excel copies on laptops "for offline." Offline copies grow password columns.

Evidence paths should be folder links (Drive, Dropbox, or a repo that is private), named by domain and date: `2026-08-17-example-com/thankyou.png`. The sheet stores the path, not the PNG binary. Do not paste images into cells. The sheet will become a 80MB monster and then crash on mobile.

If you use Notion instead of a sheet, you still need the same fields and the same password ban. Notion is not magically safer. A Notion page titled "logins" is still a leak.

If you use the SubmitLoop report plus your sheet, the sheet is the system of record for live URL QA and featured-on. The pack report is evidence of form work. Do not try to make one artifact do both jobs. You will either hide QA or hide proof.

Captcha, bot walls, and your browser profile

Use a real desktop browser you also use for banking-level sites, with a normal window size. Overly small viewports trigger "suspicious" paths on some CDNs. Do not replay the same form twenty times in a minute after an error. Read the error. Fix the field.

Disable aggressive content blockers on the directory tab if the challenge never renders. Re-enable after. Do not live with blockers off on the whole machine.

If you are tempted to rotate residential proxies to "look local," stop. You are not running a gray-hat SEO farm. You are listing a product. Local meaning is for the language filter, not for a proxy costume.

Kids' names, pet names, and birthdays as captcha-era passwords on directory accounts are how those accounts get stuffed later. Vault generates the password. You never see it if you can help it.

Email verify: IMAP, filters, and phishing

Create a filter: if the mailbox is dedicated, everything is probably a directory. Still watch for phishing that pretends to be a directory ("click here to verify your Bitcoin listing"). If you did not submit to that domain this week, do not click.

IMAP tools and assistants should only poll the dedicated mailbox. Do not grant a tool your primary mail. If you buy a pack, the verification mailbox on the business profile is part of the work. Point it at the dedicated address.

When a verify link asks you to also "upgrade to premium to finish verification," that is a paywall wearing a verify costume. Log paid_only. Do not complete it in a hurry because the button is green.

Bounce handling: if your dedicated address bounces, fix SPF/DKIM on your domain or use a mailbox that can receive. Directories will not debug your DNS.

UTM and analytics: what to look at after 30 days

Look at sessions, not vanity. Directories often send one-hit sessions. That can still be a journalist or a customer. Look at signups attributed to utm_medium=directory. If you use a product analytics tool, keep UTM parameters through the signup redirect. Many SPAs drop query strings; that is a product bug to fix before you declare directories useless.

Do not compare directory traffic to paid search in week one. Different intent. Judge directories as referring domains, brand URLs on the public web, and occasional referral signups. If after a quarter you have zero referred signups and zero indexed listings, the campaign failed or QA never happened. If you have indexed listings and no signups, you may still have the SEO layer you wanted. Say which goal you had.

UTM on app store links: if the primary URL is an app store, store analytics are a different system. Do not force web UTMs onto a store URL that will 404 or strip them. Follow the form.

Featured-on strip: design notes that keep it honest

Use grayscale marks if your designer needs them to sit still. Do not animate the row. Do not claim "as seen in" for a directory that is not press. "As seen in" is for publications. "Listed on" is for directories. The words matter to lawyers and to people who have seen too many Web3 landing pages.

Cap the row. Add a "/press" or "/listings" page if you really want to show more, and still QA every link. Footer museums of 60 logos help no one.

If a directory's mark is ugly or dated, you can omit it even if the listing is live. The strip is marketing. You are allowed to curate. You are not allowed to invent.

Pack cutoff: a simple numeric rule

Compute `hours_left * hourly_rate`. If that number is greater than $49 and the leftover queue is about 60 easy forms, Light wins. If it is greater than $99 and the leftover queue is the full guaranteed grind, Growth wins. If the leftover queue is eight editorial sites with unique essays, neither pack is the tool; you are the tool.

If hours_left is unknown, time the next five DIY rows and multiply. Founders systematically underestimate form time because they forget captcha, verify, and the one site that needed a phone.

If you will not QA, do not buy Growth. Buy nothing, or buy Light and QA sixty, or DIY fifteen you will actually check. Un-QAd volume is how the industry got a bad name.

QA: accessibility, language, and spam siblings

Open the live listing on mobile once per sprint sample. Some directories have mobile templates that truncate your name to nothing.

If the directory translated your English essay into another language automatically, read it or have a native speaker read a sample. Machine translation of "transformer" in an AI product can become a discussion of electrical equipment. Edit if you can. Note it if you cannot.

Spam siblings: some networks clone listings across many hostnames. If you submitted once and find yourself on five near-identical sites, log the family. Do not treat them as five referring domains in a client report without disclosure. Ahrefs may cluster them anyway.

Quarterly hygiene: legal, security, and people

When a person who owned the vault leaves, run hygiene immediately, not at quarter end. When a domain is sold, assume listings still point at you until QA says otherwise. When you sign an enterprise contract that restricts where the product may be advertised, compare that list to the sheet. Directories you forgot can violate a quiet restriction.

Security: directory accounts are low-value until they are not (they can change your URL to a competitor, they can post as you). Unique passwords, no sheet storage, mailbox you control. That is the whole program.

People: assign an owner for the sheet. Orphan sheets are how quarterly hygiene never happens. Put the next QA date in the company calendar, not only in a cell.

Closing: kit, queue, proof, then stop

Gather the kit. Filter the catalog. Track without passwords. Submit in tiers. Verify mail. QA live URLs. Curate the featured-on strip. Repeat quarterly. When the remaining work is repetitive guaranteed forms and the math says your hours are expensive, buy Light at $49 for 60 or Growth at $99 for 177+. Use the submission checklist generator whenever a new product or a rebrand makes the old folder untrustworthy.

The win is not a large number in a tweet. The win is a small set of true public listings, evidence for the ones a vendor claimed, and a homepage strip that does not 404 when a prospect clicks it.

11571