ChatGPT for SEO? You’re Using It Wrong.

3 months ago 29
ARTICLE AD BOX

Everyone successful SEO uses ChatGPT. But adjacent aft each these months, I inactive spot galore folks utilizing it wrong.

Let’s research 3 communal mistakes and however you tin fix them.

Mistake 1: Asking it questions it doesn’t person the reply to

Believing everything ChatGPT says is ne'er a bully idea, but believing answers to questions it doesn’t person information to reply is simply a look for disaster.

For example, here’s maine asking it for easy-to-rank-for long-tail keywords:

Asking ChatGPT for long-tail keywords

The results mightiness look good astatine archetypal glance, but if you paste them into a keyword probe instrumentality similar Ahrefs’ Keywords Explorer, you’ll spot however atrocious they are. The hunt request for each of them is truthful debased that they’re not adjacent successful our database of 16.2 cardinal keywords:

ChatGPT's long-tail keywords person  nary  hunt  demand

Google Trends shows the same thing:

ChatGPT's long-tail keywords person  nary  hunt  demand... Google Trends confirms

The crushed for this is simple: ChatGPT conscionable doesn’t person the data. It has nary hint however often things get searched oregon whether a keyword is “long tail.”

Here’s different illustration of a question ChatGPT doesn’t person the information to answer:

ChatGPT can't archer  america  the hunt  intent of a taxable   unless. It doesn't person  the data

Once again, we tally into a akin problem. To classify intent, you request to really hunt for the keyword and analyse the top-ranking pages and SERP features. ChatGPT can’t adjacent browse the web (well, GPT 4 can, but it can’t browse Google results), truthful it simply cannot bash this well.

How to hole this mistake

Keep successful caput ChatGPT’s limitations and however it works. If you cognize it doesn’t person the information to reply a question well, either refrain from asking oregon springiness it the information it needs to supply a amended answer.

For example, if you privation it to classify hunt intent, you request to springiness it thing to spell on, similar the titles of the top-ranking pages. Only past tin it adjacent effort the task at hand.

ChatGPT tin  classify hunt  intent if we springiness  it thing  to spell  on, similar  the title's of top-ranking pages

Are you an Ahrefs user?

We’ve built a overmuch much robust mentation of this thought into Keywords Explorer. Just click “Identify intents,” and AI volition provender backmost the results.

SERP intent successful  Ahrefs' Keywords Explorer

Similarly, if you privation ChatGPT to find contented gaps, you request to provender it the contented of some pages and inquire it to place gaps. Or, alternatively, run a contented spread analysis successful Ahrefs, export the keyword gaps, and inquire it to summarize them.

Mistake 2: Expecting cleanable results from your archetypal prompt

ChatGPT seldom gives you precisely what you privation connected the first try.

For example, I precocious asked for immoderate method coffee-related presumption to usage arsenic effect keywords successful Keywords Explorer.

Here’s what it gave me:

Asking ChatGPT to method  coffee-related terms

This isn’t a atrocious response, but having the descriptions determination (which I didn’t inquire for) was annoying due to the fact that it made copying and pasting the database of presumption difficult.

I faced a akin contented erstwhile I asked it to radical keywords into broader categories, and it gave maine a bulleted list.

Refining ChatGPT's output

I didn’t deliberation overmuch of this astatine first, but I soon realized I needed the results formatted arsenic a array truthful I could paste them into a Google Sheet and filter them.

How to hole this mistake

Your archetypal thought mightiness beryllium “prompt engineering” oregon buying immoderate premium ChatGPT prompts.

While these things mightiness help, you often inactive won’t get cleanable results close distant for 1 elemental reason: ChatGPT prompts are similar recipes. Even if they person a coagulated base, you inactive person to set them to your tastes.

For example, here’s a genuine premium punctual for penning a rubric and meta description:

Example of a premium ChatGPT prompt

And here’s the output:

ChatGPT's meta rubric  and description

It’s not bad, but it’s a spot adust for my liking. So, I followed up and asked ChatGPT to marque it “a spot little ceremonial and much playful.”

ChatGPT's precise  informal effort  astatine  a meta rubric  and description

Definitely better, but a small too playful this time. 

Let’s inquire it to dial it down a bit.

ChatGPT's playful but not excessively  playful effort

Not bad.

In my experience, this iterative process is the cardinal to getting bully results from ChatGPT. Keep things elemental to start, past refine arsenic you go.

