The UK’s Office for National Statistics has been rebuilding its Labour Force Survey since at least 2023. Its August 2026 update names November 2027 as the most likely switch date. Until then, the old and new surveys run side by side.
Your tracker won’t get four years and a crossover study – it gets a contract renewal and a client who reads Wave 15 next to Wave 14.
You can still migrate a tracker without losing trend data. It takes three things: a new data file built to match the old one, proof from data, and a plan agreed with the client before the first new wave goes to field.
What a Lost Wave Costs
A lost wave is one you can’t put next to the last one. The data is there. You just can’t draw a line between the two waves and trust it, and that’s the real risk when you migrate a tracker.
The bill comes in pieces. First, the wave you already paid for: fieldwork, sample and analysis, all for a number that now lives in a footnote. Then the rework. Tables get re-run, someone re-checks the script on a Saturday, and once in a while the whole thing goes back to field.
Then there’s the client call. Somebody has to explain a five-point drop, and the true answer is a setting. Nobody wants to tell a client their brand didn’t lose five points of trust. The slider did.
The trust part lasts longest. A client who doubts Wave 15 starts squinting at Wave 9 too. A Research Live opinion piece on tracker transitions makes the same point: switching suppliers can reopen questions about past results, even the ones that shaped big decisions.
So when we scope a migration, we start with the trend. Which numbers does the client watch, and what would it cost to lose them? The programming hours come after. Everyone can see the build. The wave is the thing you can’t buy back.
Where Trend Breaks Come From
When you migrate a tracker, the questions usually move intact. Trend breaks come from how the new platform asks them.
Pew Research Center saw this on a large scale. In a 2014 experiment, it asked the same question by phone and online. By phone, 62% said gays and lesbians face a lot of discrimination. Online, 48% did. Same question, same moment, a 14-point gap. Pew treated it as an effect of the method, with no change in opinion behind it.
A platform switch is a smaller change than phone to online. Still, the mechanism is the same: change how a question is shown, and some answers change with it.
So here’s a platform-sized example. In a 2011 experiment with 779 web respondents, published in Social Science Computer Review, researchers swapped radio buttons for a slider on the same rating question. On one item, middle answers fell from 21.0% to 11.6%. Slider answers also took more than twice as long.
Forsta’s own guide to tracker migration draws the line in a useful place. Cosmetic changes like logos, fonts and colours don’t have a significant impact on data. Structural changes do: a new question type, an added answer option, a different scale.
Because of that, the risk list for a buyer is short. Question types, answer lists, codes, and the data file your tables depend on. The full technical list is in our tracker migration checklist.
Three Ways to Protect Trend Data When You Migrate a Tracker
There are three common ways to handle the move. They cost very different amounts, and they prove very different things.
| Full Parallel Run | Soft Launch With Data Matching | Rebuild and Check the Screens | |
|---|---|---|---|
| Extra cost | Sample on both platforms for one or more waves | Test respondents plus a small early sample | Lowest on the quote |
| Extra time | One or more waves | Days | None upfront |
| What it proves | The size of any platform effect, on real respondents | That the new data file matches the old one, code by code | That the survey runs |
| Best for | Flagship trackers and headline metrics | Most trackers moving like for like | Nothing you plan to trend |
Forsta suggests a parallel run of at least a few waves, with the old platform still carrying most of the sample at first, for example 75%. The AYTM tracker guide recommends pulling one large sample and splitting it at random between the old and new versions. It also warns that this is “where real dollars” get spent.
For most trackers, we’d choose the middle column. The data file gets compared code by code before a single real respondent sees the new build. Then a soft launch shows whether early results sit where last wave did. A parallel run earns its cost on flagship trackers, where one headline number drives a board decision.
The third column is the cheapest one on the quote. It’s also the most expensive one in the debrief.
Convert the Script. Don’t Retype It.
When you migrate a tracker, most of the risk enters at one step: someone rebuilds it from the questionnaire. Every retyped question is a fresh chance for a new platform default to sneak in.
So we start from the code you already have. Automated conversion moves the existing script to the new platform in hours, where a hand rebuild takes days. Then our Bright MR operations team checks the result by hand: logic, routing, codes, and the platform quirks automation misses. Then comes the code-by-code data comparison. Every tracker gets checked twice before it reaches you.
So the time saved goes where it should. Into the checking, not the typing.
How much you hand over is your call:
- DIT (Do It Together): our specialists work alongside your programmers through the move. Your team keeps the know-how for the next wave.
- DIFM (Do It For Me): our team runs the migration end to end. You review the parity results and approve them before Wave 15 goes to field.
Either way, the decisions stay with you.
If a Line Moves Anyway
Migrate a tracker carefully and some lines will still move. The AYTM guide is blunt about it: “The data will never perfectly align.”
So what matters is who hears about it first. Agree with the client, before field, what counts as a real change and what counts as a platform effect. Then hold to it.
Pew shows one honest way to report it. When the method changes, the trend line continues as a dotted, labelled line with a note, for several years. Nobody pretends the switch didn’t happen. If you do need to adjust a trend, AYTM advises adjusting the old data to the new baseline. Forcing new data to match the old creates distortion that never goes away.
Five Questions to Ask Before You Sign
Ask these of any team that will migrate a tracker for you, including us:
- Will the new build start from our existing script, or from the questionnaire?
- How will you prove that Wave 15’s data file matches Wave 14’s, code by code?
- Who signs off parity, and what does “match” mean in numbers?
- Are we changing anything else at the same time?
- What do we tell the client if a line moves, and when?
Question four is the one teams skip. A new platform feels like the perfect moment to fix that grid everyone hates. IntelliSurvey’s advice is to hold changes until the transfer is confirmed. We agree. Move first. Improve next wave. Otherwise nobody can say which change moved the line.
Quick Definitions
- Trend data: the results of every wave of a tracker, lined up side by side.
- Trend break: a point where the new wave can’t be compared directly with the last one.
- Parallel run: fielding the old and new versions of a tracker at the same time and comparing results.
- Soft launch: a small first batch of real respondents, checked before full field.
- Parity: proof that the new build produces the same data file as the old one, code by code.
Where to Go Next
Before you decide how to migrate a tracker, read the specialist side of the same move: Tracker Script Migration: 3 Places a Platform Move Breaks.
It shows what breaks in the script, and where.
For the full list of what has to survive the move, use the tracker migration checklist.
