Skip to main content

Have something to say?

Tell us how we could make the product more useful to you.

Your campaign stats are hiding your results

When a client opens their campaign report, the numbers they see make the campaign look weaker than it is. Last week, a client's campaign cards showed a 0% SMS open rate and a 0.0 rating, while four new 5-star Google reviews had arrived, each within two minutes of a contact being redirected from the campaign. Clients judge the product, and our agency, on that screen. What happens today Reviews left on Google are never counted: once the contact leaves for the review site, the platform cannot see the review, so the campaign rating stays at 0.0. I matched them by hand: contact redirected at 19:18:04, Google review published at 19:18:53, same store. Four out of four matched within two minutes. SMS open rate shows 0%. SMS opens cannot be tracked at all, but "0%" reads as "nobody opened it". Email opens are undercounted. Three emails were clicked and still counted as not opened (image blocking, privacy proxies). Clicks are reliable; opens are not. Proposal Say "not tracked" instead of 0% for SMS opens, and flag email opens as an estimate (or count a click as an open, so clicked > opened never happens). Likely reviews: after a redirect, if a new review appears on that same source and location within a short window (for example 10 minutes), count it as a "likely review from this campaign" and link the two. This uses time only, not names, so usernames do not break it. Show it as a separate line, so it never claims more than the data supports. Campaign period view: reviews received on each location while the campaign runs, next to the same period before. It is the number a business owner actually cares about.

Claude can't sign in to AI Hub: discovery returns 404

I tried the new sign-in connection the day after the announcement, and again the next morning. On my white-label domain, Claude never reaches the sign-in page: it stops with an error saying it could not register with the sign-in service, so the connector stays disconnected. The token connection on the same address still works fine. What happens today In Claude, Customize → Connectors → Add custom connector, with my MCP address https:// /api/mcp. Claude detects that sign-in is required. Click Connect. Claude answers that it could not register with the sign-in service, and no sign-in page opens. Same result after deleting the connector and adding it again. Looking at the server, the three OAuth discovery documents return the right JSON, but with HTTP status 404 instead of 200: GET /.well-known/oauth-protected-resource/api/mcp -> 404 application/json (valid JSON body) GET /.well-known/oauth-protected-resource -> 404 application/json (valid JSON body) GET /.well-known/oauth-authorization-server -> 404 application/json (valid JSON body) POST /oauth/register (empty body) -> 400 invalid_redirect_uri (endpoint is alive) POST /register (empty body) -> 419 My guess: Claude reads the 404 as "no metadata", falls back to the default /register path, and fails there. I can't confirm it from my side. Proposal Serve the three.well-known documents with status 200 on white-label domains. If it already works on other domains, check what differs on mine (happy to share the domain privately).

The View as customer banner ruins demo videos

I record demo videos and screenshots for my clients from their own account, using View as customer, so they see exactly their dashboard. Every frame then carries a full-width yellow warning bar at the top, saying I'm logged in as the customer. It's useful for a second, then it pulls the eye in every recording, and my clients see a warning style on their own dashboard. What happens today (checked 19 November 2026) View as customer shows a 56 px yellow bar across the top of every page, with the customer's name, a sentence and a Leave button. It stays on screen while the page scrolls, and nothing lets me collapse or hide it. I found no setting for it in the agency settings. Proposal A collapse button that shrinks the bar to a small pill in a corner (customer name and Leave), remembered for the session. Or a presentation mode that hides it until I turn it off, with Leave still available in the user menu. Keep the full bar as the default, so nobody forgets they are viewing as a customer.

