Change Ledger

Most SEO work leaves no trace beyond a ticket. Three months later nobody can say what changed on a page, who agreed to it, or whether the lift in the report came from that change or from the season. Without a before state there is nothing to roll back to, and without a date there is nothing to measure from.

Google’s guide to traffic drops lists algorithmic updates, technical issues, seasonality and changing interests among the main causes to tell apart^[1]. A dated record of your own changes takes one cause off that list, or puts it at the top.

A ledger turns changes into evidence. It is an audit trail for a client or a legal team, the starting point of every verification, and over time the record of which kinds of change worked on this site.

The idea is old in data work. The W3C’s PROV documents describe provenance: the record of the entities, activities and people involved in producing a piece of data, so its quality and reliability can be judged^[2]. A change ledger is provenance for a website. QueryRev’s study of AI crawler access keeps the same kind of record for the web: it reads one fixed panel of 9,973 domains again every month and logs every domain that started or stopped blocking a crawler by name^[3].

Agencies and in-house teams need the same record for different reasons. An agency shows a client exactly what it did and what each change produced; an in-house team shows finance which work paid for itself. Both need the before state, the date and a result measured the same way every time.

A ledger is judged by its completeness: the share of changes with a recorded before and after state, an approver, a publish time and a verification result. An entry is closed when its outcome has been measured against a baseline, not when the change is published.

Completeness is checked on every change, not on a sample: one change missing from the ledger is one result nobody can explain.

Read the ledger by kind of change as well as by page. Over a year, the share of title rewrites, merges or redirects that produced a measured lift is the evidence for which work to repeat on this site.

Sample data: an invented site, to show the method.

An approved Update decision rewrites a product page’s title. The entry holds the old title, the new one, the approver, the path it was applied through and a 14-day verification window. One step restores the old title.

Fourteen days later the entry gains its result: the title’s click-through against the baseline, with the window and the comparison stated. Then the entry closes.