How Is Voice Search Changing Digital Marketing?
Less than the statistics suggest, in a more specific way than most articles describe. This page quotes only figures whose source and year we can name, explains the mechanism rather than the excitement and tells you plainly which businesses should care.
What Voice Search Actually Is
Most of the confusion in this subject starts with one word covering three different behaviours. Only the first of them is voice search.
Asking a device a question. This is voice search.
Somebody asks where to find something, what something costs or who does a particular job locally. A search happens and an answer comes back.
Giving a device a command. Not search at all.
Setting a timer, playing music, turning off a light. No search takes place and no business is involved. A great deal of what gets counted as voice usage is this.
Buying through an assistant. A separate thing again.
Ordering a repeat item aloud. Relevant to a narrow set of retailers and irrelevant to almost every service business.
The fourth case, which matters most. Dictating into a search box.
Tapping the microphone and speaking instead of typing. Google's John Mueller has explained that this is an ordinary web search, logged exactly like a typed one. Most so-called voice search is this. It changes nothing about how you are found.
How A Spoken Query Differs From A Typed One
People speak differently from how they type. The differences are consistent enough to be worth designing around.
Longer. Typed: emergency plumber Guildford. Spoken: who can come out and fix a leaking pipe today near me.
Phrased as a question. Typed queries are usually fragments. Spoken ones are usually whole sentences, frequently beginning with how, what, where or who.
More conversational. Complete with the small words people drop when typing, because saying them costs nothing.
More often local. A large share of spoken questions are about something nearby, since people ask aloud when they want to go somewhere or reach somebody now.
Frequently asked while doing something else. Driving, cooking, carrying something. Hands and eyes are occupied, which is the whole reason for speaking rather than typing.
Why this matters commercially. Intent is usually more immediate.
Somebody asking aloud is often closer to acting than somebody typing at a desk. That makes the interaction more valuable, while being considerably harder to see in any reporting.
The Single Answer Problem
A screen returns ten results and lets the person choose. A speaker returns one. Everything genuinely different about this subject follows from that.
What ranking second is worth on a screen. A great deal.
Somebody scans the page, forms an impression of several options and picks. Second place still gets seen, still gets clicked and still wins work.
What it is worth on a speaker. Very little.
One answer is read aloud. Nobody hears about the alternatives, so second place is functionally invisible in that moment.
What that changes about content. Answer before elaborating.
Material that states the answer plainly and briefly, near the top, before explaining the detail, is more likely to be usable as that single answer. It also happens to serve every reader better.
Why this is not a new technique. It is old advice with a new reason.
Answering a question directly rather than circling it has always been better writing. The single answer situation raises the cost of not doing it.
Where Spoken Answers Come From
An assistant does not hold facts about your business. It assembles an answer from whatever it can find. Knowing the four places it looks explains most problems people report.
The search index. Your pages, as understood.
The business profile. Hours, address, telephone number, services.
For anything local this is the most consulted source. It is also the one businesses update least often.
Structured data on the page. Information stated in a form a machine can read.
Third party directories and aggregators. The uncomfortable one.
Listings you may not remember creating, on sites you may not know exist, carrying details from years ago.
What this explains. Wrong information repeated as fact.
An out of date opening time in one listing can be read back to a customer with complete confidence. Nobody is at fault. The assistant found a source and repeated it. The customer arrived at a closed door.
What follows. Listings are not administration.
They are the raw material for answers given aloud about your business, which makes keeping them current a commercial task rather than a clerical one.
What This Actually Changes For A Business
Five things. The important point about all five is that none is a new discipline. This is the same work done more carefully.
Listing accuracy. Every place your details appear, correct and current.
The single highest return item here, costing time rather than money.
Opening hours and location data. Including exceptions.
Bank holidays, closures, seasonal changes. These get read aloud as fact, so an unmaintained set of hours actively misleads customers.
Content that answers in a sentence. Then elaborates.
Taken from the previous block. It improves the page for everybody rather than only for a speaker.
Page speed. Unchanged in principle, more consequential here.
An answer needed while somebody is driving cannot wait for a slow page.
Structured data. So the answer can be read cleanly.
Stating facts in a machine readable form removes ambiguity about what your information means.
What this adds up to. No separate strategy.
Anybody selling you a voice search strategy is selling you accurate listings, clear answers and a fast website, which you needed anyway.