Let me see the app the way my client's staff do

At most of my clients, the owner rarely logs in. The people who use the platform every day are their staff: the office manager who sends review requests, the person who answers reviews. They are the ones I train in demos and the ones who contact me for support. When one of them tells me they can't find a button or don't see a location, I can't see what they see: View as customer only opens the owner's account, with full access. What happens today (checked 19 November 2026) The customer page offers Open dashboard (their dashboard, nothing locked) and View as customer (the owner's view, with their plan's limits). For a team member, the User Directory offers Manage and Delete, and the organization's members page offers role settings and removal. Nothing lets the agency see the platform as that member. So a Manager or Responder limited to some locations can't be checked from the agency side. My only options are asking them for screenshots, or recreating a test user with the same role and locations. Proposal Add View as user to each team member, in the User Directory and on the organization's members page, showing exactly their role, permissions and locations. Keep the same banner and Leave button as View as customer. Optionally, a read-only version of this view, so a support check can't change anything by mistake.

User Directory shows roles, never location access

One of my clients runs two brands, and one staff member manages reviews for both, so they belong to two organizations: Admin in one, Manager in the other. I wanted to check which locations they can reach, and no single screen tells me. The User Directory lists their organizations and roles, but not their location access, and its Manage button only opens the settings of their first organization. What happens today (checked 19 November 2026) Agency > Customers > User Directory shows one row per user, with each organization and the role in it. Location access is not shown. For a team member, Manage opens the settings page of their first organization. For a customer (owner), it opens the customer page, which doesn't show members' location access either. Location access (all locations or specific ones) is set inside each organization, in the member's role settings or the invite form. For a user in several organizations, that means one page per organization to open, and nowhere to see the result side by side. Proposal In the User Directory, show location access next to each role: "All locations", or the number of locations with the list on hover. Make Manage open a page for that user: their organizations, with role and location access in each, editable from there. Let the directory filter by organization or location, to answer "who can see this location?". Related: Extend "Search" to include Members is about finding members across organizations; this one is about seeing and changing what each of them can access.

One brand, two organizations: only one can use its domain

When a client has two Google Business Profiles, I set them up as two organizations. Only the first one can send review requests from the brand's own email domain. The second gets "Could not create a new sender: Domain with same name already exists" and keeps sending from the platform's default address. Its customers receive a request signed with the brand's name but sent from an unrelated address: more spam folder, and it looks like phishing. What happens today Organization A (client X): Settings › Email settings › Sender domain, enter brand.com, Save. The domain is created and the DNS records are shown. Organization B (same client, same brand, its own location and Google profile): same page, enter brand.com, Save. Error: "Could not create a new sender: Domain with same name already exists". There is no way to link the existing domain, so organization B stays on the default sender. The error looks like it comes from the email provider (Brevo here), which already holds brand.com, authenticated for organization A in the same agency account. The same limit shows up inside one organization: two brands sharing an organization can't each have their own domain, since an organization takes one sender domain. Proposal If the domain already exists in the agency's provider account, let another organization of the same client use it (link instead of create), with its verification status. Better: manage sender domains at the client level, and let each organization pick its domain. At least, replace the raw error with a clear message ("This domain is already used by organization A") and what to do next. Related request from 2024, same need across locations: https://roadmap.embedmyreviews.com/en/p/e-mail-settings

Locked out by 2FA, a client user can only be deleted

A team member at one of my clients got stuck on the 2FA code screen at login. They had no authenticator code and did not remember turning 2FA on. For a white-label agency, that login screen is our brand, and I had no way to help: I could neither see that 2FA was on for them nor turn it off. The only way out was to delete the user and invite them again. What happens today (checked 19 November 2026) Two-factor authentication is a setting in each user's own profile, enabled there with an authenticator app. In the agency's user directory, the only actions are "Manage" and "Delete". "Manage" opens the organization (for a team member) or the customer page. The customer page has a password reset button, and nothing that mentions 2FA. The organization's members page lists each member's name and role, with no 2FA status; for the locked-out user, it offered no 2FA action either. The Security page mentions optional 2FA enforcement for all team members, but I couldn't find that setting in the agency settings. Proposal Show each user's 2FA status in the user directory and on the organization's members page. Add a "Reset two-factor authentication" action there, for the agency only, so the user signs in with their password and can set 2FA up again. If agency-wide 2FA enforcement exists, make it visible in the agency settings and the documentation.

