Reports built to look like progress

Every number on the report went up, and the phone rang about as often as it did last month. It is one of the most common things a business owner brings to me, and it is worth being precise about, because the report is usually not lying. It is selecting. Somebody chose which numbers went on the page, and the ones that made the cut are largely those that rise on their own, whether the work is any good or not.

A green arrow is almost never a lie

I want to be clear about that up front, because the point gets lost the moment it becomes an accusation. In almost every case nobody invented a figure. The impressions really did go up. The average position really did improve by 4.2. Those numbers came out of Google Search Console or a rank tracker and went into a template.

The problem is the template. It gets built once, applied to every account in the building, and the pressure on it is simple: it has to survive a bad month, because a report that goes red in month three triggers an awkward call and sometimes a cancellation. So over a few iterations the metrics that can go red fall off it and the ones that drift upwards by themselves stay. It is what a document becomes when its unspoken job is to keep clients calm.

So the better test is not whether the numbers are true. It is whether the report is capable of telling you bad news. If no plausible version of last month would have produced a red arrow, that is not a measurement.

A report that cannot show you a bad month is not a report. It is a subscription to feeling fine.

Metric by metric: what each one actually means

None of these are useless and each has a job. The failure is using them as the headline measure of whether your money is coming back, because every one can climb while your revenue does nothing.

Impressions

An impression is counted when a page of your site appears in a set of search results someone looked at. That is the whole definition. It does not mean anyone saw your listing, read it, or was near ready to buy. Position 74 for a search typed in another country counts the same as position two for a person three suburbs away with a burst pipe.

Impressions rise for plenty of reasons unconnected to your commercial position: Google indexes more pages, experiments with new phrases, search volume climbs seasonally, a competitor drops out. Picture a plumber whose impressions doubled because a blog post on how heat pumps work started surfacing nationally. The chart looks excellent. He is being shown to thousands of people who will never ring anybody in his city.

Keywords in the top 100

The top 100 is the first ten pages of Google, and nobody in the history of hiring a tradesperson has gone to page eight. A keyword entering the top 100 means Google has formed some opinion about one of your pages, and that is all it means.

It is also easy to grow: publish pages, wait, and it climbs on its own, because every new page picks up phrases nobody competes for. Movement from position 94 to 41 reads as improvement and is worth nothing commercially. If the count of top 100 keywords is on your report and the count in the top three is not, that choice has been made for you.

Total clicks

Clicks are the closest thing on a standard report to a real number, because a click is a human arriving on your site. But a total with no split hides the most important cut: brand versus non-brand. Brand searches are people typing your business name, who already know you exist. Non-brand searches are people typing what they need, who found you because of search work.

Sponsor a junior rugby team, get a mention in the local paper, or just serve repeat customers who cannot remember your web address, and brand clicks climb. Total clicks go green. None of it came from SEO. A report that does not separate the two is describing your marketing in general and taking credit for parts of it.

Domain authority scores

Domain authority, domain rating and the rest are numbers out of 100 invented by SEO software companies to estimate how strong a site looks, based mostly on backlinks. Fair enough for sizing up a competitor at a glance. Not a Google metric: Google does not publish it, use it, or see it.

As a progress metric it is close to worthless and mildly dangerous, because bulk links from sites that will never send you a customer push it up. The cheapest and riskiest link building moves it fastest. If that score climbs and nothing else does, understand how cheap SEO can damage your domain before treating it as good news.

Why average position is close to meaningless on its own

This one gets its own section, because it is the metric most likely to be quoted at you on a call and the quickest to fall apart under pressure.

It averages across every keyword tracked and weights them identically. Say you track 40 phrases. One is your best money term, typed by people who have decided to hire someone today, and it moved from position six to three. That single move might be worth more than everything else combined. Meanwhile twelve informational phrases you would never pick up the phone about drifted from 88 to 61. The twelve move the average far more than the one that pays.

Worse, adding keywords changes it mechanically. Add long, uncompetitive phrases where the site already sits at position eight and the average improves without a thing changing on the site. The number moved because the list moved.

Then there is the sentence itself. "Average position improved by 4.2" says nothing until you know from what, to what. From 32.1 to 27.9 is a rounding error in a place nobody looks. From 6.3 to 2.1 is a different company, because the gap between the top of the results and just below it is enormous and nothing like linear. I have written separately about the cost of being stuck at number three rather than number one. The distribution is the story; an average hides it.

The comparison period decides the story

This is the part that surprises people most. Take one month of genuinely unchanged data and you can write three honest, defensible and completely different headlines about it purely by choosing what to compare it against.

