Definition
A research backlog is the one list in which a conversion research programme records what it finds, every finding a row with the same fields. The guide to conversion research methods puts findings of all three kinds in it: problems that may be worth testing, broken things to fix, and what is already working and should survive the next redesign. Every row ends in a decision, so the backlog holds the team's conclusions and not only its observations.
Its unit is a finding, not an idea for a change. A row starts from something visitors fail to do at a particular place, backed by the methods that saw it. The change that might help comes later, if at all.
What a row records, and how the backlog differs from a test queue and a learning log
The conversion research sprint tutorial writes every row with the research hub's fields: a one-sentence statement of the problem, framed around what visitors cannot manage; where it happens; what each research method showed; how many visitors pass through that point; a range for what it probably costs; where the row stands; and the decision. Keeping the same fields on every row lets findings from different methods be read side by side.
Two neighbouring lists are easy to confuse with it. The queue of tests that the prioritisation guide orders holds hypotheses, each a proposed change with an expected effect; only some backlog rows ever become one. A learning log records experiments once they have run, losers and inconclusive results included, so it starts where the backlog's hypotheses finish.
In practice: research methods, sprints, expert reviews, recordings and accessibility
The sprint tutorial sets the backlog up before the first finding arrives and closes with rows and a short readout rather than a deck of observations. Bugs, such as a field that refuses valid input, are fixed straight away and the fix noted in the row; problems whose fix is uncertain become candidates for hypotheses; the rest are parked with a reason.
Some sources supply leads rather than evidence. The guide to expert review of a conversion journey enters each issue with the review as its source and the evidence still needed, and only issues a second method confirms move on towards a hypothesis. The tutorial on session recordings and heatmaps adds a finding once the behaviour meets an evidence rule agreed beforehand with whoever owns the backlog, then fixes it if it is a bug or plans a second source if it needs one.
Accessibility barriers are not a project of their own. The guide to accessible landing pages records each barrier in the same backlog, noting its location, which check caught it and what it stops visitors doing, and treats barriers as bugs to fix rather than ideas to test.
A stationery shop turns two weeks of research into rows
Delviax Stationery, an invented online stationery shop, spent 2 weeks researching its checkout, where 3,100 of the 5,400 visitors who reached the delivery step each month went no further. Recordings of 60 sessions, a poll on the step with 92 answers and 5 usability sessions produced 11 findings, each written as a row. One was a bug: the postcode field rejected valid postcodes typed with a space, seen in 9 recordings and 2 sessions, so it was fixed at once and the fix noted in its row. Of the rest, only 2 problems had support from more than one method, unexpected delivery charges and a confusing gift-message option, and only those became hypotheses. The other 8 rows were parked, each with a reason, some until a second source could check them.
Related terms
A HypothesisA testable statement of the change to make, the effect expected on a named metric and the evidence-based reason for expecting it. is what a strong row can become: a proposed change, the result it should have on a named metric and the reasoning behind it, which the row's evidence supplies. A Session RecordingA reconstruction of one visitor's visit, replaying their clicks, scrolling, taps and form interactions, used to see how a page is actually used. and a Usability TestA research session in which a person attempts a realistic task on a site, app or prototype while an observer notes where they struggle. are two of the methods whose observations feed rows, and a finding seen by one method alone is treated as a lead until another confirms it. Once a hypothesis runs as an A/B TestControlled comparison of two variants., its result belongs in the Learning LogA single, searchable record of every experiment, losers and inconclusive results included, with the hypothesis, design, result, decision and what it taught., while the backlog row's status shows where that problem stands.
Common confusions
- It is the test roadmap. Rows hold problems with evidence, and a row can end as a fix or a parked note without ever becoming a test. Ordering tests is the prioritisation guide's job, applied to hypotheses.
- Bugs need a test too. A broken field or a wrong price is repaired without a test, with a note of the repair in its row.
- One method is enough. A row can start from a single source, but that source is a lead until a second method backs it up.
- Only problems belong in it. The methods guide also records what is working, so that a redesign does not remove it by accident.
- It is the same as a content or SEO backlog. Those lists hold planned pieces of work. This one holds conversion research findings and what was decided about each.