For example, I precocious tried creating a UK taxation calculator with ChatGPT. Instead of penning retired insanely elaborate instructions from the get-go, I started with a elemental prompt: “create a uk taxation calcalator successful HTML, CSS, and JS”

Here’s the calculator it created:

ChatGPT's archetypal  effort  astatine  creating a UK taxation  calculator

Is it perfect? No. I wanted it to cipher the taxation automatically based connected UK bands alternatively than asking the idiosyncratic for their taxation rate. But with conscionable a fewer iterations, the occupation was solved, and I had a beauteous neat and close calculator.

The taxation  calculator I got ChatGPT to make  aft  a fewer  iterations

Looking to usage AI to constitute meta titles and descriptions?

Check retired Ahrefs’ escaped title generator and meta statement generator. Simply archer it what your leafage is about, prime a tone, and it’ll make everything for you.

Ahrefs' escaped  AI rubric  generator

Mistake 3: Asking it to “write an article…”

ChatGPT volition happily enactment 1 connection successful beforehand of the different if you inquire it to constitute something, but the contented volition conscionable beryllium a mishmash of what’s already been said a cardinal times. In different words, boring boilerplate content.

Case in point:

ChatGPT's mega dull effort  astatine  "writing an article"

(Talk about dull…)

How to hole this mistake

Don’t conscionable inquire ChatGPT to “write an article.” Give it much guidance and usage it to brainstorm ideas.

For example, accidental I’m penning a station connected nexus gathering tips. Rather than asking it to constitute the full nonfiction without guidance, I tin commencement by asking for immoderate outreach tips to get my originative juices flowing.

Brainstorming with ChatGPT, my caller   champion  friend

I don’t emotion immoderate of these, but I deliberation you could harvester #1 and #4 for a unsocial tip: grounds and see a abbreviated video transportation successful your email.

Now I person the idea, I tin inquire ChatGPT to grow connected it and constitute it up:

ChatGPT's effort  astatine  polishing the extremity   we came up   with together

Is this perfect? No. But it’s surely not bad. In fact, if we chopped immoderate of the fluff and walk 5 minutes polishing the copy, it’s beauteous decent:

X. Record Short Video Pitches

Personalized video messages adhd a quality interaction to your pitches, and you tin make them effortlessly with a inexpensive webcam and escaped Loom account.

For example, accidental you’re moving a breached nexus gathering campaign. Instead of sending a long-winded email explaining the issue, wherefore not amusement them precisely what you mean? Fire up Loom and springiness them a ocular rundown.

  1. Point retired the breached nexus connected their page.
  2. Show them precisely wherever it is and wherefore it matters.
  3. Suggest a replacement nexus from your website and explicate wherefore it’s a great fit.

Here’s an example:

{insert video}

It mightiness lone beryllium 1-2 minutes long, but it’s ace helpful, and the idiosyncratic interaction tin marque a large quality successful however your transportation is received.

ChatGPT is besides large astatine helping you amended and optimize existing content. Again, you conscionable request to springiness it the close direction…

For example, we precocious recovered that we ranked #7 for “blogging tips” but didn’t ain the featured snippet.

Our website ranking #7 for "blogging tips". Data via Ahrefs' Site Explorer

Sam theorized that, among different things, it mightiness beryllium due to the fact that the top-ranking page’s subheadings intelligibly communicated what their tips were, portion galore of ours were vague:

Our vague subheadings

He decided to have ChatGPT rewrite them for clarity:

ChatGPT's polished subheadings

Did it work? You’ll person to ticker the video to find out, but honestly, it doesn’t matter. It’s evident to maine that the ChatGPT “optimized” ones are overmuch clearer than what we had before, which means ChatGPT improved our content.

ChatGPT is besides large astatine proofreading, simplifying unnecessarily lengthy sentences, and thing on those lines. You know, each the worldly that improves contented prime and nary uncertainty indirectly influences idiosyncratic signals.

Final thoughts

ChatGPT is insanely almighty and useful—but lone if you recognize however to use it.

If you’re hoping for it to constitute large contented with nary absorption whatsoever, support hoping. It can’t bash that yet. But it tin assistance with brainstorming, polishing human-written content, and galore different SEO tasks. You conscionable request to support successful caput its limitations.

Got questions? Give maine a outcry connected Twitter X.