• Against the previous month. The common default, and the noisiest, because the month before may have had a public holiday, a different number of working weekdays, or sat at a different point in your season.
• Against the same month last year. Usually the most honest comparison for any seasonal business, because it holds the season constant. It is also the one most likely to expose a flat year.
• Against a rolling three-month average. Good for seeing a trend rather than a wobble. It also blurs a sharp recent drop into something gentle.

Picture a landscaping business here in New Zealand looking at August. August against July looks fantastic, because people think about their gardens as spring approaches and demand climbs on its own every year. August against August last year tells you whether the search work did anything. Same data, opposite conclusions, neither technically false.

The failure is not picking one of the three. It is picking a different one each month, whichever looks best. And be wary of any figure framed as a running total since the campaign began, because a cumulative number can only go up. It is incapable of reporting a bad quarter.

Keyword sets that quietly change

The tracked keyword list is the ruler you measure with. If it changes length between measurements you no longer have a measurement, just two unrelated numbers side by side.

Lists change for reasons that are rarely sinister. Someone tidies the tracker, a content push adds fifteen phrases the new pages target, a term gets dropped as a duplicate. Each edit is defensible alone. The cumulative effect is a report that structurally cannot decline, because anything performing badly can leave the list and anything easy can join it.

The fix is simple and any decent provider will agree on the spot. Fix a core set of money terms, keep it stable, report it identically every month, and flag additions as additions, shown separately for a few months.

The ten-minute self-check

No SEO knowledge needed, and nothing to log into. You need last month's report and this month's open side by side, and about ten minutes.

1. Compare the two keyword lists. Same phrases, same number of them, same order? If terms have appeared or vanished with no explanation, the documents are not comparable and no trend across them means anything.
2. Check the comparison period on each. Are both against the previous month, or has one switched to year on year or a rolling average? Changing the basis between reports is the easiest way to make a flat month look good.
3. Check the length of the periods. A 28 day window against a full month, or a five-Monday month against a four-Monday one, is not like for like in a business that trades on weekdays.
4. Count the lines about enquiries, phone calls, form submissions, bookings or sales, rather than rankings, impressions and clicks. On a lot of reports that count is zero.
5. Look for a brand versus non-brand split on the clicks. Without it you cannot tell how much growth came from people who already knew you.
6. Find one number that got worse. Any number. If nothing is red anywhere across two consecutive months, this is not a document designed to inform you.

That last one is the real test and the one I would run first. Every account has a bad month: competitors invest, Google updates, a page slips. Anyone genuinely measuring your business will occasionally hand you something they have to explain.

What a report for a local business should actually contain

It does not need to be long. The best reporting I have seen for a small service business fits on two pages. It needs to be comparable with the last one, and connected to money.

• A stable tracked set of money terms, usually 20 to 40 for a local business, identical every month, with additions clearly marked as new.
• The current position for each term with last month's beside it, so you see the distribution rather than an average. Map pack positions shown separately from the standard blue links, because they behave differently.
• Enquiries: calls, form submissions, bookings, quote requests. The actual count, with the previous period beside it.
• Which pages produced those enquiries. This is the line that turns a report into a decision, because it says where the next piece of work should go.
• Non-brand organic clicks separated from brand clicks.
• A short plain English note on what was done, what it was meant to achieve, and whether it did - including anything stopped or reversed.

One honest caveat on the enquiry lines: attribution in a local business is never perfect. Someone finds you on Google, thinks for three weeks, then rings a number off the side of a van. No tracking catches that cleanly, and a report claiming flawless attribution is its own kind of fiction. What you want is a good faith count from call tracking and forms, plus a "how did you hear about us" question asked consistently by whoever answers the phone.

The questions to put on your next call

Three questions, none of which need to be combative. You are not catching anybody out, you are finding out whether the information exists at all.

1. "Could you send me the same keyword list, in the same order, for the last six months, in one spreadsheet?" Anyone tracking a stable set can do this in ten minutes, because the data already exists in that shape. If it takes two weeks or comes back reshaped, the set was not stable.
2. "Which pages produced enquiries last month, and how do you know?" The second half is the important half: it tells you whether enquiries are measured or estimated from traffic.
3. "What on this report got worse, and what are you doing about it?" This is the one that separates a person measuring your business from a person producing a document about it.

A good provider will enjoy these, because they already ask them of themselves. Defensive or vague answers prove nothing dishonest, but they tell you the reporting has drifted into a monthly reassurance exercise. Better to name that early than a year in, when the harder conversation is about having paid for SEO and gone backwards.

This failure belongs to a family of related ones, so read it next to the wider picture of why SEO agencies fail, and if you are weighing up a change, the things worth asking are in how to choose an SEO provider. Reporting is where the relationship either stays about your revenue or quietly stops being about it, and it takes ten minutes and two documents to find out which you have.