An annual website review is only useful when it joins two perspectives: what people did on your site and what happened once they wrote to you. Together, they show which pages bring conversations, where customers need help and whether the team kept up at peak times. This guide gives you twelve questions for the turn of the year and an order of work that fits into one afternoon.
This is not a monthly customer service report, and it is not goal setting for next year. An annual review answers what changed over twelve months and why. Before you answer anything, check one technical detail: whether the detailed records for January still exist.
Twelve questions for your annual review
Each question can be checked in the data. The third column tells you what to watch out for before you write the answer down. Order matters: website first, then support, then data quality.
Scroll the table sideways to see all columns.
| Question | Where to look | Watch out for |
|---|---|---|
| 1. Which landing pages produced the most conversations, not just visits? | Landing pages with conversation counts | High traffic with no conversations means something different from low traffic with many |
| 2. Which sources brought visits that ended in a conversation or goal? | Sources and funnel | Label sessions without a recorded source separately |
| 3. Which months peaked in visits, and which in conversations? | Daily chart across the year | The two peaks may fall in different months |
| 4. How did first response time change? | Median and p90 by quarter | An average hides long waits at peak times |
| 5. How did resolution time and reopen counts change? | Service report | Faster resolution with more reopens is not progress |
| 6. How many callback requests were completed, and how many waited? | Callback requests and completions | Older entries may lack a completion date |
| 7. Which tags were most common, and did their ranking change? | Tag list and inbox filter | A changed tag definition changes the result |
| 8. Which goals were reached, and from which landing pages? | Goals in the traffic report | A goal added mid-year does not cover a full year |
| 9. On which pages did conversations most often concern a problem? | Conversations, recordings, heatmaps | A few loud cases are not yet a pattern |
| 10. How did CSAT change, and on what sample? | Score distribution | A small sample moves for no reason |
| 11. How was team workload spread during peaks? | Team report | Workload is not a rating of individuals |
| 12. In which periods was the data incomplete? | Coverage and freshness | A tracking gap looks like a decline |
Before you answer: check what is left in the data
Most tools keep two kinds of data. Details are individual visits, sessions and events you can drill into from a chart. Aggregates are daily totals, and year-over-year comparisons rely on them. Details usually expire sooner, so in December you may still have the total visits for March but no longer the March sessions behind them.
Google explains that a standard GA4 property can keep user-level and event-level data for 2 or 14 months, and that this setting affects explorations and funnel reports, not standard aggregated reports. Google Analytics Help. It does not show your property's setting; check that in Admin.
In Meetlead Analytics, detailed visit history is kept for 30, 180 or 365 days on the standard plans, and daily aggregates for 24 months. You can therefore compare years on aggregates, but drilling into specific sessions from early in the year depends on your plan. Export what you cannot rebuild later before the review.
- Landing pages and sources with conversation counts, quarter by quarter.
- Business goal results, together with the key or pattern that defines each goal.
- Median and p90 for first response time and resolution time for every quarter.
- Your tag list with a short note on how the team understood each tag this year.
- Coverage and freshness information for any period that looks doubtful.
Better still, do not wait for December. A monthly scheduled report sent as XLSX builds an archive that does not depend on retention periods, and at year end you only need to merge twelve files into one table.
Questions 1–3 and 8: pages, sources and goals
Start with landing pages, because that is where the gap between traffic and conversations shows fastest. Set the range to the full year and sort pages by conversations, not page views. A fictional custom furniture workshop might find that its post on wardrobe dimensions brings plenty of visits and few questions, while a less visited page on assembly prices leads to one conversation in three. It simply tells you which page to review first.
Compare monthly peaks with care. A month with more working days, a campaign or a tracking change can pass for a trend; our guide on seasonality in traffic and enquiries explains how to tell fluctuation from change. For goals, note when each was created. A goal set up in September says nothing about the first half of the year, and next year's comparison is only fair if the definition stays the same.
Questions 4–7 and 10–11: support speed and quality
Show first response time and resolution time by quarter, as the median and p90. The median describes a typical contact, while p90 shows how long the unluckiest customers waited. If the team changed its hours or its counting method during the year, mark that in the table. Our guide on ticket resolution time helps you keep definitions consistent, including pauses while you wait for the customer.
Tags tell you what people asked about, but only if they meant the same thing all year. The counter next to a tag is not an annual breakdown, so pick the three most common tags, filter the inbox by each and read a few conversations from the start and end of the year. If the categories drifted, tidy them up with our guide to tagging conversations and tickets before the new year begins.
Always report the number of ratings next to the CSAT score. Read team workload as input for planning peak-time cover, not as a judgement of individual people. Ask which week the queue outgrew the team's replies, and what helped then.
Questions 9 and 12: trouble spots and tracking gaps
To find pages where customers most often wrote about a problem, link conversations to the page where they started. List five such pages and, for each, watch a few session recordings and check the heatmap if your plan includes them. Recordings are kept for less time than visit history, so for earlier months you will usually have only the conversations. Look for a repeated sticking point, such as a form field or unclear delivery details, not proof that the page is bad.
The last question protects you from false conclusions. If a script, consent banner or page template changed in a given month, the figures for that period may be too low. When one tool disagrees with another, check definitions and date ranges first, as described in our guide on why the data does not match. Label any period with incomplete coverage clearly in the review.
Data for [period] is incomplete because [reason]. We show the results for this period but do not compare them with the previous year.
How to run the review in one afternoon
- Collect the files.Export the checklist items to CSV or XLSX before the details expire, and put the date range in each file name.
- Answer briefly.Write one sentence per question, with a figure and a comparison to the previous year if the aggregates allow it.
- Drill into examples.For the three biggest changes, open a few conversations or sessions to check the number describes what you think it does.
- Write down conclusions.On one page: three things that worked, three problems and the periods with incomplete data.
The review document is a starting point for planning, but setting priorities is separate work, covered in our guide to customer service goals for the new year. Keep the twelve questions unchanged, so next year's comparison is one you can trust.
Questions about the annual review
Is an annual review worth doing if I only have a few months of data? Yes, but without a year-over-year comparison. Answer the questions for the period you have, label it clearly and treat the result as a baseline for next year. Draw seasonal conclusions only after seeing the same month twice.
What if the details for the start of the year have already expired? Work with the aggregates and note that you cannot drill into individual sessions for that period. Missing examples do not invalidate the totals, but they limit the explanations. Going forward, set up a monthly scheduled report so the archive builds itself.