My AI assistant can analyze every client account, but I still set each one up by hand

The AI Hub is great for reading: with Claude I can review a client's reviews, campaigns, contacts and metrics in minutes, and draft replies or Google Posts that wait for approval. But as soon as a client needs setting up or changing (a new location, a campaign, its texts, a review source, the reply prompt), the assistant has no tool for it. So I end up driving the dashboard with an AI that clicks through the browser, which is slow and fragile, or doing it myself. In my done-for-you work, this is where most of the time goes: onboarding a client and keeping its setup right. The guardrails are already there (scoped tokens, write permission, approval queue, audit log), which is exactly what makes it safe to go further. What happens today MCP: 57 tools to read data, draft review replies, tag reviews, run Local Search Grid scans, prepare Google Posts and manage SMS providers. REST API: can also create and update customers, organizations and locations, issue Google or Facebook connect links and send review invites. None of these are available as MCP tools. Neither the MCP nor the API can create or edit a campaign (steps, email and SMS texts, delays, sending window, logo), pause or resume it, import a contact list (the API sends invites one at a time), add a review source other than Google or Facebook, change the review form and page, set the AI reply prompt or auto-respond rules, or invite a user with a role. Blueprints help with a first setup, but the assistant can't apply one or adjust what comes after. Driving the dashboard instead: the active organization is shared across tabs and sessions, so a change in one place switches the client everywhere else (reported here: https://roadmap.embedmyreviews.com/en/p/switching-client-in-one-tab-silently-switches-every-other-tab). During one of those switches, a save I had made was lost without any message. Proposal (simplest first) Bring to the MCP what the REST API already does: create and update organizations and locations, issue connect links, send invites. Campaign setup tools: apply a blueprint package to an organization, create a campaign from a blueprint, edit its steps (texts, sender, logo, delays), schedule, pause and resume, and import contacts into a paused campaign. Same approval and audit rules as Google Posts. Account settings tools: review sources (custom and third-party URLs), review form and page texts, AI reply prompt per organization, auto-respond rules, user invitations with a role. Make every write tool explicit and safe for agents: an explicit organization_id on every write call, never tied to the dashboard's active organization; a preview (dry run) before saving; the saved state returned after saving; everything in the audit log. The dashboard side (one client per tab) is covered in the request linked above. What it would change: onboarding a client becomes a conversation with an assistant that prepares everything, with a person approving what goes live; fewer setup mistakes across many clients; and setup no longer depends on knowing every screen of the dashboard. Related: Use the organization name as the default location name, where another agency mentions managing clients through the MCP.

Standard SMS under 160 characters counted as 3 SMS

Hi Mannie, Quick one on SMS campaigns. My message is about 136 characters (template text + the auto-added “Cliquez ici : {{ short_link }}”), and the editor shows “3 SMS messages will be sent”. I think it’s because “ê” in “être” isn’t in the GSM-7 charset, so the whole message switches to Unicode (70 chars max, 67 per segment). Can you confirm that’s how segments are calculated? If so, small feature request: show a warning in the editor when a non-GSM character is detected (and which one), plus the real character count after the link is added. Salon owners won’t figure this out on their own and will burn 3x the credits. Thanks! JG

“Add to Phone Wallet” for QR codes

Please add an “Add to Wallet” option to the “Start Collecting Reviews with QR Codes” section, so any QR code can be saved as an Apple or Google Wallet Card on the business owner’s or staff member’s phone. How it works: After a job or visit, they open the card, show it to their customer, and the customer scans the QR code. That takes them straight into the feedback form to collect the review. Why it’s useful: It’s another simple way to start collecting reviews, especially for tradespeople visiting customers at home. It works just as well in hospitality, for example a waitress showing the card on her phone when she brings the bill. Something Like: https://addtowallet.co/ they create Apple and Google Wallet passes. Building something similar into the QR Codes section would keep everything in one place, with wallet scans tracked alongside the existing QR code stats.

