Privacy at CRM Better

Effective October 15, 2026 · Contact conversion, support, CRM requests, new export formats and AI review

What happens when you choose files

Your browser downloads the converter software from our website. It then reads and converts your selected files on your device, using programmed rules and your selections. Previewing, updating mappings, merging duplicates and generating downloads happen in a browser worker, with temporary working files held in browser memory.

Contact files, field values and converted output are not sent to our conversion server. There is no automatic server fallback. If your browser cannot complete a conversion, you can cancel, try a smaller batch or use another supported browser.

CRM Better does not edit original files on your device, connect to your source CRM, or change contacts in HOME. The output can differ from the source, for example when matching duplicates are merged or phone numbers are normalized. You decide whether to download and import the result.

No AI contact analysis

The conversion itself uses no AI: nothing reads, classifies, maps, merges or converts your contacts with AI, and the conversion does not send your contact data to an AI service or use it to train AI models. The exception is a file you choose to attach to a support report or CRM request: we use it to reproduce the problem, and that can involve AI coding tools, as described under AI review of what you send. Buyer-search names and prompts are assembled from available criteria using programmed rules. If you choose to paste a prompt into Home’s AI assistant, that is a separate action in Home, subject to its own policies. Reports you choose to send us are reviewed by AI; see AI review of what you send.

Local working data and website logs

The worker removes its temporary working files after each conversion. Selected files, previews and choices remain in the current browser page while you work. Clear this session or close the tab to release the session. We do not save contact data to browser local storage or an IndexedDB database. If you answer “No thanks” to the new-format card, the browser stores only a fingerprint of that file’s column layout, so it does not ask again. When you download a converted file or export a mapping, the browser also keeps the mapping choices you used for that CRM (column names, and the stages, tags and other values you chose to translate, and tags you added to a stage; never contact details or tags you add to every contact) so it can offer them on your next file from that CRM. Clearing this site’s data in your browser removes them. Downloaded files remain on your device until you delete them. Your browser may cache the converter software; that cache does not contain your contact files.

Our server provides the website and converter software. Hosting and network providers can process technical connection information for those requests. Public conversion-upload endpoints are disabled. Previous versions processed temporary copies on the server; this version processes conversions locally.

Optional usage analytics

When configured, PostHog and Google Analytics help us understand visits, conversion steps, completed previews, mapping changes, downloads, and general error categories. The usage-analytics notice is displayed when tracking is configured. You can turn analytics off using that notice; we also honor Global Privacy Control and Do Not Track. Conversion works with analytics disabled or blocked.

We send counts, timings, source CRM categories, recognized standard column labels and mapping destinations, and random browser and run identifiers. Unknown or custom column headings are grouped as “other.” Confirmed mapping events include unchanged suggestions as well as edits, so we can compare which defaults users keep or change. The first completed import download counts once per run; repeat downloads do not add mapping votes. A download does not prove an import into HOME succeeded, and these observations do not automatically change mappings.

Analytics do not include uploaded files, contact names or values, file names, custom headings, notes, tags you enter, buyer criteria, support descriptions, raw error messages, or contact-file fingerprints. We do not record your screen or replay sessions. Page addresses are sent without query strings or fragments. Google receives only the referring site’s origin, not its full page address.

Analytics providers receive technical connection information, including your IP address. PostHog receives a random browser identifier stored locally and is asked not to enrich location from the IP. Google Analytics may use its own cookies and technical device information. Advertising personalization and Google Signals are disabled. Events are retained according to the configured project settings. See PostHog’s privacy policy and Google’s privacy policy.

Support reports you choose to send

The Support button sends a report to Kyle at kylenorthup@gmail.com. It includes your description, current step, displayed errors, file names and types, column headings and mapping destinations, counts, browser information, and local conversion status and recent request status. You can inspect the diagnostics before sending. Descriptions, file names, headings, or error messages can contain personal information; review them before submission.

Every report asks for your email so we can tell you when the problem is fixed. It is used to reply about your report, not to subscribe you to marketing. After you send a report, this browser remembers the address (in local storage, on your device only) and fills it in on your next report; clear your browser’s site data to remove it. Original uploaded files and contact preview rows are attached only if you explicitly select that option. Attachments are limited to 10 MB total. If you attach them, we use them to reproduce the problem, as described under AI review of what you send.

Support emails and approved attachments are delivered through Resend and retained in Kyle’s Gmail for troubleshooting. They are not covered by conversion-file deletion. A separate private issue log stores limited technical summaries: report ID and dates, CRM and file types, workflow step, recognized standard field labels, error categories and HTTP statuses, counts, browser family, screen-size category, and issue/delivery status. It excludes your description, reply address, file names, raw custom column headings, raw error messages, contact rows, attachments, and IP address. Similar summaries are grouped to help find recurring problems.

Support mail and issue-log records remain until manually deleted. To request access to or deletion of a support report, email Kyle and include its reference number if available. We will address records under our control; provider backups and records may follow their own retention policies.

