Compare chat statistics over time with Rematch

1. Compare the measure, not the story around it

Compare chat statistics over time by using the same indicator definitions on an earlier and later report, then inspecting only values available in both. A delta shows that the measured value changed in those samples. It does not prove why it changed.

Tellune groups a later Rematch with the earlier report from the same conversation. The comparison works on aggregate indicator values, not a new reading of hidden intent. If you need the definitions first, start with the guide to reading a WhatsApp chat.

2. Three checks decide whether a delta belongs

Rematch keeps a value only when it passes three compatibility checks:

1. The calculation definition and version match. Reports created under different definitions are not compared.
2. The indicator is available in both reports. A missing sample on either side removes that value from the delta list.
3. The unit matches. A percentage is compared with a percentage; minutes are compared with minutes.

These checks prevent the app from filling a missing value or comparing unlike measures. They also explain why two complete-looking reports may produce only a few deltas.

Direction is a separate layer. Tellune looks at the goal selected for the later report. Movement toward 50 can serve a balance goal for some indicators; lower or higher can serve other goals; some changes remain unclear. An increase is not automatically better, and a decrease is not automatically worse.

3. Different indicators keep different units

Each row keeps its own definition, sample, and unit.

Indicator	Comparable value	What it does not establish
Message balance	Your share of confirmed messages · %	Whether the conversation is healthy
Who starts	Your share of session starts · %	Why either person reopened the chat
Reply pace	Your median burst-to-reply time · min	How much either person cares
Depth trend	Later-half average · messages per session	Emotional depth or closeness

Here is a hypothetical example: message balance moves from 62% to 55% for “you,” a change of seven percentage points, while reply pace remains measured in minutes. Those values belong in separate rows. The first moved closer to an even split; that does not tell you whether the conversation improved.

For the underlying measures, see how message balance is calculated and why reply pace is not a caring score.

4. A changed number does not identify the cause

Many things can change between two imports. Rematch does not observe those outside events and does not isolate one cause.

The defensible sentence is: “This indicator changed between these two reports.” A stronger sentence, such as “this action fixed the conversation,” is not supported by the delta.

You can still make the comparison useful. Record what you intended to observe, read the targeted indicator first, and keep other changes as context. That is disciplined observation, not causal proof.

5. Not enough comparable data is a valid answer

An indicator may be unavailable because the later import lacks enough sessions, timed replies, bursts, or timestamped messages. A version change or unit mismatch can also block comparison.

Tellune returns no delta for that row. It does not estimate what the value might have been.

The next move is to check the unavailable reason. If the source has lost timestamps, another import of the same text will not repair a timing measure. If the conversation simply lacks enough supported events, wait for a later real sample rather than manufacturing one.

6. Keep the next comparison narrow

Use a flat or unavailable value to narrow the next question, then choose one defined measure and its limitation for the next comparison.

Rematch is a Premium feature. Tellune’s iPhone release is still in preparation.