Markopolo

What is Markopolo AI?

Markopolo AI is a customer interaction and marketing automation platform provided by Markopolo AI Inc. to e-commerce, DTC, and retail brands. It integrates first-party behavior tracking, customer profiles, audience forecasting, content generation, omnichannel outreach, and revenue attribution within a single system.

The focus of current products has shifted from traditional ad management to autonomous customer interaction. The system uses MarkTag to detect events such as views, clicks, time spent on a page, items added to the cart, and purchases; the built-in Athena model then analyzes the user’s intentions in order to determine the best time, channel, product, and content to present to each visitor.

Main functions

• MarkTag behavior tracking: It collects events such as page views, clicks, scrolling, time spent on pages, use of the shopping cart, checkout processes, and purchases, in order to create session and behavior profiles for anonymous visitors. Teams can initially use client-side scripts for quick integration, or they can set up their own subdomains and server-side tracking methods to reduce data losses caused by ad blockers.
• Composable customer data platform: It integrates anonymous behavior data, identified users, CRM information, and transaction records into a unified profile, thereby providing a consistent context for the same customer across different channels. Information such as name, email address, or phone number is included in the identified profile only when the business uses the identification interfaces intentionally or inputs that information manually.
• Prediction of Athena’s behavior: The platform describes Athena as a large-scale event model with 4 billion parameters, which uses real-time micro-behaviors to determine purchasing intent, price sensitivity, category preferences, the likelihood of customer churn, and response probabilities. The outputs are used to rank customers and trigger appropriate actions; however, the prediction scores still need to be verified using the company’s own business data.
• Personalization on a one-by-one basis: Different messages, product recommendations, as well as timing and channels for delivery are chosen based on each visitor’s browsing history, shopping cart contents, purchase history, and current intentions. This approach offers a greater level of detail compared to sending messages to entire groups in a uniform manner; nevertheless, the team should still restrict the use of sensitive data and avoid any form of unfair differential pricing.
• Browse Abandonment Recovery: Identifies visitors who leave after viewing a product, and sends them relevant products, reminders, or further information via appropriate channels. This approach is suitable for stores with high traffic but low add-to-cart rates; it is important to control the frequency of such messages to avoid causing annoyance.
• Recovery efforts for abandoned carts and checkout processes: An automated sequence is triggered when a user adds items to their cart or starts the checkout process but then abandons it; personalized messages are generated based on the products selected, the amount owed, the time elapsed since the user left, and their past behavior. The system can use various channels such as email, text messages, WhatsApp, push notifications, and voice calls to reach the user, with the choice of channel depending on the package chosen, the contact information available, and the user’s preferences.
• Automatic email sending: AutoSend is used for sending emails at various stages of the customer lifecycle, such as welcome messages, messages for customers who abandoned their browsing session, attempts to encourage them to complete their purchase, updates regarding stock levels, messages aimed at retaining customers, follow-up purchases, and product recommendations. AI can generate the subject lines, body text, and dynamic content, but the brand still needs to verify the prices, inventory levels, validity of any discounts, and the process for unsubscribing.
• SMS and WhatsApp: Send messages with a high intent to remind customers, send recovery messages, or share information about events – provided that these customers have valid phone numbers and have given their consent. The amount of SMS messages that can be sent is limited by the package plan; WhatsApp automation is part of the customization options available for businesses. The costs associated with sending messages, as well as the rules applicable in different regions, must be confirmed in the settlement details or contract.
• Push notifications: They are used to reach authorized users again through web or mobile alerts, and are suitable for sending updates regarding stock levels, price changes, the shopping cart contents, and delivery deadlines. Push notifications are part of a customized solution; the client side needs to store and transmit the device token.
• AI voice calls: Real-time, conversational follow-up calls are made to shoppers who have phone numbers; the scripts for these calls can be generated based on the shopper’s name, items in their shopping cart, amount spent, time of departure, and purchase history. This step is skipped if no phone number is available, and it is also subject to the regulations regarding telemarketing, recording, and automated calls in each region before it can be used.
• Smart product recommendations and dynamic pages: Products are selected based on the user’s intentions, and the banners, home screen content, action buttons, or various page elements are updated dynamically. Before going live, it is necessary to set limits on available inventory, minimum profit margins, brand guidelines, and content approval processes.
• Analysis and attribution: The dashboard tracks conversion rates, average order value, interaction rates, clicks on emails or messages, as well as the revenue generated through attribution, thereby helping the team determine the contribution of each campaign and channel. First-party tracking can help fill some data gaps, but transactions across different devices, cases where users refuse to provide consent, and offline transactions still affect the accuracy of attribution.