Where It Matters And Where It Does Not
This subject is worth attention for some businesses and worth almost none for others. Being specific about which is more useful than urging everybody to prepare.
Matters most: local services people need now.
Emergency trades, taxis, locksmiths, breakdown services. Somebody with a problem, speaking rather than typing because their hands are busy or they are outside.
Matters: food, hospitality and opening hours.
Restaurants, pubs, cafes, anywhere people ask whether it is open and how to get there.
Matters: directions and quick factual questions.
Where you are, when you close, whether you do a particular thing.
Matters least: considered purchases.
Kitchens, extensions, cosmetic procedures, professional services, anything expensive or irreversible. Nobody chooses a surgeon by asking a speaker. Those buyers want to compare on a screen, read reviews and think.
Matters least: anything visual.
If the decision depends on seeing something, a spoken answer cannot serve it. A business selling on appearance should spend its attention elsewhere.
Voice And The Move To Assistant Answers
The single answer dynamic is no longer confined to speakers. Generated answers on screen do the same thing, which makes this subject larger than voice while making voice itself less distinctive.
What has changed. Screens summarise too.
A question typed into a search box increasingly returns a composed answer above the results rather than only a list. The person reads one answer instead of choosing among ten.
Why that matters more than voice. Volume.
Spoken questions are a minority of what happens. Typed questions receiving a summarised answer are not.
What serves both. The same writing.
Content stating an answer plainly, in a form that can be quoted without distortion, is what gets used by a speaker and by a generated summary alike. One piece of work, two surfaces.
Where this goes next. Beyond the scope of this page.
How material is written to be used by answer systems is a subject of its own, covered by our AI optimisation service rather than here. The wider map is in marketing disciplines explained.
What Cannot Be Measured
There is no reliable way to know how much of your search traffic was spoken. That is not a gap in your setup. It is how the reporting works.
No split between spoken and typed. None available.
Search Console does not distinguish them. A dictated query is logged identically to a typed one, because that is what it is.
Some interactions are not recorded at all. The stronger point.
Google's John Mueller has explained that where an assistant reads an answer aloud from a page, the impression is not logged in Search Console. Those interactions leave no trace in your reporting.
What that means for any quoted figure. Treat it carefully.
If the platforms cannot separate spoken from typed, anybody quoting a precise voice conversion rate for your sector is working from something other than measurement. Ask where the number came from.
What you can watch instead. Proxies, labelled as proxies.
Long question shaped queries. Local pack visibility. Calls and direction requests from your business profile. None of these is voice traffic. All of them move when the underlying work is right.
The Myths
This subject carries more unsourced figures than any other in marketing. Rather than repeat them as fact, here is the method for checking one, applied to the two most quoted.
Step one. Find the original publisher.
Step two. Check the year.
Step three. Check whether it measured what the headline says it measured.
The most quoted figure of all. Real, also old.
Sundar Pichai, then Google's chief executive, said at Google I/O in May 2016 that twenty per cent of queries on the Google mobile app and on Android devices were voice searches. That is a genuine, sourced, dated statement. It is a decade old. It described the Google mobile app and Android devices rather than all searches everywhere.
The forecast still quoted as a fact. A prediction about a year long past.
The claim that half of all searches would be voice by 2020 was a forecast, made years before that date. The date has come and gone. It is still repeated in the present tense.
What this teaches. The check takes two minutes.
Most figures in this subject fail at step two or step three. A supplier quoting one without a source and a year has not done the check.
Common Mistakes
Each of these is a reasonable response to the hype rather than to the mechanism.
Treating voice as a separate strategy. Commissioning a distinct programme for what is accurate listings, clear answers and a fast site. The work is already on your list under other headings.
Question headings with no answer underneath. Pages full of headings phrased as questions where the answer arrives three paragraphs later, if at all. This is the commonest attempt at voice optimisation and it produces worse pages.
Neglecting the business listing. Spending on content while the most consulted source about your business carries last year's hours.
Ignoring page speed. Particularly for anything urgent or local, where the person asking is moving.
Chasing the trend instead of the fundamentals. The one that costs most. A business whose site is slow, whose listings are wrong and whose pages answer nothing does not have a voice search problem. It has the problems described throughout what is digital marketing. The material itself is covered in what is a content strategy. The full series is on the digital marketing guide.