AI review replies can't see the business facts we already confirmed for Google Posts

For Google Posts, EmbedMyReviews now asks us to confirm each business's facts (services, guarantees, availability, areas served, offers) and grounds every AI post in them. AI review replies don't use any of it. They only get a prompt, so a reply to a complaint or a question stays generic ("please contact us directly") exactly where a real fact would help most. I've tried the workaround many times: typing each client's facts into its reply prompt by hand. The prompts get very long, they have to be updated client by client, and the replies still rarely use the facts well. Some other review tools I've looked at ground replies in a structured business profile, and it works very well. What happens today Posts › Settings › Business facts: per location, filled from the Google profile, each fact confirmed by a person. Every number, price and date in an AI post is checked against them, approvers can report a wrong fact, and a scheduled post waits if a fact it uses changes. Post voice: agency playbook for every location, overridden per location, plus "Make this a standing rule" from client feedback. Review replies: a Tone & Style prompt (name + free text) with six variables: review, author, rating, organization, location, language. Nothing from the business facts, nothing from the post voice. I tested it on a demo location: I confirmed four facts (furniture insured up to €50,000, free survey within 48 hours, areas served, main service), then generated an AI reply to a 1-star review about a TV damaged during a move. The reply (translated from French): "We understand your frustration about your damaged TV and your claim. Feel free to contact us directly so we can look at your situation together." No guarantee, no contact channel, nothing from the facts. Multi-location: facts are entered location by location. For a client with 20 locations sharing the same services and guarantees, that's 20 entries for posts, plus the same information again in the reply prompt. When a guarantee or an area changes, every copy has to be found and updated. Screenshots in the first comment: the business facts and playbook used by posts, then the AI reply window and the prompt form for the same location. Proposal (simplest first) Let AI replies read the confirmed business facts of the review's location, with the same rule as posts: use only confirmed facts, never invent a number, a price or a date. Add the few facts replies need to the same screen: how to reach the business after a complaint (phone, email, named contact role), how claims and guarantees work, frequent customer questions with their answers, things never to say. One business profile, used by posts, replies, AI Review Assistant and AI Insights. Share the voice: one Voice and Instructions page with a posts part and a replies part, in the same three layers (agency playbook, client, location). Multi-location: let facts and voice be set at client (organization) level and inherited by every location, with location overrides (address, areas, local offer). Edit once, every location follows; show who changed what and when, so the profile can evolve without breaking anything. Close the loop: "Make this a standing rule" on an edited AI reply, as on posts. What it would change: replies that answer with the business's real guarantees, contacts and services instead of "contact us"; one place to keep a client's information right, used everywhere; posts and replies that never contradict each other; and setup that scales with a network instead of growing with every location. Related: Location-based AI Prompt Management (https://r

Email notification when a customer connects GBP

We send out emails to clients to connect their GBP, and there is no way to know if they connected or let the link expire after a few days without manually checking. Would be great to get pinged when they connect their GBP