Input, processing, and output

Usage process

1. Create an account and select a store or a custom technology stack; first, determine the expected number of events, emails, text messages, as well as the channels required.
2. Create data sources and MarkTag containers in the Data Room, and decide whether to use client-side tracking or server-side tracking with a custom subdomain.
3. Shopify can be installed without writing any code through apps, while WooCommerce makes use of dedicated plugins; for other websites, a Web SDK can be used, and for mobile applications, Flutter or React Native SDKs can be chosen.
4. Events such as page creation, adding to the cart, checking out, making a purchase, and requesting a refund are handled, with consistent handling of product identifiers, amounts, currencies, customer IDs, and consent status.
5. Connect to an existing CRM or import customer data, and first test whether anonymous sessions and profile information after logging in can be merged correctly, in order to avoid duplicate contacts.
6. Set the audience, objectives, frequency, silent periods, allowed channels, discount rules, and exit conditions, while reserving approval for sensitive or high-value campaigns.
7. Start with a scenario of abandonment, preview the messages and recommendations generated by AI, and use test contacts to verify links, variables, unsubscribe options, and channel delivery.
8. After publication, it is necessary to monitor revenue recovery, gross profit, cancellations, complaints, rejected calls, customer retention, and attribution; success should not be judged solely based on the amount of revenue recovered.
9. Regularly verify package events as well as the sending quotas, permissions, domain name configurations, SDK versions, consent logs, and data deletion requests.

MarkTag integration method

Prices and packages

The Starter page indicates that it can be started without a credit card; the Growth, Pro, and Custom plans all offer a 30-day free trial. The page showing the public prices contains some duplicates and misaligned fields. Press releases have also mentioned that, in addition to the basic subscription, fees might be calculated based on revenue generated, so the actual costs, additional pricing, message costs, and commission amounts should be referred to on the account settlement page or in the sales contract.

Cancellation, Refunds & Migration

• The pricing page states that the team can migrate customer data, audiences, automation settings, and campaigns for free, but the scope of the migration, the time required for it, and procedures for dealing with any failures must be specified in a written plan.
• The official legal page states that it is possible to cancel the subscription when no campaigns are running; it also mentions that refunds will be processed within seven working days, and the advertising credits will become invalid temporarily.
• The same legal text states that, unless otherwise agreed, termination generally does not entitle one to a refund, and the outstanding amount still has to be paid; in the case of termination due to a serious breach, the prepaid fees for the period after the date of termination can be refunded.
• The Legal page makes extensive use of terminology related to older products such as advertising content, advertising platforms, and advertising credits; it is not fully in line with the current customer interaction packages. It is necessary to verify whether these cancellation terms apply to the current subscription, message usage, and revenue sharing before making a purchase.

Which users are it suitable for

• A e-commerce brand that aims to integrate emails, text messages, WhatsApp, push notifications, and voice calls into the same customer journey.
• Shopify and WooCommerce stores with a high rate of page views, add-to-carts, or abandonment at the checkout stage, that wish to automatically recover customers based on their real-time intentions.
• They already have a CRM, but the behavioral data and communication channels are scattered; DTC teams that aim to create a unified customer profile.
• Growth and data teams that require tracking by a first-party entity or the server side, and that have the capabilities for DNS management, SDK usage, and consent management.
• Partners who wish to integrate MarkTag into e-commerce platforms, website building tools, or agency systems.

Advantages

• It creates a closed loop that spans behavior collection, prediction, and cross-channel execution, thereby reducing the need to transfer audiences between CDPs, email, and messaging tools.
• The free plan offers a higher quota for events and emails, making it suitable for testing tracking mechanisms, attribution methods, and abandonment processes before opting for a paid version.
• It also provides clients, servers, website plugins, mobile SDKs, and partner APIs, covering various engineering capabilities.
• The anonymous-first design separates the default behavior data from the identity information provided by the merchant, thereby facilitating data minimization.
• The Partners API allows for the creation of merchants, the generation of tracking tags, domain verification, and the retrieval of events; it is suitable for integration in multi-tenant platforms.

Capabilities boundaries and limitations

• The figures provided by the platform regarding the recovery rate, model benchmarks, and marketing improvements are stated in the product descriptions or press releases; they do not guarantee that specific stores will achieve identical results.
• First-party tracking on the server side can reduce technical interference, but it does not mean that cookies, electronic marketing practices, and user consent requirements can be ignored.
• Predictions depend on events, identity merging, and the quality of product data; incorrect currency values, duplicate purchases, or lost refunds can distort the attribution of revenues.
• Fully automated messages can result in incorrect discounts, repeated deliveries, or an inappropriate tone; important campaigns require testing, frequency limits, and manual approval.
• AI-powered calls, text messages, and WhatsApp messages are subject to limitations related to the quality of the numbers used, the region, the service provider, template approvals, and marketing regulations; therefore, not all contacts can use them.
• The mobile SDK does not capture all of the default events of the Web SDK automatically; the development team must send the relevant events explicitly, as specified in the documentation.
• The pricing rules, as well as the procedures for cancellation, on the price page, the press releases, and the page detailing old laws are not entirely consistent; therefore, it is not possible to estimate the total cost based on just one of these pages.

