The AI recommender reads the tools a website already runs and predicts which new technology it is likely to add. Type your product. Get a ranked list of sites that fit, with the reason for each pick.
Large retailers recommend products based on what you already bought. The recommender does the same for websites and technologies.
Instead of shoppers and products, it learns from millions of websites and the 4,000+ tools they run.
The same model answers questions for sellers, for store owners and for whole lists of domains.
Type your product. Get the sites most likely to add it.
Enter a domain, or type the tools you run. Get up to 1,000 recommendations.
Send up to 500 domains. Get recommendations for every one.
Most prospecting targets competitor users. That works, but every deal starts by asking someone to rip out a tool that mostly works.
Adding a tool is a smaller decision than replacing one. There is no migration to plan.
You are not fighting a contract price, so you do not need to undercut it.
"Sites like yours usually add this next" is easier to hear than "switch from your vendor".
Your rivals mine each other's users. Few are mining the empty slots.
Use competitor customer lists around rival price rises. Run both motions side by side.
A score alone is hard to act on. Each recommendation names the technology already installed that most drove it.
The reason becomes your first line: "since you already use X, Y tends to be the next step".
Sales teams ignore black-box scores. A visible reason gets the list worked.
When one tool keeps showing up as a driver, it is a partner or integration worth building.
| Recommended tool type | Typical driver | How to use the reason |
|---|---|---|
| Live chat | A helpdesk or ticketing tool | "Turn your tickets into real-time answers." |
| Loyalty program | Email automation and reviews | "You already retain customers, now reward them." |
| SMS marketing | Email automation | "Add the channel your email list already opens most." |
| Pay-later financing | High prices and a premium theme | "Lower the barrier on your biggest baskets." |
| Product reviews | Email marketing and upsell apps | "Give your emails social proof to link to." |
| Site search | A large catalog and analytics | "Help visitors find products in a big catalog." |
Drivers in the table are illustrative patterns. The actual driver is shown per site in the dashboard.
Combine the fit ranking with store size and momentum to decide who gets which treatment. Rep time is the scarcest resource on any sales team, so the matrix below exists to spend it where it pays back most.
Research the store, write by hand, offer a call or demo. These are your best accounts.
AI-drafted emails, self-serve trial or onboarding. Volume with relevance.
Check the record by hand. A special reason may still make it worth a call.
Keep for newsletters or retargeting. Spend rep time elsewhere.
Every one of 5 million domains carries an IAB category. That shows where your type of tool is used above average.
The result is a list that fits on two axes at once: vertical and stack.
Shopify usage in Style & Fashion compared with that category's share of all domains. Shopping, Food & Drink, Hobbies, Pets, Sports and Home & Garden also sit above 1.0.
From login to first replies in about a week. The steps are the same whether you sell a $20 app or a $2,000 platform; only the filters and the level of personalization change.
Stores if you sell to merchants. Popular domains if you sell more widely.
Type your own product, or the category leader if you are not listed yet.
Country, vertical and store size that match your pricing.
Split by fit and size into the four treatments above.
Lead with the driver tool. Use AI drafts for the long tail.
Track replies by driver. Double down on the drivers that convert.
Prospecting is the obvious use. Teams that keep the recommender for a year usually find these six jobs as well, often in other departments.
When your tool is an unusual fit for a customer's stack, that customer is more likely to leave.
Tools that often drive recommendations for yours are the integrations users expect.
Share recommender picks with a partner whose tool keeps appearing as a driver.
Show a client store the tools its closest peers run that it does not.
Enter your stack and see what similar stores use that you do not.
Export high-fit domains for account-based advertising.
The recommender behaves differently for every kind of tool. These are the patterns we typically see when vendors in each category run it.
High-ticket and technical catalogs where shoppers ask before buying.
Stores with email automation that lack social proof on product pages.
Beauty, pet and food stores already investing in retention.
Stores with mature email programs ready for a second channel.
Furniture, electronics and fitness stores with high average prices.
Coffee, supplements, pet food and personal care stores.
Stores with hundreds or thousands of products and steady traffic.
Stores with international shipping or multiple currencies.
Patterns are typical, not guaranteed. Your own run shows the real drivers for your product.