Manual upload with “Send immediately” still schedules new contacts 3h later

Hi Mannie, I’d like to report a bug on Organization 57 (Salon Prestige), campaign 91 “SMS PRESTIGE”. My client regularly does manual bulk uploads of contacts and selects “Send immediately”. According to him, part of the batch always goes out right away, but another part is scheduled 3 hours later. The contact list shows “Envoi dans 3 heures” (sending in 3 hours) on these contacts, even though he picked “Send immediately”. What I see in the data for today’s upload (2026-11-14, around 17:17 UTC): Contacts that already existed in list 53 were processed immediately at 17:17 UTC. In the examples I checked (contacts 83765, 83482, 82281), they were marked “ignored” (not sent), which is expected because of our duplicate filter. All 47 contacts newly created by this upload (for example IDs 84401 to 84417) got send_at = subscribed_at + 3h exactly. Contact 84417 was added at 17:17:31 UTC and has step 1 scheduled for 20:30 UTC. The 3h delay matches the campaign’s normal delay for contacts added through the API (for example contact 83765, added 10:22 UTC on Nov 13 and invited at 13:33 UTC). So it looks like “Send immediately” on a manual upload does not override the campaign’s step delay for newly created contacts. It only seems to apply to contacts that already existed. Expected behavior: when “Send immediately” is selected on a manual upload, every contact in the batch (new or existing) should be sent right away, within the campaign’s send window, without the campaign delay. Could you take a look and let me know if this is a bug or a setting I’m missing? Thanks, Jean-Gabriel

Incorrect TXT Record Instructions for Amazon SES Domain Verification

The Amazon SES domain verification section incorrectly instructs users to add a TXT record for verification. Amazon SES DKIM verification requires only the three CNAME records provided by Amazon SES. A TXT record is not required. Expected Fix: Remove the incorrect TXT record instructions. Display only the three required DKIM CNAME records. Correct the verification guidance to match Amazon SES requirements. Ensure the same incorrect instructions are not displayed elsewhere in the platform. Impact: The incorrect instructions can confuse users, cause unnecessary DNS configuration changes, and prevent users from completing their email domain setup smoothly. Please correct this issue.

Add Copy Link / Share Button for Opt-in Pages

Currently, under Campaign Settings → Opt-in Page, there is only a Preview Live Form option. Users can open the form in a new browser tab and manually copy the URL from the browser address bar. We would like to request a dedicated Copy Link / Share Opt-in Page button that allows users to easily copy or share their Opt-in Page URL directly from the settings. Expected Functionality: Add a Copy Link button next to the Preview Live Form option. Allow users to copy the Opt-in Page URL with a single click. Optionally provide a Share button for easier distribution. Business Benefits: This improvement would make it easier for businesses to distribute their opt-in forms through WhatsApp, email, websites, social media, and QR codes, improving convenience and user experience.

Edited Google reviews don't update, so they never show as Needs Response

When a reviewer edits or updates their existing Google review, the change isn’t picked up. The review keeps its old text, old date and the business’s old reply, so it stays marked as Replied and never appears under Needs Response. The client only finds out by checking Google directly. Example from one of our clients (Allesley Park Dental Practice): a reviewer updated their Google review about 2 weeks ago, but in the dashboard it still shows their original review from May 2022 with the practice’s old reply. Same for another reviewer whose original review is from May 2021. A third updated review doesn’t appear in the dashboard at all. Expected: when a Google review is edited, the text and date update and it’s flagged as Needs Response if the new version hasn’t been replied to. Happy to share the reviewer names privately.

1Review Sources

Feature Improvement Request – Automatic Country Code Detection

Hi Team, We would like to request the following two improvements related to phone number fields: 1. Automatic Country Code Detection on the Signup Page Currently, the country code is not automatically detected when a user enters their phone number during signup. It would be helpful to implement automatic country detection based on the user's location or browser region, so the country code is preselected accordingly. 2. Automatic Country Code Detection in Feedback Forms We would also like the same functionality in feedback forms. When a customer opens a feedback form, the phone number field should automatically detect and preselect the appropriate country code instead of requiring users to select it manually. Expected Behaviour: Automatically detect the user's country. Preselect the corresponding international dialing code. Allow users to manually change the country code if needed. Apply this functionality consistently to both signup pages and feedback forms. These improvements would make the phone number entry process smoother and reduce friction for international users. Please let us know if these improvements can be implemented. Thank you.

