Search used to be one box. Now it is two: the familiar blue links, and the AI answers that ChatGPT, Perplexity, Copilot, Gemini, Claude and Grok hand back before anyone clicks a thing. A monthly SEO report that sits in a PDF cannot keep pace with either of them. Waggle is our answer: one service that keeps you found across both, and keeps making it you as the data moves.
Waggle is a discoverability service, not a dashboard you have to remember to open. It watches where you show up across search engines and answer engines, works out what to change, and then actually makes the change — publishing pages, fixing titles and schema, closing the gaps where a rival is cited and you are not. The rule underneath it is short: the report is the work, so skip the report and do the work.
The problem: search split in two
Ranking fourth on Google no longer means what it did. When someone asks an assistant instead of typing a query, the ten blue links never load — a model reads the web, decides who to cite, and answers in a paragraph. You can be the best result on Google and completely invisible in that paragraph. Being found now means being found on both surfaces, and each moves on its own clock.
One signal, five streams
Most SEO tools look at one thing — keywords, or backlinks, or analytics — and leave you to join the dots. Waggle joins them. It pulls five streams into a single picture: search data (what you rank for and where), analytics from Scout (who actually arrives and what they do), edge crawler logs (which of your pages the AI crawlers are reading), the answer engines themselves (who they cite for your queries), and raw server logs (what really got requested). Siloed, each is a hunch. Together they are one signal good enough to act on.
Three modes: active, reactive, conditional
What Waggle does with that signal falls into three modes, running at once.
- Active. Always working, not a monthly report. Continuous optimisation, publishing and fixes across the whole site, every week.
- Reactive. It moves when the data moves. A ranking slips, a question starts trending, an algorithm shifts, an engine stops citing you — Waggle responds, often the same day.
- Conditional. The next move depends on what it finds. If a page is close, push it. If a query has no home, build one. If an engine cites a rival, close the gap.
The numbers it watches
Like the rest of the studio, Waggle only reports numbers you can point at the data for. Three sit at the front: your AI-answer share of voice (how often the engines cite you for the queries that matter), the count of pages read by AI crawlers (are the models even seeing your content?), and time-to-index for new and updated pages (how fast a change actually lands). When Waggle ships something, these are what move.
How it works, end to end
Underneath the service the shape is deliberately small:
- One collector. Search, Scout analytics, edge crawler logs, answer-engine citations and server logs are pulled on a schedule and normalised into a single per-page, per-query picture.
- One signal. Those streams are reconciled into one view of where you stand — on Google and in every answer engine — so a decision rests on all of them, not one.
- One loop. The active, reactive and conditional modes turn that signal into work: pages published, titles and schema fixed, gaps closed — then measured against share of voice, crawler reads and time-to-index.
The stack is the same one we run across the studio: SvelteKit on Cloudflare Workers, with the scheduled collector living in the Worker's scheduled() handler, Neon Postgres with Drizzle for the history, and Scout supplying the analytics stream. Nothing about it is a black box: every change Waggle makes is logged, attributable and reversible.
Why we built it
Waggle exists because “do some SEO” stopped being enough the moment assistants started answering for us. Being found is now a moving target across two surfaces at once, and a quarterly audit will always be looking at yesterday. It is one of several internal tools we have grown into products, alongside Scout, SwarmGen and Apiary — and it shares their instinct that the only number worth reporting is one you can point at the data for. If being found where people actually search now is the kind of thing you want handled for you, the way we work is a good place to start, the rest of the hive's notes are worth a read, or see the live service itself.