Why We Do Not Score Leads

A lead score compresses several facts about a prospect into a single number, and the compression is the point: it makes a long list sortable. What lead scoring throws away is when each of those facts was observed. In outbound to ecommerce stores the observation date is most of the information, so we do not score leads.

The disclosure first. We build Keaz Signals, which sells an ecommerce leads database with live buying signals attached, and we sell no scoring model of any kind. We are a competitor to tools that do. We have an obvious interest in you finding the score less useful than the event, so read the rest with that in mind.

What follows is a structural argument about what a number can carry, plus a description of our own product behaviour as published. There is no study here and no benchmark. Where we have not measured something, this article says so rather than estimating it.

Scoring earns its keep under three conditions together: more inbound than your team can work, a long consideration cycle, and a CRM that already holds months of history on each account. Where all three hold, a score is doing real work. It ranks a queue you cannot get through, using evidence you genuinely accumulated.

Picture a company whose demo form fills faster than the sales team empties it. Every account carries a trail: pages visited, emails opened, a webinar attended, a job title a rep corrected by hand. Fit a model to closed-won history and the queue it produces beats arrival order. That is a real problem, genuinely solved, and nothing here argues otherwise.

None of the three conditions describes an agency prospecting ecommerce stores. Our own site puts the qualifier plainly: the clearest sign we fit is that you have paid sales capacity that is not fully booked, which is the opposite of an inbound queue. There is no accumulated history on a store you have never contacted. And the cycle is short, because someone installed an email tool last week and is deciding what to do about it now.

A signal is an event with a date attached: this store started running Meta ads, launched a product, or installed an email tool, on this day. A score is a standing property with no date on it. That date is the part of an outreach message which could not have been written a month ago.

Two stores can hold the same score for opposite reasons. One has looked quietly qualified for a year. The other rebuilt its storefront on Tuesday. Compressed into a number they are indistinguishable, and only the second has a reason to reply this week.

Stability is the other cost. A score changes slowly, so the list it produces is roughly the same list every week, and every team running a similar model writes to the same founders. An event expires. The set of stores that changed something in the last seven days is smaller, different each week, and far less crowded. The event types we watch are listed in the buying signal catalogue.

We surface dated events and stop there. The shipped list: new Meta ads running, an active ad count rising, a product launch, an email marketing tool installed, social growth across Instagram and TikTok, newsletter activity, and store or site changes. Each arrives with a date, and nothing combines them into a rank.

As of September 2026 we watch 4.1 million Shopify and WooCommerce storefronts across Europe and North America, re-scraped weekly, with roughly 41,200 carrying a fresh signal in any seven-day window. Those are our own moving figures and they move, which is why they are dated. Stores arrive with contacts attached, including founder addresses that are not publicly listed, in the ecommerce leads database.

What is not shipped is stated rather than hidden: registry and firmographic triggers, and hiring signals, are marked planned. The sales agent writes per-store copy from the signal and your own knowledge base, and sending runs through your own Instantly workspace. None of that produces a probability that a store will buy, and we do not offer one.

We are not going to tell you how much better a dated signal converts than a lead score. That is the number you would most want from an article with this title. We did not run the comparison, we are not aware of a defensible published version of it, and an invented one would be worse than none.

We are also not going to publish an accuracy figure for any scoring model, neither a competitor's nor a hypothetical one of our own. Accuracy is a property of a model fitted to one company's history, and a figure produced under someone else's conditions is not evidence about yours. Any number we put here would be a measurement we did not take.

And no reply rates. Not for signal-led outreach in general and not for any signal type in particular. A reply rate is a property of your offer, your segment and your sending domains together, and a vendor controls one of those at most.

Without a score the sorting happens earlier, in the segment rather than in the queue. You define who is worth hearing from once, by country, platform, follower band and category, then work whatever fires inside that set this week. The ranking question mostly disappears because the set is already small.

Building that segment is a different exercise from fitting a model; we covered it in Customer Segmentation When the Customer Is a Storefront. Once a store fires, the signal type decides the opening rather than a number: one template per event, as in Cold Email Templates, One Per Buying Signal, with follow-up timed against the event instead of the calendar.

For the wider version of this argument, see Intent Data: What Each Category Can Actually See.

If you have an inbound queue, a score beats us and it is not close. We have nothing to say about ranking hand-raisers, we do not read your CRM, and we do not model your closed-won history. A company with those assets and a real prioritisation problem should score, and should not take advice about it from us.

The rest of where we lose belongs in the same breath. We do not export: leads move into campaigns and stay there, which for some teams is disqualifying and should be. There are no LinkedIn signals and no LinkedIn sending, and none are planned. We watch Shopify and WooCommerce stores in Europe and North America and nothing else, so a store on another platform or in another market is invisible to us however well it would score. We are GDPR-conscious by design and will not claim more than that; responsibility for what you send stays with you.

If the question you actually have is which stores moved this week, that is the instrument we built. Signups start with 1,000 free leads, which is enough to check whether your segment fires often enough to be worth a workflow.

• Our own product behaviour and the shipped signal types, as published on the buying signal catalogue, retrieved 17 September 2026.
• The audience qualifier about unbooked paid sales capacity, and the scope of platforms and markets, as published on the ecommerce leads database page, retrieved 17 September 2026.
• Own pool figures: 4.1 million Shopify and WooCommerce stores under watch across Europe and North America, re-scraped weekly, approximately 41,200 carrying a fresh signal in a seven-day window. As of September 2026. These figures move and are not permanent.
• No competitor accuracy figures, no third-party benchmarks, no signal-versus-score conversion comparison and no reply rates are cited here. That is deliberate, and the reason is in the section above.