CRM requests

The “Don’t see your CRM? Request it.” form sends the CRM name, your email, and either the header row you paste or the contacts export you attach. It also sends browser information (your browser’s user agent, window size and language) and, for an attached file, its name, size and type. It goes to Kyle at kylenorthup@gmail.com through the same path as support reports, with the same delivery, retention and deletion described above. Your email is used to reply about that CRM, not to subscribe you to marketing.

A header row is only column names and should not contain contact details. An attached export can contain client data. It is attached to the email only, limited to 10 MB, and not stored on our server. If you attach one, we use it to check that your CRM converts correctly. The private issue log keeps the same limited technical summary as a support report, never the CRM name, header row, file name or file.

New export formats

If you upload an export from software we have not seen before, the preview may show a card asking whether you want us to add it. Nothing is sent until you open the card’s form, review the report and press Send. Clicking No thanks, or ignoring the card, sends nothing.

Only if you choose to send a report, we email Kyle at kylenorthup@gmail.com: the software name you enter, your email, the file’s column names, how many rows fill each column, the file type, row count and how the converter read it, and, unless you untick them, 10 sample rows scrubbed by an automatic rule shown to you first. The form shows the exact text before you send it, along with browser information (your browser’s user agent, window size and language).

Recognized name columns and ambiguous unlabeled name-like columns become placeholders. Elsewhere, only known CRM categories, dates and nonletter values can remain after at least 3 repeated rows. Other names and custom labels are masked. Emails, phone numbers, web addresses and street addresses become look-alike placeholders; a link's domain or an email's +tag can remain when three or more rows share it. Long notes become text-length placeholders. Column headings are retained. This does not guarantee that all identifying information is removed: a name matching a retained category can remain outside a recognized name column, and headings, domains, tags and your description can contain identifying text. Read the exact report before sending it.

The email is delivered and retained the same way as support reports. The private issue log stores nothing from it except a fingerprint of the column layout (a code computed from the column names, not the names themselves) and counts, so we can tell how many uploads brought the same format.

AI review of what you send

Every support report, CRM request and new-format report you send is reviewed by our AI, Claude, provided by Anthropic, to build a fix. Each form says so before you send it. There is no setting to skip the AI review. If you do not want a report reviewed this way, do not send it.

The AI receives your description, the column headings and where they were mapped, fill counts, the step you were on, error messages, browser details, and the sizes and types of your files. For a CRM request that includes the CRM name and any header row you paste; for a new export format, the text the card shows you before you send, including the scrubbed sample rows unless you untick them. A file name reaches it only where your description, an error message or a CRM request mentions it, such as the name of an export you attach. It does not receive contact preview rows or the files you attach.

What is scrubbed first: before the AI reads a report, every email address and every phone number written in a US format is replaced, wherever it appears in what you send: your description, column headings, error messages and the rest of the troubleshooting details. Your reply email is removed from the report, including from your description, before the AI reads it; any other email address becomes “[email]”, and every digit of a phone number becomes 0, so the AI sees how the number was written but not the number. Sample rows from a new export format are also scrubbed by the rule described under New export formats, which replaces names and other values in them unless three or more rows share the value. Names you type yourself, in a description or anywhere else, are not detected or removed, so keep client names out of what you type.

If you attach your files, we use them to reproduce the problem. We run them through the converter on our own computers to find what went wrong. That work can be done with AI coding tools (also Claude) that run the converter and read the results, such as how many rows convert and which values a column holds. Attached files are not sent to the automatic AI review.

The AI prepares a fix and tests it in an isolated copy of the converter. It does not change your current import, and nothing it prepares goes live until a person on our team has checked it. Reports, source snapshots and AI results are kept in a private support queue until manually deleted. Anthropic processes what it receives under its privacy policy. To request deletion, contact Kyle with the report reference.

Service providers and maps

Cloudflare routes and protects the site and hosts the support endpoint and private issue database. It can process technical information such as IP addresses and request metadata under its privacy policy. Support uses Resend and Gmail to deliver and retain the information you choose to send. A short-lived hashed IP key is used to limit repeated support submissions; it is not added to the private issue log.

When a buyer-search map is displayed, your browser requests map tiles from OpenStreetMap. Those requests reveal your IP address and the map area being viewed. CRM Better does not send contact names or the full exported search boundary to the tile service. See the OpenStreetMap Foundation privacy policy.

The converter does not use session replay or create advertising audiences from your contacts. We use uploaded contact data to provide the conversion you request, not to sell contact lists or create advertising audiences.

Other forms and your choices

The older converter’s optional signup and CRM Better’s separate booking forms are distinct from conversion and support. If you use them, the details you submit may be sent to our business CRM, Cloze, and used to respond to that signup or booking. You do not need to use either form to convert contacts or request support.

Only upload files you are authorized to process. Send the minimum information needed in a support request. For questions about this policy or your information, contact kylenorthup@gmail.com. We will update this page when these practices change.