Dashboards

Number

One headline figure — responses, completion rate, an average score. The workhorse of the top row.

How it's counted

• Pick one measure — either a ready-made one (every one is listed under Headline number measures) or one worked out from a question: an average, total, lowest, highest, middle value, lower or upper quarter, or spread of a number question, or the % who gave one particular answer.
• % who gave an answer = people who picked it ÷ people who answered the question — people who skipped the question are left out of both sides.
• The vs the previous period chip compares with the same-length window just before the range you picked (30 days → the 30 days before). It appears only once a time range is chosen.

What you can set

• Colour rules: up to three, such as red below 40 and green at 80 or above. The first matching rule wins, and a word (“Needs attention”) always appears beside the colour.
• Show change vs the previous period.
• Only count… (up to two conditions).

Progress to target

A dial showing how close a number is to a goal you set — for example 600 responses against a target of 1,000.

How it's counted

• The number is any measure a Number widget can show. The dial fill is value ÷ target, capped at a full dial — going past the target does not overflow it.
• The percentage under the number is the same sum, rounded to a whole number, and is not capped (130% shows as 130%).

What you can set

• Target: any number above zero.
• Colour rules and Only count…, as for Number.

Bar chart

Compare groups side by side. Stack each bar by a second question, put two rounds next to each other, rank the groups, or draw a target line.

How it's counted

• Each response goes into one bar, chosen by its answer to the Group responses byquestion. The bar's length is the measure for the responses in that bar — by default the number of responses.
• In the picture, the measure is average of Numeracy score: add up the scores in each district and divide by how many scored. The dashed line is a target of 60; two of five groups reach it.
• A question left blank by a respondent goes into a (no answer) bar so the counts still add up.
• Only the largest groups get their own bar (10 by default, up to 50). Smaller ones are combined into “Other” — but only when counting responses, because an average of leftovers would mislead.

What you can set

• Group by, Measure and Show top groups.
• Stack by / Compare side by side by — a second question, response status, or year/month.
• Order groups: largest first, highest value first or lowest value first.
• Target line (bar charts only): the chart prints how many groups have reached it.

Pie / donut

How responses split across a few options. Best for four or five choices; use a bar chart for more.

How it's counted

• One slice per answer; the size is the measure (by default the number of responses). The percentage is slice ÷ all slices shown: 140 ÷ 220 = 64%.
• Past five slices the rest fold into Other, and (no answer) gets a muted slice of its own, so identity never rests on colour alone — every slice is also named in the list.
• A donut cannot be split by a second question.

What you can set

• Group by, Measure and Show top groups.

Trend over time

Responses — or an average, a total, a rate — by day, week, month or year. Optionally a running total.

How it's counted

• A response is placed on the day, week, month or year it was submitted. Days, weeks (Monday start) and months are calendar periods in UTC, so a response filed late evening in India can land on the next UTC day.
• When counting responses, days or weeks with no responses are drawn as zero rather than skipped, so quiet stretches are visible.
• Running total turns each point into this period + every period before it. It needs day, week, month or year, and a count or a total.
• Day of the week and Hour of the day are also available here and on bar charts. They use India Standard Time (UTC+5:30) and always show every weekday or all 24 hours.

What you can set

• Group by time, Measure, Show a running total, Only count….

Table

Exact numbers per group — or a cross-tab of two questions, optionally shaded like a heatmap.

How it's counted

• Same grouping and measure as a bar chart, but as numbers. With a count, an extra Share column shows group ÷ all responses (left out when people could pick several options, because shares would add up past 100%).
• A cross-tab has one column per answer of the second question and a Total — or Overall when the measure is an average or rate. A dash means nobody in that cell answered.
• Shade cells by size darkens each cell in proportion to the largest one, so one huge value does not wash out the rest.

What you can set

• Group by, Measure, Cross-tab by, Order groups, Shade cells by size, Show top groups.

Distribution

How scores or numbers are spread from lowest to highest — to see whether most children cluster near the middle or at the extremes.