Status of Partners API, SDK, and open source versions

Markopolo provides public development documentation; MarkTag covers Web, Flutter, React Native, and server-side events. The Partners API is intended for platforms and agents, and it uses Bearer Tokens to manage merchants, generate and verify MarkTags, as well as read events. Partner tokens must be obtained by contacting the team, and the base addresses mentioned in the documentation still contain alpha identifiers; stability in production and service levels need to be confirmed separately.

1. Create a MarkTag container in the Markopolo workspace, and select client, server, or pre-verified integration.
2. Ordinary merchants install the SDK or plugins provided by the platform; partners contact the relevant team to request API tokens, storing the keys solely on the server side.
3. Create a merchant, save the identifier returned for that merchant, and then generate a corresponding tracking tag for it.
4. Install the client script, or configure DNS for first-party tracking; initialize the SDK on mobile devices and explicitly log events.
5. Send a test event to verify the status code, user identifier, product, amount, and currency, and check on the dashboard to see if it has been recorded.
6. Before going live, implement retry, rate limiting, key rotation, consent gating, request deletion, and filtering of sensitive fields.

• The public npm package for React Native is currently licensed under the MIT license; it includes TypeScript declarations and points to the repository of the Markopolo organization.
• Making the SDK package available publicly under the MIT license does not mean that the Markopolo SaaS platform, the Athena model, the backend services, the marketing materials, or the trademarks are made open source.
• The edit links on the document pages point to the Markopolo organization, but some of those edit addresses are currently inaccessible; therefore, it cannot be concluded that the entire document repository is publicly available.
• The specific licenses for the Flutter SDK, Web SDK, WordPress plugins, and server-side components must be checked separately; the MIT license of the React Native package cannot be applied in these cases.

Privacy, Security, and Data

• Markopolo acts as the controller of data on its own website and accounts, while it typically functions as a processor for the data that merchants send in through trackers, SDKs, or other systems; the merchants remain responsible for ensuring compliance with legal requirements and informing users accordingly.
• By default, it collects anonymous session identifiers, page views, time, clicks, scrolling, dwell time, browser information, system details, device information, and approximate location when permitted.
• Name, email address, phone number, and customer ID are processed only when the merchant provides them voluntarily through calls, logins, forms, or other explicit integration methods; the platform states that it does not automatically extract such identity information from pages or forms.
• The privacy policy requires clients to manage consent for unnecessary tracking, and to control the initiation of data collection through consent mechanisms when necessary. The use of proprietary domains or first-party cookies does not exempt them from this responsibility.
• The platform states that TLS 1.3 is used for data transmission, and AES-256 is employed for static data; it also incorporates mechanisms for authentication, role-based access control, monitoring, logging, as well as measures to ensure employee confidentiality. Enterprises should still request verifiable security and audit documentation.
• The retention period for data varies depending on the data type as well as contractual and legal obligations, and there is no fixed number of days; once the retention period expires, the data is deleted, anonymized, or aggregated.
• Data may be processed across borders; for data transferred from the European Economic Area, the UK, or Switzerland, standard contract clauses or other legal mechanisms are employed.
• Users may request access, correction, deletion, restriction, portability, objection, or withdrawal of consent; when the platform acts as a processor, such requests should generally first be submitted to the relevant business.
• Input masks can help prevent passwords or payment details from being captured, but teams should still proactively remove any unnecessary sensitive data related to health, finance, minors, and other areas.

Content, Copyright, and Commercial Liability

• Businesses must possess the necessary rights to use customer lists, product materials, trademarks, and advertising content, and must ensure that the information is accurate and lawful.
• When creating email, SMS, and voice scripts, it is necessary to verify facts, prices, inventory levels, promotional terms, disclaimers, and the brand’s tone.
• Each channel should provide a legitimate mechanism for unsubscribing or refusing to receive messages, and should keep records of consent, deliveries, unsubscriptions, and complaints.
• The legal provisions grant the platform broad permission to use feedback, while the rights related to the service itself remain with Markopolo; customer data, the content that is generated, and the purposes for model training shall be further specified in the current contract and data processing agreement.
• The use of the platform for personal or household purposes by consumers is prohibited; the terms specify that the service is intended for businesses and professional users only.

Summary

Markopolo AI is suitable for e-commerce brands that wish to utilize real-time behavioral data to enable cross-channel recovery strategies and personalized approaches. Its key features include MarkTag, Athena, and a set of tools for autonomous operation across multiple channels. Before going live, it is advisable to first use the free version to test the data and the incremental value it offers; attention should also be paid to any additional costs, revenue sharing arrangements, permissions for automated outreach, consent management, and the suitability of existing legal provisions for the current product.