The Summary Said "Mixed Feedback." Your Comment Said Something Else.
Photo by Photo by david jun on Unsplash on Unsplash
Back in March you did the thing. You opened the city's survey link, clicked through nine questions about the parks master plan or the transit study or the budget priority exercise, and then you got to the box marked "Anything else you'd like us to know?" and you actually typed something. Three sentences. Maybe four.
By July there was a results page. It said the city received more than 1,400 responses and that "key themes included connectivity, maintenance, and cost." Your three sentences were not among the key themes, or at least not in a form you'd recognize. That doesn't automatically mean nobody read it. It means it went through a process that turns sentences into tally marks, and the tally marks are what the council saw.
This is a guide to walking that path backward.
The Short Trip From Your Keyboard to a Bullet Point
Most municipal surveys run on a platform — PublicInput, MetroQuest, SurveyMonkey, or a homegrown form bolted onto the city website. When the survey closes, staff or a consultant exports the results. Closed-ended answers tabulate themselves. Open-ended answers don't. Someone has to read them, decide what each one is "about," and assign it to a category.
That's called coding, and it's normal. Nobody expects a planner to hand a council member 1,400 free-text paragraphs.
What varies wildly is what happens to the paragraphs afterward. Some cities keep a verbatim appendix and attach it to the agenda item. Some keep it in the project file and never publish it. Some keep only the coding log — the count of how many comments mentioned each theme — and let the original text sit in a folder until the retention schedule says it can go.
The public-facing result is a summary memo, usually five to ten pages, often with a presentation deck on top of it. That document is what gets described as "what we heard." It's a translation, and translations can be checked.
Ask for the Raw Set in Plain Language
You don't need to cite a statute to make a public records request, though naming your state's open records law in the first line tends to shorten the back-and-forth. What you need is specificity. Vague requests get vague answers, and vague answers get deadlines extended.
Ask for the following, all in one request:
- Every open-text response submitted through the survey, with the date submitted and a response ID (or whatever identifier the platform assigned), delivered as a CSV or Excel file.
- Any coding framework, codebook, or theme list used to categorize those responses.
- The final summary memo and any slide deck or presentation derived from it.
- Any consultant deliverable that the memo draws on, including drafts if the contract required them.
Add a line asking for a written fee estimate before work begins, and a line asking for a fee waiver on public-interest grounds if your state allows one. Ask for the records electronically — many agencies will do it by default, but saying so removes an excuse.
One thing worth knowing: if the survey never collected names, there's nothing to redact, and a city that claims a privacy exemption for anonymous comments is worth pushing back on. If names were collected, expect them to be stripped. The substance of the comments should survive either way.
Put the Two Documents Side by Side
Once you have both, the comparison is mostly arithmetic and patience.
Start with the counts. If 1,400 people answered the closed-ended questions but only 300 left a comment, that's a real finding — and it may explain why the summary leans on the multiple-choice data. If 900 people left comments and the summary mentions "several" of them, you've found a gap worth asking about.
Then look at the buckets. A good summary will tell you how comments were distributed across themes, including how many landed in "other" or "general comments." That last number matters. In a lot of coding schemes, "other" quietly absorbs whatever didn't fit the framework the consultant brought in — and sometimes that's where the sharpest feedback lives.
Then check the units. Are the percentages percentages of people or of comments? One resident who writes five paragraphs about drainage can register as five separate comments. A theme described as "frequently mentioned" may rest on a dozen responses out of a thousand.
Finally, look for the questions that disappeared. If the survey asked about something specific, and the summary's theme list never touches it, that's not proof of anything on its own — but it's a fair question for the dais.
Three Patterns That Look Like Something
When feedback goes sideways, it usually goes one of three ways.
Grouped away. A distinct concern — say, accessibility at a specific trailhead — gets folded into a broad theme like "maintenance" and becomes invisible. The comment was counted. The point wasn't carried.
Counted but not connected. The summary lists the theme accurately, then the recommendation section doesn't mention it. Readers skim the themes and assume the recommendations reflect them. Check whether the memo draws an explicit line between the two, or just places them near each other.
Absorbed by a later document. The engagement memo says one thing, and the staff report attached to the actual vote cites a different, earlier analysis. Both documents are public. Only one of them gets read aloud.
None of these require bad faith. Coding is judgment, and judgment compresses. But compression is exactly the thing you're entitled to inspect.
The Free Follow-Up
You don't have to file anything to raise this. If the summary is presented at a public meeting, that meeting has a comment period, and "could staff attach the verbatim appendix to the agenda item" is a two-sentence request that's hard to refuse on the record.
If you want to go further, ask that the raw export be posted alongside the summary on the project page. A handful of cities already do this by default. It costs nothing, it ends the argument, and it makes the next coding pass more careful.
One more thing, and it's the practical one: the survey for the next phase of whatever this is — the design round, the implementation plan, the follow-up study — is probably being drafted right now. Asking these questions in November about a survey that closed in March is not a complaint about the past. It's a note to whoever is writing question ten this time.