How it's counted

• Takes one number, rating, slider or NPS question. Blank answers are skipped, and a note says how many of the responses answered it.
• If the answers are whole numbers with few different values (a 1–5 rating, a 0–10 score) there is one bar per value. Otherwise the range from the lowest to the highest answer is cut into equal-width groups, and each bar counts the answers falling in it.

What you can set

• Question, and Number of groups (3 to 20).

Net Promoter Score

Promoters, passives and detractors from an NPS question (a 0–10 “would you recommend” scale).

How it's counted

• Each answer is a promoter (9–10), a passive (7–8) or a detractor (0–6).
• NPS = % promoters − % detractors, from −100 to +100. With 50 promoters and 20 detractors out of 100: 50 − 20 = +30. Passives count in the total but not in the score.

What you can set

• NPS question, Only count….

Quiz results

Pass rate, average score and how scores are spread — for forms that are quizzes. You can set your own pass mark and choose which parts to show.

How it's counted

• Graded means the response has a score. Pass rate = passed ÷ graded: 154 ÷ 220 = 70%.
• Passed is the result the form itself recorded at submission — unless you tick Use my own pass mark, in which case a response passes when score ÷ maximum score ≥ your pass mark. The pass mark is shown on the card.
• Average score is the mean of score ÷ maximum score over graded responses — each response counts equally, however many marks the quiz is worth.
• The Scores chart counts responses in bands of 5%, 10% (the default) or 20% of the maximum, e.g. 0–9%, 10–19%… 90–100%.

What you can set

• Parts to show (pass rate, average score, score chart — at least one).
• Use my own pass mark (1–100%).
• Score band size.
• Only count….

Completion funnel

How many people open the form, start answering and submit — and where they drop away.

How it's counted

• Three counts from the form's own counters, not from responses: views (form opened), starts (first answer given) and submissions.
• Each percentage is step ÷ the step above: 364 ÷ 1,008 = 36%, then 220 ÷ 364 = 60%. Or choose step ÷ the first step to read every step against the people who opened the form.
• It shows a dash while a filter on answers is on, because views and starts are not tied to anyone's answers.

What you can set

• Steps to show (at least two).
• Show percentages as: % of the previous step, or % of the first step.

Data quality

Spots responses that look wrong, so you can fix collection while there is still time — duplicates, missing answers and impossible numbers.

How it's counted

• Duplicates: responses whose answers are exactly the same as another response's. The extra copies are counted, so three identical responses count as 2.
• Missing answers: a response is flagged if any ticked question was left blank. Pick the questions that must never be empty, such as the district or school.
• Out of range: a number below your lowest allowed or above your highest allowed value — for example a score of 150 on a 0–100 scale. Responses with no number are not flagged here.
• Each check shows the count, flagged ÷ all responses checked, and a word — Clear or Check. If a question used by a check is later deleted, the check shows “Unknown” instead of breaking the dashboard.

What you can set

• Up to eight checks, of any mix.
• Only count….

Location plot

Where responses came from, so you can see coverage and gaps — rounded so it shows clusters, not doorsteps.

How it's counted

• Takes one location question. Each response with a valid position becomes a dot.
• Positions are rounded to two decimal places — about 1 km — before they leave the server. At most 500 dots are drawn; the note says how many there were.
• There is no map underneath on purpose: nothing leaves the browser, and it works offline in a field office.
• Never shown on a shared public link.

What you can set

• Location question, Only count….

The five schools that need most help

Which schools should a coordinator visit first?

How it's counted

• Add a Bar chart. Group by School, Measure Average of Numeracy score.
• Set Order groups to Lowest value first, and Show top groups to 5.
• Optionally add a Target line at the score you are aiming for — the chart says how many schools have reached it.

Baseline vs endline

Did learning improve between rounds?

How it's counted

• Add a Bar chart. Group by District, Measure Average of Numeracy score.
• Set Compare side by side by to Year (or Month).
• Each pair of bars is the same measure for the responses submitted in that year. A district with no responses in a year simply has one bar.