One Trustpilot page for several stores: every review shows up 3 times and the stats are inflated

Most multi-location clients have one Google profile per store but a single Trustpilot page for the whole brand. To offer Trustpilot on each store's review form, I connect that Trustpilot page on every location, and EMR then imports each Trustpilot review once per location. For a 3-store client, about 63 Trustpilot reviews appear 190 times. The "All locations" dashboard shows 2,295 reviews instead of about 2,170, every 1-star Trustpilot review counts three times in the star breakdown, and the Reviews inbox shows each Trustpilot review as three separate cards, each with its own "Reply" button. The client's figures no longer match Trustpilot, and I have to explain why. What happens today Store A, B and C each have their own Google source; Trustpilot is one page for the brand. The same Trustpilot page is connected on A, B and C so it shows on each store's review form. Each sync imports the same review (same Trustpilot review ID) once per location. Every "All locations" view adds the copies up: total reviews, average rating, star breakdown, response rate, Reviews inbox, export. Sources are connected per location (Integrations › Sources), and I found no setting to share a source across locations or to merge identical reviews. My workaround today: keep the real Trustpilot source on one location only, and on every other location create a Custom Source pointing to the Trustpilot "write a review" link. The review forms keep their Trustpilot button and nothing is imported twice. But the brand's Trustpilot reviews then belong to a single store, and the setup has to be repeated by hand for each location of each client. Screenshots in the first comment (demo organization, same Trustpilot page connected on two locations): the same review appears once per location, and the "All locations" total jumps from 760 to 960. Proposal (simplest first) In "All locations" views, count a review once when the same source and the same review ID are attached to several locations (dashboard, statistics, Reviews inbox, export). In the Reviews inbox, show it as one card tagged with its locations ("Store A · Store B · Store C"), with a single reply. Let a source be connected once at the organization level (brand-wide Trustpilot, Facebook page or website), available to every location's review forms and widgets, and imported once. Let the agency choose, per location, whether brand-level reviews count in that location's own statistics.

No safe way to prepare a launch: a paused campaign can't be tested or loaded, so we go live to test

When I onboard a new client, I build their campaigns paused. Then I'm stuck: a paused campaign can't send a test, and its contacts can't be imported. The only way to see the real email and SMS on my phone, or to load the client's list, is to switch the campaign on. If that happens during its sending window, it starts sending to real customers right away. For a client's very first campaign, where nothing may go wrong, that is exactly backwards. What happens today (9 November 2026) On a paused campaign, "Send test" is greyed out, with no explanation. On an active campaign outside its sending window, "Send now" on the test is postponed to the next window. "Invite multiple contacts" (CSV import) only lists active campaigns, and its button is "Import and start sending": the daily batch starts today. The tester is enrolled in the whole sequence, so after testing I have to find and delete my own contact before launch. Proposal Allow "Send test" on paused campaigns, sent immediately whatever the sending window, and only the step being tested (or a "test the whole sequence now" option). Allow importing a CSV into a paused campaign: contacts wait, and sending starts with the daily limit when the campaign is resumed. A "start date" for the import would also do. (Related: Upload csv of contacts without adding to campaign, Campaign schedule on bulk import.) Remove test contacts automatically once their test is sent, or keep them out of the contact list.

The email preview shows the logo picked in the step, but customers receive a different one

Each email step of a campaign has its own logo setting, and the preview shows the logo you pick there. The email that actually goes out ignores it and uses the organization's logo from Settings. I only found out when a client's first test email arrived with the wrong logo, after I had checked the preview. For an agency running several brands, or a client with a wide and a square logo, the preview gives false confidence: what you approve is not what customers get. What happens today (reproduced on a demo organization, 9 November 2026) Campaign editor → email step 1 → Logo: upload and select a test image ("STEP LOGO"). The preview shows it. Send a test: the email received shows the organization's logo (Settings → Organization), not "STEP LOGO". On a client account, changing the organization's logo changed the logo received; the step setting had no effect. Proposal Send the logo selected in the step. Or, if the organization logo is meant to win, remove the step setting and show the organization logo in the preview.