How we built an SEO intelligence layer for LeverX.
LeverX competes for organic search against a field of global SAP integrators, and the intelligence that guided it was pulled by hand: every week a specialist exported SEMrush reports and rebuilt the competitive picture in spreadsheets that were already a week old. We gave the marketing team an intelligence layer instead. One self-updating platform tracks leverx.com against a hand-picked competitor watchlist across four markets every day, and turns raw ranking data into the analysis the team actually acts on: content gaps, striking-distance keywords, cannibalization, and newly published competitor pages, with reporting no one could assemble by hand. It is multi-user by design, so the whole marketing team works on it with role-based access instead of one person's spreadsheet, and a plain-language AI assistant answers any question grounded in the platform's own live data. The dashboard never calls SEMrush live: every view is computed on read from an owned Postgres cache, refreshed on its own every day. It runs in production on LeverX's own cloud, and they own every line of it.
anfloy.
The challenge
LeverX competes for organic search against a field of global SAP integrators, and the intelligence that guided it was pulled by hand. Every week, a marketing specialist logged into SEMrush, exported rankings, and rebuilt the competitive report in spreadsheets.
The data was a week old the moment it was done, the analysis only happened when that one person had time, and the whole competitive picture lived in files no one else could open. The team could not self-serve, and it could not answer a question the moment it mattered. That was the problem.
One intelligence layer over their SEO
We replaced the manual workflow with a single intelligence layer over LeverX's entire search footprint. One platform tracks leverx.com against a hand-picked competitor watchlist every day, and turns raw SEMrush data into decisions: rankings, gaps, alerts, and trends in one place, always current.
The decision underneath it: the platform never calls SEMrush live. Ranking data is ingested into its own Postgres cache, and every view is computed from that cache on read, then refreshed on its own by a daily scheduler. The result is one source of truth that is instant, always available, and never a week behind again.
Built for the whole team
It is multi-user by design. The whole marketing team works on one platform through role-based access, owner, admin, and viewer, each person seeing exactly what they should, with one-time temporary passwords and a full audit trail of who did what.
The expertise that used to live in one specialist's spreadsheets now lives in a system the entire team can open, query, and act on. Competitive intelligence stopped being one person's job and became something the whole team runs on.
Reporting no one could do by hand
On top of the data sits the reporting that makes it powerful. Content gaps surface the keywords competitors rank for that LeverX does not, scored by opportunity. Striking-distance finds the keywords sitting one push from page one and drafts a meta title, description, and content angle for each. Cannibalization flags where two LeverX pages compete for the same term, and new-page tracking resolves a competitor page's true publish date, so the team sees what rivals actually just shipped.
All of it pivots by market across the US, UK, DE, and PL, holds a full history of every snapshot, and exports clean. It is reporting no one could assemble by hand, produced automatically, every day.
Ask the intelligence layer
The whole platform is queryable in plain language. A streaming AI assistant answers questions about LeverX's rankings grounded only in the platform's own live data, and it reads the exact numbers on the dashboard, so it can never drift from what the team sees.
Ask where a competitor overtook us last month, or which gaps are worth chasing this quarter, and the answer comes back built from the real data and cited to it, not a generic guess.
Built like production, built to own
It is built like production software, not a demo. Role-based access and an audit log, shipped as Docker containers with a continuous-integration pipeline that boots the whole stack and smoke-tests it on every change, plus a full set of handover docs so LeverX's team can run it without us.
It runs in production on LeverX's own cloud, Railway for the backend, Vercel for the front end, with their own keys. LeverX owns every line of it.