You never see the model, only its output. These are the kinds of pattern a good stack-based recommendation reflects.
Some tools almost always travel together. A site with one half of the pair is a strong candidate for the other.
Stacks grow in a typical order. A site's current stage hints at its next purchase.
What is standard in fashion is rare in electronics. Gaps against the norm stand out.
A stack full of paid tools signals a team that buys software readily.
Heads of sales are rightly skeptical of AI lead scores. Here is what they usually ask before a first campaign.
They do not have to. Every pick shows the driver tool, so a rep can judge it in two seconds.
Intent tells you when. The recommender tells you who fits. Use the fit list as the base and intent as a timing layer.
Run the recommender for the closest known tool in your category. Fit transfers well between similar products.
Filter by popularity rank and catalog size. The top tiers of 6M+ sites still give long lists.
Send up to 500 of your accounts for a bulk run. You learn which ones to prioritize before buying more data.
Run a recommender list next to your usual list for a month. Compare replies and meetings per hundred contacts.
Small enough to run alongside your normal pipeline work, big enough to give a real answer by Friday.
| Day | Task | Output |
|---|---|---|
| Monday | Run the recommender for your product in your main country. Apply size filters that match your pricing. | One filtered list of 100 to 500 sites |
| Tuesday | Read the driver column. Group sites by their top driver tool. | Three to five driver segments |
| Wednesday | Write one short message per segment, built around its driver. | A tested opener for each segment |
| Thursday | Generate AI drafts for the long tail. Hand-write the top 20 accounts. | A loaded sequence |
| Friday | Send the first batch. Set a reminder to compare replies with your usual list in four weeks. | A running experiment with a clear test |
Most teams already use the first two. The third closes the gap they leave open, and it is the one your competitors are least likely to be working.
| Industry + size list | Competitor users | Recommender picks | |
|---|---|---|---|
| Who is on it | Companies matching a sector and headcount | Sites running a rival tool | Sites whose stack predicts your tool |
| Fit to your product | Weak | Strong | Strong |
| Incumbent to displace | Unknown | Always | None in your slot |
| Built-in reason to reach out | No | Yes, the rival | Yes, the driver tool |
| How many rivals use the same list | Many | Many | Few |
| Ranking | None | None by default | By likelihood |
The recommender is part of both plans, Advanced and Enterprise.
Send up to 500 domains to [email protected] for recommendations on each.
Which technologies a website is likely to add, given the technologies it already runs. Flipped around, it predicts which websites are likely to add a given technology.
More than 100 million technology usage data points: 4,000+ technologies across millions of websites.
No. Recommendations are for technologies a site does not run yet.
Yes. Run the recommender for the closest well-known product in your category. The sites that fit it usually fit you too.
Up to 1,000 technology recommendations for a site. For a technology, the list of sites is capped by your plan: 100 rows per report on Advanced, 5,000 on Enterprise.
For each recommendation, we name the installed technology most responsible for it. You see why a site was picked.
Yes, by the same fields as other lists: country, category, size, age and authority.
Yes. Recommendations also cover the five million most popular non-ecommerce domains.
Send up to 500 domains and we return recommendations for each. It is a quick way to plan cross-sell.
It is a different motion. Recommender picks have no incumbent in your slot, competitor lists do. Most teams run both.
Yes. Export to CSV, Excel or PDF and use the reason column in your sequences.
Advanced at $999 per year is enough to test the motion. Move to Enterprise when 100 rows per report is no longer enough.
Ad lookalikes are built from people and stay inside an ad platform. The recommender works on websites and their tools, and gives you an exportable list with reasons.
Yes. Country, language, category, product count, price, age and authority all work on recommender results.
Monthly is a good rhythm. Stacks change, and sites drop out of the list once they adopt your category of tool.
Yes. Agencies run it for each client's product and deliver the exports. See the agency page.
It is a relative ranking signal, not a purchase probability. Use it to order your list: higher scores go first.
Stacks hide patterns people rarely notice. Check the driver: it usually explains the pick, and surprising picks often turn into untapped segments.
Sites that already run your technology are never recommended for it. For customers on other plans or brands, remove them from the export before you send.
Let 100M+ data points rank your next prospects, with a reason for every one.